{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.5780730897009967,
  "eval_steps": 500,
  "global_step": 950,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016611295681063123,
      "grad_norm": 69.42809295654297,
      "learning_rate": 5.4054054054054056e-08,
      "loss": 0.7218,
      "step": 1
    },
    {
      "epoch": 0.0033222591362126247,
      "grad_norm": 84.8204345703125,
      "learning_rate": 1.0810810810810811e-07,
      "loss": 0.7005,
      "step": 2
    },
    {
      "epoch": 0.0049833887043189366,
      "grad_norm": 62.26895523071289,
      "learning_rate": 1.6216216216216215e-07,
      "loss": 0.6448,
      "step": 3
    },
    {
      "epoch": 0.006644518272425249,
      "grad_norm": 77.79634094238281,
      "learning_rate": 2.1621621621621622e-07,
      "loss": 0.6623,
      "step": 4
    },
    {
      "epoch": 0.008305647840531562,
      "grad_norm": 76.27490234375,
      "learning_rate": 2.702702702702703e-07,
      "loss": 0.599,
      "step": 5
    },
    {
      "epoch": 0.009966777408637873,
      "grad_norm": 76.31900787353516,
      "learning_rate": 3.243243243243243e-07,
      "loss": 0.6139,
      "step": 6
    },
    {
      "epoch": 0.011627906976744186,
      "grad_norm": 72.4638442993164,
      "learning_rate": 3.783783783783784e-07,
      "loss": 0.6561,
      "step": 7
    },
    {
      "epoch": 0.013289036544850499,
      "grad_norm": 45.92433547973633,
      "learning_rate": 4.3243243243243244e-07,
      "loss": 0.5786,
      "step": 8
    },
    {
      "epoch": 0.014950166112956811,
      "grad_norm": 55.887176513671875,
      "learning_rate": 4.864864864864865e-07,
      "loss": 0.568,
      "step": 9
    },
    {
      "epoch": 0.016611295681063124,
      "grad_norm": 34.738040924072266,
      "learning_rate": 5.405405405405406e-07,
      "loss": 0.5106,
      "step": 10
    },
    {
      "epoch": 0.018272425249169437,
      "grad_norm": 23.600587844848633,
      "learning_rate": 5.945945945945947e-07,
      "loss": 0.4439,
      "step": 11
    },
    {
      "epoch": 0.019933554817275746,
      "grad_norm": 22.552448272705078,
      "learning_rate": 6.486486486486486e-07,
      "loss": 0.5118,
      "step": 12
    },
    {
      "epoch": 0.02159468438538206,
      "grad_norm": 17.616506576538086,
      "learning_rate": 7.027027027027027e-07,
      "loss": 0.4026,
      "step": 13
    },
    {
      "epoch": 0.023255813953488372,
      "grad_norm": 21.766603469848633,
      "learning_rate": 7.567567567567568e-07,
      "loss": 0.4315,
      "step": 14
    },
    {
      "epoch": 0.024916943521594685,
      "grad_norm": 21.652666091918945,
      "learning_rate": 8.108108108108108e-07,
      "loss": 0.3857,
      "step": 15
    },
    {
      "epoch": 0.026578073089700997,
      "grad_norm": 29.43855094909668,
      "learning_rate": 8.648648648648649e-07,
      "loss": 0.5083,
      "step": 16
    },
    {
      "epoch": 0.02823920265780731,
      "grad_norm": 27.099016189575195,
      "learning_rate": 9.18918918918919e-07,
      "loss": 0.3577,
      "step": 17
    },
    {
      "epoch": 0.029900332225913623,
      "grad_norm": 33.78343963623047,
      "learning_rate": 9.72972972972973e-07,
      "loss": 0.3735,
      "step": 18
    },
    {
      "epoch": 0.03156146179401993,
      "grad_norm": 37.29399871826172,
      "learning_rate": 1.0270270270270269e-06,
      "loss": 0.4251,
      "step": 19
    },
    {
      "epoch": 0.03322259136212625,
      "grad_norm": 28.423860549926758,
      "learning_rate": 1.0810810810810812e-06,
      "loss": 0.4173,
      "step": 20
    },
    {
      "epoch": 0.03488372093023256,
      "grad_norm": 27.742494583129883,
      "learning_rate": 1.135135135135135e-06,
      "loss": 0.3403,
      "step": 21
    },
    {
      "epoch": 0.036544850498338874,
      "grad_norm": 25.273988723754883,
      "learning_rate": 1.1891891891891893e-06,
      "loss": 0.3046,
      "step": 22
    },
    {
      "epoch": 0.03820598006644518,
      "grad_norm": 29.61273193359375,
      "learning_rate": 1.2432432432432432e-06,
      "loss": 0.3323,
      "step": 23
    },
    {
      "epoch": 0.03986710963455149,
      "grad_norm": 24.798965454101562,
      "learning_rate": 1.2972972972972972e-06,
      "loss": 0.2518,
      "step": 24
    },
    {
      "epoch": 0.04152823920265781,
      "grad_norm": 25.87079620361328,
      "learning_rate": 1.3513513513513513e-06,
      "loss": 0.2818,
      "step": 25
    },
    {
      "epoch": 0.04318936877076412,
      "grad_norm": 32.44166564941406,
      "learning_rate": 1.4054054054054054e-06,
      "loss": 0.3059,
      "step": 26
    },
    {
      "epoch": 0.044850498338870434,
      "grad_norm": 40.5885124206543,
      "learning_rate": 1.4594594594594594e-06,
      "loss": 0.3353,
      "step": 27
    },
    {
      "epoch": 0.046511627906976744,
      "grad_norm": 29.34911346435547,
      "learning_rate": 1.5135135135135135e-06,
      "loss": 0.2943,
      "step": 28
    },
    {
      "epoch": 0.04817275747508306,
      "grad_norm": 34.42272186279297,
      "learning_rate": 1.5675675675675676e-06,
      "loss": 0.2938,
      "step": 29
    },
    {
      "epoch": 0.04983388704318937,
      "grad_norm": 32.65325927734375,
      "learning_rate": 1.6216216216216216e-06,
      "loss": 0.3359,
      "step": 30
    },
    {
      "epoch": 0.05149501661129568,
      "grad_norm": 38.35554504394531,
      "learning_rate": 1.6756756756756755e-06,
      "loss": 0.3983,
      "step": 31
    },
    {
      "epoch": 0.053156146179401995,
      "grad_norm": 27.567651748657227,
      "learning_rate": 1.7297297297297298e-06,
      "loss": 0.3107,
      "step": 32
    },
    {
      "epoch": 0.054817275747508304,
      "grad_norm": 26.41943359375,
      "learning_rate": 1.7837837837837836e-06,
      "loss": 0.3461,
      "step": 33
    },
    {
      "epoch": 0.05647840531561462,
      "grad_norm": 20.003856658935547,
      "learning_rate": 1.837837837837838e-06,
      "loss": 0.3131,
      "step": 34
    },
    {
      "epoch": 0.05813953488372093,
      "grad_norm": 17.432525634765625,
      "learning_rate": 1.8918918918918918e-06,
      "loss": 0.2034,
      "step": 35
    },
    {
      "epoch": 0.059800664451827246,
      "grad_norm": 18.504539489746094,
      "learning_rate": 1.945945945945946e-06,
      "loss": 0.2837,
      "step": 36
    },
    {
      "epoch": 0.061461794019933555,
      "grad_norm": 15.921092987060547,
      "learning_rate": 2e-06,
      "loss": 0.2069,
      "step": 37
    },
    {
      "epoch": 0.06312292358803986,
      "grad_norm": 17.68560791015625,
      "learning_rate": 1.999996376504091e-06,
      "loss": 0.2074,
      "step": 38
    },
    {
      "epoch": 0.06478405315614617,
      "grad_norm": 18.591819763183594,
      "learning_rate": 1.9999855060426223e-06,
      "loss": 0.3209,
      "step": 39
    },
    {
      "epoch": 0.0664451827242525,
      "grad_norm": 21.39113998413086,
      "learning_rate": 1.9999673886943732e-06,
      "loss": 0.2902,
      "step": 40
    },
    {
      "epoch": 0.0681063122923588,
      "grad_norm": 29.161853790283203,
      "learning_rate": 1.9999420245906396e-06,
      "loss": 0.306,
      "step": 41
    },
    {
      "epoch": 0.06976744186046512,
      "grad_norm": 12.444762229919434,
      "learning_rate": 1.9999094139152346e-06,
      "loss": 0.2314,
      "step": 42
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 25.82412338256836,
      "learning_rate": 1.999869556904488e-06,
      "loss": 0.3418,
      "step": 43
    },
    {
      "epoch": 0.07308970099667775,
      "grad_norm": 18.117881774902344,
      "learning_rate": 1.9998224538472425e-06,
      "loss": 0.2559,
      "step": 44
    },
    {
      "epoch": 0.07475083056478406,
      "grad_norm": 17.093626022338867,
      "learning_rate": 1.999768105084854e-06,
      "loss": 0.2327,
      "step": 45
    },
    {
      "epoch": 0.07641196013289037,
      "grad_norm": 19.649778366088867,
      "learning_rate": 1.999706511011188e-06,
      "loss": 0.2659,
      "step": 46
    },
    {
      "epoch": 0.07807308970099668,
      "grad_norm": 20.985107421875,
      "learning_rate": 1.999637672072616e-06,
      "loss": 0.3078,
      "step": 47
    },
    {
      "epoch": 0.07973421926910298,
      "grad_norm": 14.487998962402344,
      "learning_rate": 1.9995615887680127e-06,
      "loss": 0.2029,
      "step": 48
    },
    {
      "epoch": 0.08139534883720931,
      "grad_norm": 12.239130020141602,
      "learning_rate": 1.9994782616487534e-06,
      "loss": 0.2548,
      "step": 49
    },
    {
      "epoch": 0.08305647840531562,
      "grad_norm": 12.82898998260498,
      "learning_rate": 1.9993876913187095e-06,
      "loss": 0.2234,
      "step": 50
    },
    {
      "epoch": 0.08471760797342193,
      "grad_norm": 14.681973457336426,
      "learning_rate": 1.9992898784342433e-06,
      "loss": 0.2229,
      "step": 51
    },
    {
      "epoch": 0.08637873754152824,
      "grad_norm": 19.037193298339844,
      "learning_rate": 1.9991848237042032e-06,
      "loss": 0.2545,
      "step": 52
    },
    {
      "epoch": 0.08803986710963455,
      "grad_norm": 12.438526153564453,
      "learning_rate": 1.999072527889921e-06,
      "loss": 0.2018,
      "step": 53
    },
    {
      "epoch": 0.08970099667774087,
      "grad_norm": 13.439962387084961,
      "learning_rate": 1.9989529918052027e-06,
      "loss": 0.1708,
      "step": 54
    },
    {
      "epoch": 0.09136212624584718,
      "grad_norm": 17.36454200744629,
      "learning_rate": 1.998826216316326e-06,
      "loss": 0.265,
      "step": 55
    },
    {
      "epoch": 0.09302325581395349,
      "grad_norm": 22.255950927734375,
      "learning_rate": 1.998692202342032e-06,
      "loss": 0.3088,
      "step": 56
    },
    {
      "epoch": 0.0946843853820598,
      "grad_norm": 16.178789138793945,
      "learning_rate": 1.998550950853518e-06,
      "loss": 0.2184,
      "step": 57
    },
    {
      "epoch": 0.09634551495016612,
      "grad_norm": 12.416084289550781,
      "learning_rate": 1.998402462874433e-06,
      "loss": 0.1486,
      "step": 58
    },
    {
      "epoch": 0.09800664451827243,
      "grad_norm": 14.603720664978027,
      "learning_rate": 1.9982467394808674e-06,
      "loss": 0.2007,
      "step": 59
    },
    {
      "epoch": 0.09966777408637874,
      "grad_norm": 15.751521110534668,
      "learning_rate": 1.9980837818013486e-06,
      "loss": 0.261,
      "step": 60
    },
    {
      "epoch": 0.10132890365448505,
      "grad_norm": 11.762495040893555,
      "learning_rate": 1.9979135910168287e-06,
      "loss": 0.1878,
      "step": 61
    },
    {
      "epoch": 0.10299003322259136,
      "grad_norm": 11.098094940185547,
      "learning_rate": 1.997736168360679e-06,
      "loss": 0.1474,
      "step": 62
    },
    {
      "epoch": 0.10465116279069768,
      "grad_norm": 17.337011337280273,
      "learning_rate": 1.9975515151186805e-06,
      "loss": 0.2041,
      "step": 63
    },
    {
      "epoch": 0.10631229235880399,
      "grad_norm": 25.151264190673828,
      "learning_rate": 1.9973596326290133e-06,
      "loss": 0.254,
      "step": 64
    },
    {
      "epoch": 0.1079734219269103,
      "grad_norm": 17.115270614624023,
      "learning_rate": 1.9971605222822486e-06,
      "loss": 0.1631,
      "step": 65
    },
    {
      "epoch": 0.10963455149501661,
      "grad_norm": 11.817288398742676,
      "learning_rate": 1.996954185521337e-06,
      "loss": 0.1729,
      "step": 66
    },
    {
      "epoch": 0.11129568106312292,
      "grad_norm": 14.577704429626465,
      "learning_rate": 1.9967406238415996e-06,
      "loss": 0.1962,
      "step": 67
    },
    {
      "epoch": 0.11295681063122924,
      "grad_norm": 18.282690048217773,
      "learning_rate": 1.996519838790716e-06,
      "loss": 0.2224,
      "step": 68
    },
    {
      "epoch": 0.11461794019933555,
      "grad_norm": 14.57044792175293,
      "learning_rate": 1.996291831968714e-06,
      "loss": 0.149,
      "step": 69
    },
    {
      "epoch": 0.11627906976744186,
      "grad_norm": 19.73041343688965,
      "learning_rate": 1.996056605027956e-06,
      "loss": 0.2655,
      "step": 70
    },
    {
      "epoch": 0.11794019933554817,
      "grad_norm": 18.137643814086914,
      "learning_rate": 1.995814159673132e-06,
      "loss": 0.229,
      "step": 71
    },
    {
      "epoch": 0.11960132890365449,
      "grad_norm": 15.201904296875,
      "learning_rate": 1.995564497661239e-06,
      "loss": 0.1941,
      "step": 72
    },
    {
      "epoch": 0.1212624584717608,
      "grad_norm": 15.047739028930664,
      "learning_rate": 1.9953076208015772e-06,
      "loss": 0.1623,
      "step": 73
    },
    {
      "epoch": 0.12292358803986711,
      "grad_norm": 16.15547752380371,
      "learning_rate": 1.9950435309557302e-06,
      "loss": 0.1894,
      "step": 74
    },
    {
      "epoch": 0.12458471760797342,
      "grad_norm": 16.512256622314453,
      "learning_rate": 1.994772230037556e-06,
      "loss": 0.1993,
      "step": 75
    },
    {
      "epoch": 0.12624584717607973,
      "grad_norm": 14.934798240661621,
      "learning_rate": 1.994493720013169e-06,
      "loss": 0.1411,
      "step": 76
    },
    {
      "epoch": 0.12790697674418605,
      "grad_norm": 14.899330139160156,
      "learning_rate": 1.9942080029009296e-06,
      "loss": 0.185,
      "step": 77
    },
    {
      "epoch": 0.12956810631229235,
      "grad_norm": 17.317094802856445,
      "learning_rate": 1.993915080771427e-06,
      "loss": 0.1947,
      "step": 78
    },
    {
      "epoch": 0.13122923588039867,
      "grad_norm": 12.058066368103027,
      "learning_rate": 1.9936149557474663e-06,
      "loss": 0.1272,
      "step": 79
    },
    {
      "epoch": 0.132890365448505,
      "grad_norm": 12.605445861816406,
      "learning_rate": 1.9933076300040505e-06,
      "loss": 0.1624,
      "step": 80
    },
    {
      "epoch": 0.1345514950166113,
      "grad_norm": 15.506592750549316,
      "learning_rate": 1.9929931057683666e-06,
      "loss": 0.1833,
      "step": 81
    },
    {
      "epoch": 0.1362126245847176,
      "grad_norm": 11.711981773376465,
      "learning_rate": 1.9926713853197696e-06,
      "loss": 0.165,
      "step": 82
    },
    {
      "epoch": 0.1378737541528239,
      "grad_norm": 13.009001731872559,
      "learning_rate": 1.9923424709897644e-06,
      "loss": 0.2145,
      "step": 83
    },
    {
      "epoch": 0.13953488372093023,
      "grad_norm": 14.122442245483398,
      "learning_rate": 1.992006365161991e-06,
      "loss": 0.1774,
      "step": 84
    },
    {
      "epoch": 0.14119601328903655,
      "grad_norm": 9.544992446899414,
      "learning_rate": 1.991663070272206e-06,
      "loss": 0.1555,
      "step": 85
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 15.784918785095215,
      "learning_rate": 1.9913125888082632e-06,
      "loss": 0.2485,
      "step": 86
    },
    {
      "epoch": 0.14451827242524917,
      "grad_norm": 14.848572731018066,
      "learning_rate": 1.9909549233100998e-06,
      "loss": 0.1986,
      "step": 87
    },
    {
      "epoch": 0.1461794019933555,
      "grad_norm": 19.732912063598633,
      "learning_rate": 1.990590076369715e-06,
      "loss": 0.2783,
      "step": 88
    },
    {
      "epoch": 0.1478405315614618,
      "grad_norm": 10.916641235351562,
      "learning_rate": 1.9902180506311514e-06,
      "loss": 0.1609,
      "step": 89
    },
    {
      "epoch": 0.14950166112956811,
      "grad_norm": 7.3824782371521,
      "learning_rate": 1.9898388487904764e-06,
      "loss": 0.0938,
      "step": 90
    },
    {
      "epoch": 0.1511627906976744,
      "grad_norm": 20.201457977294922,
      "learning_rate": 1.989452473595762e-06,
      "loss": 0.2579,
      "step": 91
    },
    {
      "epoch": 0.15282392026578073,
      "grad_norm": 11.529504776000977,
      "learning_rate": 1.989058927847067e-06,
      "loss": 0.1761,
      "step": 92
    },
    {
      "epoch": 0.15448504983388706,
      "grad_norm": 12.745220184326172,
      "learning_rate": 1.9886582143964143e-06,
      "loss": 0.1881,
      "step": 93
    },
    {
      "epoch": 0.15614617940199335,
      "grad_norm": 8.80517292022705,
      "learning_rate": 1.9882503361477703e-06,
      "loss": 0.1003,
      "step": 94
    },
    {
      "epoch": 0.15780730897009967,
      "grad_norm": 16.047895431518555,
      "learning_rate": 1.9878352960570256e-06,
      "loss": 0.1476,
      "step": 95
    },
    {
      "epoch": 0.15946843853820597,
      "grad_norm": 13.64163875579834,
      "learning_rate": 1.987413097131972e-06,
      "loss": 0.2103,
      "step": 96
    },
    {
      "epoch": 0.1611295681063123,
      "grad_norm": 10.822211265563965,
      "learning_rate": 1.9869837424322827e-06,
      "loss": 0.1423,
      "step": 97
    },
    {
      "epoch": 0.16279069767441862,
      "grad_norm": 11.291969299316406,
      "learning_rate": 1.9865472350694867e-06,
      "loss": 0.1375,
      "step": 98
    },
    {
      "epoch": 0.1644518272425249,
      "grad_norm": 10.820338249206543,
      "learning_rate": 1.9861035782069496e-06,
      "loss": 0.18,
      "step": 99
    },
    {
      "epoch": 0.16611295681063123,
      "grad_norm": 12.342602729797363,
      "learning_rate": 1.985652775059849e-06,
      "loss": 0.123,
      "step": 100
    },
    {
      "epoch": 0.16777408637873753,
      "grad_norm": 11.896181106567383,
      "learning_rate": 1.985194828895152e-06,
      "loss": 0.1846,
      "step": 101
    },
    {
      "epoch": 0.16943521594684385,
      "grad_norm": 12.606119155883789,
      "learning_rate": 1.9847297430315903e-06,
      "loss": 0.1322,
      "step": 102
    },
    {
      "epoch": 0.17109634551495018,
      "grad_norm": 10.485321044921875,
      "learning_rate": 1.984257520839637e-06,
      "loss": 0.1443,
      "step": 103
    },
    {
      "epoch": 0.17275747508305647,
      "grad_norm": 11.102760314941406,
      "learning_rate": 1.983778165741483e-06,
      "loss": 0.1322,
      "step": 104
    },
    {
      "epoch": 0.1744186046511628,
      "grad_norm": 12.000479698181152,
      "learning_rate": 1.983291681211011e-06,
      "loss": 0.1558,
      "step": 105
    },
    {
      "epoch": 0.1760797342192691,
      "grad_norm": 15.679694175720215,
      "learning_rate": 1.98279807077377e-06,
      "loss": 0.1671,
      "step": 106
    },
    {
      "epoch": 0.1777408637873754,
      "grad_norm": 14.487334251403809,
      "learning_rate": 1.9822973380069507e-06,
      "loss": 0.1675,
      "step": 107
    },
    {
      "epoch": 0.17940199335548174,
      "grad_norm": 12.584924697875977,
      "learning_rate": 1.9817894865393597e-06,
      "loss": 0.1733,
      "step": 108
    },
    {
      "epoch": 0.18106312292358803,
      "grad_norm": 8.71336841583252,
      "learning_rate": 1.9812745200513923e-06,
      "loss": 0.1175,
      "step": 109
    },
    {
      "epoch": 0.18272425249169436,
      "grad_norm": 14.50503158569336,
      "learning_rate": 1.9807524422750064e-06,
      "loss": 0.1571,
      "step": 110
    },
    {
      "epoch": 0.18438538205980065,
      "grad_norm": 12.778355598449707,
      "learning_rate": 1.9802232569936956e-06,
      "loss": 0.1631,
      "step": 111
    },
    {
      "epoch": 0.18604651162790697,
      "grad_norm": 14.556724548339844,
      "learning_rate": 1.979686968042461e-06,
      "loss": 0.2194,
      "step": 112
    },
    {
      "epoch": 0.1877076411960133,
      "grad_norm": 13.824010848999023,
      "learning_rate": 1.979143579307784e-06,
      "loss": 0.1673,
      "step": 113
    },
    {
      "epoch": 0.1893687707641196,
      "grad_norm": 11.80858325958252,
      "learning_rate": 1.9785930947275985e-06,
      "loss": 0.1661,
      "step": 114
    },
    {
      "epoch": 0.19102990033222592,
      "grad_norm": 15.89432144165039,
      "learning_rate": 1.9780355182912623e-06,
      "loss": 0.2334,
      "step": 115
    },
    {
      "epoch": 0.19269102990033224,
      "grad_norm": 11.047170639038086,
      "learning_rate": 1.977470854039527e-06,
      "loss": 0.1506,
      "step": 116
    },
    {
      "epoch": 0.19435215946843853,
      "grad_norm": 12.432543754577637,
      "learning_rate": 1.9768991060645096e-06,
      "loss": 0.1775,
      "step": 117
    },
    {
      "epoch": 0.19601328903654486,
      "grad_norm": 9.709572792053223,
      "learning_rate": 1.976320278509663e-06,
      "loss": 0.1365,
      "step": 118
    },
    {
      "epoch": 0.19767441860465115,
      "grad_norm": 11.285073280334473,
      "learning_rate": 1.9757343755697456e-06,
      "loss": 0.1765,
      "step": 119
    },
    {
      "epoch": 0.19933554817275748,
      "grad_norm": 10.247356414794922,
      "learning_rate": 1.9751414014907913e-06,
      "loss": 0.1558,
      "step": 120
    },
    {
      "epoch": 0.2009966777408638,
      "grad_norm": 14.953573226928711,
      "learning_rate": 1.974541360570079e-06,
      "loss": 0.1802,
      "step": 121
    },
    {
      "epoch": 0.2026578073089701,
      "grad_norm": 14.226633071899414,
      "learning_rate": 1.9739342571560996e-06,
      "loss": 0.1492,
      "step": 122
    },
    {
      "epoch": 0.20431893687707642,
      "grad_norm": 9.25733757019043,
      "learning_rate": 1.973320095648527e-06,
      "loss": 0.1563,
      "step": 123
    },
    {
      "epoch": 0.2059800664451827,
      "grad_norm": 11.132588386535645,
      "learning_rate": 1.9726988804981845e-06,
      "loss": 0.1634,
      "step": 124
    },
    {
      "epoch": 0.20764119601328904,
      "grad_norm": 9.98374080657959,
      "learning_rate": 1.972070616207013e-06,
      "loss": 0.16,
      "step": 125
    },
    {
      "epoch": 0.20930232558139536,
      "grad_norm": 14.084647178649902,
      "learning_rate": 1.971435307328039e-06,
      "loss": 0.2341,
      "step": 126
    },
    {
      "epoch": 0.21096345514950166,
      "grad_norm": 10.617352485656738,
      "learning_rate": 1.9707929584653408e-06,
      "loss": 0.1376,
      "step": 127
    },
    {
      "epoch": 0.21262458471760798,
      "grad_norm": 12.762201309204102,
      "learning_rate": 1.9701435742740146e-06,
      "loss": 0.1543,
      "step": 128
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 11.197344779968262,
      "learning_rate": 1.9694871594601435e-06,
      "loss": 0.131,
      "step": 129
    },
    {
      "epoch": 0.2159468438538206,
      "grad_norm": 24.01343536376953,
      "learning_rate": 1.9688237187807594e-06,
      "loss": 0.2743,
      "step": 130
    },
    {
      "epoch": 0.21760797342192692,
      "grad_norm": 11.367233276367188,
      "learning_rate": 1.9681532570438117e-06,
      "loss": 0.1446,
      "step": 131
    },
    {
      "epoch": 0.21926910299003322,
      "grad_norm": 10.252062797546387,
      "learning_rate": 1.967475779108131e-06,
      "loss": 0.1093,
      "step": 132
    },
    {
      "epoch": 0.22093023255813954,
      "grad_norm": 16.38922691345215,
      "learning_rate": 1.9667912898833952e-06,
      "loss": 0.1282,
      "step": 133
    },
    {
      "epoch": 0.22259136212624583,
      "grad_norm": 14.078662872314453,
      "learning_rate": 1.966099794330091e-06,
      "loss": 0.1436,
      "step": 134
    },
    {
      "epoch": 0.22425249169435216,
      "grad_norm": 12.31657600402832,
      "learning_rate": 1.9654012974594813e-06,
      "loss": 0.1442,
      "step": 135
    },
    {
      "epoch": 0.22591362126245848,
      "grad_norm": 16.30408477783203,
      "learning_rate": 1.9646958043335675e-06,
      "loss": 0.1845,
      "step": 136
    },
    {
      "epoch": 0.22757475083056478,
      "grad_norm": 18.23689842224121,
      "learning_rate": 1.9639833200650524e-06,
      "loss": 0.2381,
      "step": 137
    },
    {
      "epoch": 0.2292358803986711,
      "grad_norm": 14.732518196105957,
      "learning_rate": 1.9632638498173037e-06,
      "loss": 0.1844,
      "step": 138
    },
    {
      "epoch": 0.23089700996677742,
      "grad_norm": 8.49440860748291,
      "learning_rate": 1.9625373988043164e-06,
      "loss": 0.1314,
      "step": 139
    },
    {
      "epoch": 0.23255813953488372,
      "grad_norm": 15.13712215423584,
      "learning_rate": 1.9618039722906746e-06,
      "loss": 0.1931,
      "step": 140
    },
    {
      "epoch": 0.23421926910299004,
      "grad_norm": 18.702835083007812,
      "learning_rate": 1.961063575591515e-06,
      "loss": 0.1516,
      "step": 141
    },
    {
      "epoch": 0.23588039867109634,
      "grad_norm": 6.8990559577941895,
      "learning_rate": 1.960316214072486e-06,
      "loss": 0.0791,
      "step": 142
    },
    {
      "epoch": 0.23754152823920266,
      "grad_norm": 11.587632179260254,
      "learning_rate": 1.9595618931497105e-06,
      "loss": 0.1103,
      "step": 143
    },
    {
      "epoch": 0.23920265780730898,
      "grad_norm": 20.253131866455078,
      "learning_rate": 1.9588006182897455e-06,
      "loss": 0.1988,
      "step": 144
    },
    {
      "epoch": 0.24086378737541528,
      "grad_norm": 15.596393585205078,
      "learning_rate": 1.958032395009545e-06,
      "loss": 0.1549,
      "step": 145
    },
    {
      "epoch": 0.2425249169435216,
      "grad_norm": 10.316424369812012,
      "learning_rate": 1.9572572288764154e-06,
      "loss": 0.1564,
      "step": 146
    },
    {
      "epoch": 0.2441860465116279,
      "grad_norm": 9.298800468444824,
      "learning_rate": 1.95647512550798e-06,
      "loss": 0.1488,
      "step": 147
    },
    {
      "epoch": 0.24584717607973422,
      "grad_norm": 15.457275390625,
      "learning_rate": 1.955686090572136e-06,
      "loss": 0.2725,
      "step": 148
    },
    {
      "epoch": 0.24750830564784054,
      "grad_norm": 13.5877685546875,
      "learning_rate": 1.9548901297870124e-06,
      "loss": 0.2401,
      "step": 149
    },
    {
      "epoch": 0.24916943521594684,
      "grad_norm": 10.776034355163574,
      "learning_rate": 1.95408724892093e-06,
      "loss": 0.149,
      "step": 150
    },
    {
      "epoch": 0.25083056478405313,
      "grad_norm": 13.359519958496094,
      "learning_rate": 1.9532774537923615e-06,
      "loss": 0.1533,
      "step": 151
    },
    {
      "epoch": 0.25249169435215946,
      "grad_norm": 10.034915924072266,
      "learning_rate": 1.9524607502698843e-06,
      "loss": 0.1987,
      "step": 152
    },
    {
      "epoch": 0.2541528239202658,
      "grad_norm": 11.459542274475098,
      "learning_rate": 1.9516371442721425e-06,
      "loss": 0.1631,
      "step": 153
    },
    {
      "epoch": 0.2558139534883721,
      "grad_norm": 9.19011402130127,
      "learning_rate": 1.9508066417678018e-06,
      "loss": 0.1789,
      "step": 154
    },
    {
      "epoch": 0.2574750830564784,
      "grad_norm": 9.150303840637207,
      "learning_rate": 1.9499692487755076e-06,
      "loss": 0.1464,
      "step": 155
    },
    {
      "epoch": 0.2591362126245847,
      "grad_norm": 10.107464790344238,
      "learning_rate": 1.9491249713638394e-06,
      "loss": 0.1705,
      "step": 156
    },
    {
      "epoch": 0.260797342192691,
      "grad_norm": 7.740807056427002,
      "learning_rate": 1.948273815651269e-06,
      "loss": 0.1257,
      "step": 157
    },
    {
      "epoch": 0.26245847176079734,
      "grad_norm": 8.180129051208496,
      "learning_rate": 1.947415787806115e-06,
      "loss": 0.09,
      "step": 158
    },
    {
      "epoch": 0.26411960132890366,
      "grad_norm": 12.505861282348633,
      "learning_rate": 1.946550894046498e-06,
      "loss": 0.1999,
      "step": 159
    },
    {
      "epoch": 0.26578073089701,
      "grad_norm": 16.3990535736084,
      "learning_rate": 1.9456791406402963e-06,
      "loss": 0.2219,
      "step": 160
    },
    {
      "epoch": 0.26744186046511625,
      "grad_norm": 23.716087341308594,
      "learning_rate": 1.944800533905099e-06,
      "loss": 0.2294,
      "step": 161
    },
    {
      "epoch": 0.2691029900332226,
      "grad_norm": 23.61831283569336,
      "learning_rate": 1.943915080208163e-06,
      "loss": 0.2393,
      "step": 162
    },
    {
      "epoch": 0.2707641196013289,
      "grad_norm": 17.86264991760254,
      "learning_rate": 1.943022785966363e-06,
      "loss": 0.1994,
      "step": 163
    },
    {
      "epoch": 0.2724252491694352,
      "grad_norm": 10.839683532714844,
      "learning_rate": 1.9421236576461487e-06,
      "loss": 0.0614,
      "step": 164
    },
    {
      "epoch": 0.27408637873754155,
      "grad_norm": 10.387090682983398,
      "learning_rate": 1.941217701763495e-06,
      "loss": 0.1148,
      "step": 165
    },
    {
      "epoch": 0.2757475083056478,
      "grad_norm": 22.82048797607422,
      "learning_rate": 1.9403049248838576e-06,
      "loss": 0.2027,
      "step": 166
    },
    {
      "epoch": 0.27740863787375414,
      "grad_norm": 12.700292587280273,
      "learning_rate": 1.9393853336221225e-06,
      "loss": 0.1516,
      "step": 167
    },
    {
      "epoch": 0.27906976744186046,
      "grad_norm": 16.656648635864258,
      "learning_rate": 1.9384589346425605e-06,
      "loss": 0.2115,
      "step": 168
    },
    {
      "epoch": 0.2807308970099668,
      "grad_norm": 20.727684020996094,
      "learning_rate": 1.9375257346587774e-06,
      "loss": 0.1832,
      "step": 169
    },
    {
      "epoch": 0.2823920265780731,
      "grad_norm": 12.764060020446777,
      "learning_rate": 1.936585740433665e-06,
      "loss": 0.1757,
      "step": 170
    },
    {
      "epoch": 0.2840531561461794,
      "grad_norm": 17.68997573852539,
      "learning_rate": 1.9356389587793544e-06,
      "loss": 0.1494,
      "step": 171
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 11.008881568908691,
      "learning_rate": 1.934685396557165e-06,
      "loss": 0.1242,
      "step": 172
    },
    {
      "epoch": 0.287375415282392,
      "grad_norm": 8.055181503295898,
      "learning_rate": 1.9337250606775536e-06,
      "loss": 0.1221,
      "step": 173
    },
    {
      "epoch": 0.28903654485049834,
      "grad_norm": 9.147614479064941,
      "learning_rate": 1.9327579581000665e-06,
      "loss": 0.1163,
      "step": 174
    },
    {
      "epoch": 0.29069767441860467,
      "grad_norm": 9.311761856079102,
      "learning_rate": 1.931784095833289e-06,
      "loss": 0.1026,
      "step": 175
    },
    {
      "epoch": 0.292358803986711,
      "grad_norm": 7.291942119598389,
      "learning_rate": 1.930803480934792e-06,
      "loss": 0.1058,
      "step": 176
    },
    {
      "epoch": 0.29401993355481726,
      "grad_norm": 10.639616012573242,
      "learning_rate": 1.9298161205110838e-06,
      "loss": 0.1921,
      "step": 177
    },
    {
      "epoch": 0.2956810631229236,
      "grad_norm": 12.382287979125977,
      "learning_rate": 1.9288220217175583e-06,
      "loss": 0.1336,
      "step": 178
    },
    {
      "epoch": 0.2973421926910299,
      "grad_norm": 9.517138481140137,
      "learning_rate": 1.92782119175844e-06,
      "loss": 0.1173,
      "step": 179
    },
    {
      "epoch": 0.29900332225913623,
      "grad_norm": 10.612733840942383,
      "learning_rate": 1.9268136378867365e-06,
      "loss": 0.1215,
      "step": 180
    },
    {
      "epoch": 0.30066445182724255,
      "grad_norm": 11.6936616897583,
      "learning_rate": 1.9257993674041814e-06,
      "loss": 0.1511,
      "step": 181
    },
    {
      "epoch": 0.3023255813953488,
      "grad_norm": 9.376453399658203,
      "learning_rate": 1.9247783876611857e-06,
      "loss": 0.0859,
      "step": 182
    },
    {
      "epoch": 0.30398671096345514,
      "grad_norm": 15.27945327758789,
      "learning_rate": 1.9237507060567802e-06,
      "loss": 0.2082,
      "step": 183
    },
    {
      "epoch": 0.30564784053156147,
      "grad_norm": 13.213935852050781,
      "learning_rate": 1.9227163300385662e-06,
      "loss": 0.1624,
      "step": 184
    },
    {
      "epoch": 0.3073089700996678,
      "grad_norm": 14.251786231994629,
      "learning_rate": 1.921675267102657e-06,
      "loss": 0.1719,
      "step": 185
    },
    {
      "epoch": 0.3089700996677741,
      "grad_norm": 21.97607421875,
      "learning_rate": 1.920627524793628e-06,
      "loss": 0.2718,
      "step": 186
    },
    {
      "epoch": 0.3106312292358804,
      "grad_norm": 12.712850570678711,
      "learning_rate": 1.9195731107044594e-06,
      "loss": 0.1694,
      "step": 187
    },
    {
      "epoch": 0.3122923588039867,
      "grad_norm": 7.244543075561523,
      "learning_rate": 1.9185120324764805e-06,
      "loss": 0.1159,
      "step": 188
    },
    {
      "epoch": 0.313953488372093,
      "grad_norm": 8.671590805053711,
      "learning_rate": 1.917444297799317e-06,
      "loss": 0.1121,
      "step": 189
    },
    {
      "epoch": 0.31561461794019935,
      "grad_norm": 8.755648612976074,
      "learning_rate": 1.916369914410834e-06,
      "loss": 0.1329,
      "step": 190
    },
    {
      "epoch": 0.31727574750830567,
      "grad_norm": 9.643033027648926,
      "learning_rate": 1.9152888900970783e-06,
      "loss": 0.1697,
      "step": 191
    },
    {
      "epoch": 0.31893687707641194,
      "grad_norm": 10.84849739074707,
      "learning_rate": 1.9142012326922246e-06,
      "loss": 0.1537,
      "step": 192
    },
    {
      "epoch": 0.32059800664451826,
      "grad_norm": 8.020485877990723,
      "learning_rate": 1.913106950078517e-06,
      "loss": 0.0776,
      "step": 193
    },
    {
      "epoch": 0.3222591362126246,
      "grad_norm": 14.385054588317871,
      "learning_rate": 1.9120060501862128e-06,
      "loss": 0.1708,
      "step": 194
    },
    {
      "epoch": 0.3239202657807309,
      "grad_norm": 12.023110389709473,
      "learning_rate": 1.9108985409935247e-06,
      "loss": 0.14,
      "step": 195
    },
    {
      "epoch": 0.32558139534883723,
      "grad_norm": 20.190410614013672,
      "learning_rate": 1.9097844305265622e-06,
      "loss": 0.1881,
      "step": 196
    },
    {
      "epoch": 0.3272425249169435,
      "grad_norm": 13.03503704071045,
      "learning_rate": 1.908663726859275e-06,
      "loss": 0.1258,
      "step": 197
    },
    {
      "epoch": 0.3289036544850498,
      "grad_norm": 12.031476020812988,
      "learning_rate": 1.907536438113394e-06,
      "loss": 0.1018,
      "step": 198
    },
    {
      "epoch": 0.33056478405315615,
      "grad_norm": 12.906877517700195,
      "learning_rate": 1.9064025724583707e-06,
      "loss": 0.1026,
      "step": 199
    },
    {
      "epoch": 0.33222591362126247,
      "grad_norm": 12.325488090515137,
      "learning_rate": 1.905262138111321e-06,
      "loss": 0.1089,
      "step": 200
    },
    {
      "epoch": 0.3338870431893688,
      "grad_norm": 9.743961334228516,
      "learning_rate": 1.9041151433369623e-06,
      "loss": 0.0776,
      "step": 201
    },
    {
      "epoch": 0.33554817275747506,
      "grad_norm": 22.17953109741211,
      "learning_rate": 1.902961596447557e-06,
      "loss": 0.2423,
      "step": 202
    },
    {
      "epoch": 0.3372093023255814,
      "grad_norm": 8.794547080993652,
      "learning_rate": 1.90180150580285e-06,
      "loss": 0.1216,
      "step": 203
    },
    {
      "epoch": 0.3388704318936877,
      "grad_norm": 8.961956977844238,
      "learning_rate": 1.9006348798100086e-06,
      "loss": 0.1306,
      "step": 204
    },
    {
      "epoch": 0.34053156146179403,
      "grad_norm": 12.606147766113281,
      "learning_rate": 1.8994617269235614e-06,
      "loss": 0.1489,
      "step": 205
    },
    {
      "epoch": 0.34219269102990035,
      "grad_norm": 15.009194374084473,
      "learning_rate": 1.8982820556453384e-06,
      "loss": 0.1878,
      "step": 206
    },
    {
      "epoch": 0.3438538205980066,
      "grad_norm": 8.010025024414062,
      "learning_rate": 1.8970958745244075e-06,
      "loss": 0.1143,
      "step": 207
    },
    {
      "epoch": 0.34551495016611294,
      "grad_norm": 9.553529739379883,
      "learning_rate": 1.8959031921570135e-06,
      "loss": 0.1271,
      "step": 208
    },
    {
      "epoch": 0.34717607973421927,
      "grad_norm": 7.424415588378906,
      "learning_rate": 1.8947040171865157e-06,
      "loss": 0.0986,
      "step": 209
    },
    {
      "epoch": 0.3488372093023256,
      "grad_norm": 7.113452434539795,
      "learning_rate": 1.8934983583033252e-06,
      "loss": 0.1211,
      "step": 210
    },
    {
      "epoch": 0.3504983388704319,
      "grad_norm": 8.20436954498291,
      "learning_rate": 1.8922862242448428e-06,
      "loss": 0.1339,
      "step": 211
    },
    {
      "epoch": 0.3521594684385382,
      "grad_norm": 11.628143310546875,
      "learning_rate": 1.8910676237953931e-06,
      "loss": 0.2292,
      "step": 212
    },
    {
      "epoch": 0.3538205980066445,
      "grad_norm": 11.920955657958984,
      "learning_rate": 1.8898425657861643e-06,
      "loss": 0.1387,
      "step": 213
    },
    {
      "epoch": 0.3554817275747508,
      "grad_norm": 7.538236141204834,
      "learning_rate": 1.8886110590951415e-06,
      "loss": 0.1015,
      "step": 214
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 12.414472579956055,
      "learning_rate": 1.8873731126470434e-06,
      "loss": 0.1664,
      "step": 215
    },
    {
      "epoch": 0.3588039867109635,
      "grad_norm": 9.726411819458008,
      "learning_rate": 1.8861287354132582e-06,
      "loss": 0.0867,
      "step": 216
    },
    {
      "epoch": 0.36046511627906974,
      "grad_norm": 17.113433837890625,
      "learning_rate": 1.8848779364117772e-06,
      "loss": 0.2126,
      "step": 217
    },
    {
      "epoch": 0.36212624584717606,
      "grad_norm": 10.119791030883789,
      "learning_rate": 1.8836207247071307e-06,
      "loss": 0.1147,
      "step": 218
    },
    {
      "epoch": 0.3637873754152824,
      "grad_norm": 7.584885597229004,
      "learning_rate": 1.882357109410322e-06,
      "loss": 0.0545,
      "step": 219
    },
    {
      "epoch": 0.3654485049833887,
      "grad_norm": 7.976661682128906,
      "learning_rate": 1.8810870996787599e-06,
      "loss": 0.0746,
      "step": 220
    },
    {
      "epoch": 0.36710963455149503,
      "grad_norm": 16.05130958557129,
      "learning_rate": 1.8798107047161952e-06,
      "loss": 0.0874,
      "step": 221
    },
    {
      "epoch": 0.3687707641196013,
      "grad_norm": 16.22881507873535,
      "learning_rate": 1.8785279337726517e-06,
      "loss": 0.2398,
      "step": 222
    },
    {
      "epoch": 0.3704318936877076,
      "grad_norm": 15.935908317565918,
      "learning_rate": 1.87723879614436e-06,
      "loss": 0.1306,
      "step": 223
    },
    {
      "epoch": 0.37209302325581395,
      "grad_norm": 20.93619155883789,
      "learning_rate": 1.8759433011736894e-06,
      "loss": 0.1476,
      "step": 224
    },
    {
      "epoch": 0.37375415282392027,
      "grad_norm": 8.026069641113281,
      "learning_rate": 1.8746414582490822e-06,
      "loss": 0.1122,
      "step": 225
    },
    {
      "epoch": 0.3754152823920266,
      "grad_norm": 11.861761093139648,
      "learning_rate": 1.8733332768049826e-06,
      "loss": 0.1487,
      "step": 226
    },
    {
      "epoch": 0.3770764119601329,
      "grad_norm": 13.299816131591797,
      "learning_rate": 1.8720187663217712e-06,
      "loss": 0.1201,
      "step": 227
    },
    {
      "epoch": 0.3787375415282392,
      "grad_norm": 12.491223335266113,
      "learning_rate": 1.870697936325695e-06,
      "loss": 0.169,
      "step": 228
    },
    {
      "epoch": 0.3803986710963455,
      "grad_norm": 12.5487060546875,
      "learning_rate": 1.8693707963887976e-06,
      "loss": 0.1829,
      "step": 229
    },
    {
      "epoch": 0.38205980066445183,
      "grad_norm": 14.099169731140137,
      "learning_rate": 1.8680373561288519e-06,
      "loss": 0.1489,
      "step": 230
    },
    {
      "epoch": 0.38372093023255816,
      "grad_norm": 6.0192718505859375,
      "learning_rate": 1.8666976252092877e-06,
      "loss": 0.1093,
      "step": 231
    },
    {
      "epoch": 0.3853820598006645,
      "grad_norm": 8.559925079345703,
      "learning_rate": 1.8653516133391248e-06,
      "loss": 0.1205,
      "step": 232
    },
    {
      "epoch": 0.38704318936877075,
      "grad_norm": 7.076784610748291,
      "learning_rate": 1.8639993302728998e-06,
      "loss": 0.1206,
      "step": 233
    },
    {
      "epoch": 0.38870431893687707,
      "grad_norm": 6.830137729644775,
      "learning_rate": 1.8626407858105972e-06,
      "loss": 0.1168,
      "step": 234
    },
    {
      "epoch": 0.3903654485049834,
      "grad_norm": 10.434679985046387,
      "learning_rate": 1.8612759897975776e-06,
      "loss": 0.1848,
      "step": 235
    },
    {
      "epoch": 0.3920265780730897,
      "grad_norm": 11.184035301208496,
      "learning_rate": 1.8599049521245066e-06,
      "loss": 0.1652,
      "step": 236
    },
    {
      "epoch": 0.39368770764119604,
      "grad_norm": 10.515231132507324,
      "learning_rate": 1.8585276827272827e-06,
      "loss": 0.1524,
      "step": 237
    },
    {
      "epoch": 0.3953488372093023,
      "grad_norm": 7.663083076477051,
      "learning_rate": 1.8571441915869662e-06,
      "loss": 0.1314,
      "step": 238
    },
    {
      "epoch": 0.39700996677740863,
      "grad_norm": 7.605090618133545,
      "learning_rate": 1.8557544887297061e-06,
      "loss": 0.1261,
      "step": 239
    },
    {
      "epoch": 0.39867109634551495,
      "grad_norm": 8.647713661193848,
      "learning_rate": 1.8543585842266674e-06,
      "loss": 0.1128,
      "step": 240
    },
    {
      "epoch": 0.4003322259136213,
      "grad_norm": 13.598785400390625,
      "learning_rate": 1.8529564881939587e-06,
      "loss": 0.1606,
      "step": 241
    },
    {
      "epoch": 0.4019933554817276,
      "grad_norm": 10.329269409179688,
      "learning_rate": 1.8515482107925587e-06,
      "loss": 0.1587,
      "step": 242
    },
    {
      "epoch": 0.40365448504983387,
      "grad_norm": 21.867544174194336,
      "learning_rate": 1.850133762228242e-06,
      "loss": 0.0797,
      "step": 243
    },
    {
      "epoch": 0.4053156146179402,
      "grad_norm": 8.419551849365234,
      "learning_rate": 1.8487131527515059e-06,
      "loss": 0.0986,
      "step": 244
    },
    {
      "epoch": 0.4069767441860465,
      "grad_norm": 17.915225982666016,
      "learning_rate": 1.8472863926574954e-06,
      "loss": 0.1774,
      "step": 245
    },
    {
      "epoch": 0.40863787375415284,
      "grad_norm": 16.778846740722656,
      "learning_rate": 1.8458534922859294e-06,
      "loss": 0.2271,
      "step": 246
    },
    {
      "epoch": 0.41029900332225916,
      "grad_norm": 11.009391784667969,
      "learning_rate": 1.8444144620210253e-06,
      "loss": 0.1626,
      "step": 247
    },
    {
      "epoch": 0.4119601328903654,
      "grad_norm": 14.870036125183105,
      "learning_rate": 1.8429693122914234e-06,
      "loss": 0.1578,
      "step": 248
    },
    {
      "epoch": 0.41362126245847175,
      "grad_norm": 13.489202499389648,
      "learning_rate": 1.8415180535701125e-06,
      "loss": 0.1642,
      "step": 249
    },
    {
      "epoch": 0.4152823920265781,
      "grad_norm": 7.866860866546631,
      "learning_rate": 1.8400606963743516e-06,
      "loss": 0.1444,
      "step": 250
    },
    {
      "epoch": 0.4169435215946844,
      "grad_norm": 11.771472930908203,
      "learning_rate": 1.8385972512655977e-06,
      "loss": 0.1617,
      "step": 251
    },
    {
      "epoch": 0.4186046511627907,
      "grad_norm": 11.763538360595703,
      "learning_rate": 1.8371277288494245e-06,
      "loss": 0.1606,
      "step": 252
    },
    {
      "epoch": 0.420265780730897,
      "grad_norm": 6.283024311065674,
      "learning_rate": 1.8356521397754493e-06,
      "loss": 0.1329,
      "step": 253
    },
    {
      "epoch": 0.4219269102990033,
      "grad_norm": 7.202626705169678,
      "learning_rate": 1.8341704947372544e-06,
      "loss": 0.1043,
      "step": 254
    },
    {
      "epoch": 0.42358803986710963,
      "grad_norm": 9.756691932678223,
      "learning_rate": 1.8326828044723085e-06,
      "loss": 0.1469,
      "step": 255
    },
    {
      "epoch": 0.42524916943521596,
      "grad_norm": 9.103363037109375,
      "learning_rate": 1.8311890797618913e-06,
      "loss": 0.1666,
      "step": 256
    },
    {
      "epoch": 0.4269102990033223,
      "grad_norm": 7.448887825012207,
      "learning_rate": 1.8296893314310138e-06,
      "loss": 0.1172,
      "step": 257
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 9.971731185913086,
      "learning_rate": 1.8281835703483392e-06,
      "loss": 0.1809,
      "step": 258
    },
    {
      "epoch": 0.43023255813953487,
      "grad_norm": 7.47666597366333,
      "learning_rate": 1.826671807426106e-06,
      "loss": 0.1046,
      "step": 259
    },
    {
      "epoch": 0.4318936877076412,
      "grad_norm": 9.57524299621582,
      "learning_rate": 1.8251540536200482e-06,
      "loss": 0.1566,
      "step": 260
    },
    {
      "epoch": 0.4335548172757475,
      "grad_norm": 7.479125022888184,
      "learning_rate": 1.8236303199293145e-06,
      "loss": 0.075,
      "step": 261
    },
    {
      "epoch": 0.43521594684385384,
      "grad_norm": 10.383878707885742,
      "learning_rate": 1.822100617396391e-06,
      "loss": 0.1494,
      "step": 262
    },
    {
      "epoch": 0.4368770764119601,
      "grad_norm": 11.652271270751953,
      "learning_rate": 1.8205649571070193e-06,
      "loss": 0.103,
      "step": 263
    },
    {
      "epoch": 0.43853820598006643,
      "grad_norm": 8.859437942504883,
      "learning_rate": 1.8190233501901166e-06,
      "loss": 0.0748,
      "step": 264
    },
    {
      "epoch": 0.44019933554817275,
      "grad_norm": 16.176801681518555,
      "learning_rate": 1.8174758078176961e-06,
      "loss": 0.2191,
      "step": 265
    },
    {
      "epoch": 0.4418604651162791,
      "grad_norm": 14.912988662719727,
      "learning_rate": 1.815922341204785e-06,
      "loss": 0.1775,
      "step": 266
    },
    {
      "epoch": 0.4435215946843854,
      "grad_norm": 18.418682098388672,
      "learning_rate": 1.8143629616093419e-06,
      "loss": 0.2055,
      "step": 267
    },
    {
      "epoch": 0.44518272425249167,
      "grad_norm": 21.120506286621094,
      "learning_rate": 1.812797680332179e-06,
      "loss": 0.205,
      "step": 268
    },
    {
      "epoch": 0.446843853820598,
      "grad_norm": 22.294757843017578,
      "learning_rate": 1.811226508716877e-06,
      "loss": 0.2917,
      "step": 269
    },
    {
      "epoch": 0.4485049833887043,
      "grad_norm": 19.390213012695312,
      "learning_rate": 1.8096494581497032e-06,
      "loss": 0.183,
      "step": 270
    },
    {
      "epoch": 0.45016611295681064,
      "grad_norm": 5.7725629806518555,
      "learning_rate": 1.8080665400595302e-06,
      "loss": 0.0768,
      "step": 271
    },
    {
      "epoch": 0.45182724252491696,
      "grad_norm": 10.425414085388184,
      "learning_rate": 1.806477765917753e-06,
      "loss": 0.1679,
      "step": 272
    },
    {
      "epoch": 0.45348837209302323,
      "grad_norm": 8.793322563171387,
      "learning_rate": 1.804883147238204e-06,
      "loss": 0.1279,
      "step": 273
    },
    {
      "epoch": 0.45514950166112955,
      "grad_norm": 9.122625350952148,
      "learning_rate": 1.803282695577072e-06,
      "loss": 0.1291,
      "step": 274
    },
    {
      "epoch": 0.4568106312292359,
      "grad_norm": 14.53525161743164,
      "learning_rate": 1.8016764225328177e-06,
      "loss": 0.1598,
      "step": 275
    },
    {
      "epoch": 0.4584717607973422,
      "grad_norm": 10.3401460647583,
      "learning_rate": 1.8000643397460879e-06,
      "loss": 0.1736,
      "step": 276
    },
    {
      "epoch": 0.4601328903654485,
      "grad_norm": 8.310847282409668,
      "learning_rate": 1.7984464588996339e-06,
      "loss": 0.1426,
      "step": 277
    },
    {
      "epoch": 0.46179401993355484,
      "grad_norm": 8.352697372436523,
      "learning_rate": 1.7968227917182247e-06,
      "loss": 0.2232,
      "step": 278
    },
    {
      "epoch": 0.4634551495016611,
      "grad_norm": 7.920416355133057,
      "learning_rate": 1.7951933499685632e-06,
      "loss": 0.1649,
      "step": 279
    },
    {
      "epoch": 0.46511627906976744,
      "grad_norm": 10.16816234588623,
      "learning_rate": 1.7935581454592002e-06,
      "loss": 0.1324,
      "step": 280
    },
    {
      "epoch": 0.46677740863787376,
      "grad_norm": 6.508622646331787,
      "learning_rate": 1.7919171900404497e-06,
      "loss": 0.1132,
      "step": 281
    },
    {
      "epoch": 0.4684385382059801,
      "grad_norm": 6.778066635131836,
      "learning_rate": 1.790270495604302e-06,
      "loss": 0.1386,
      "step": 282
    },
    {
      "epoch": 0.4700996677740864,
      "grad_norm": 8.793827056884766,
      "learning_rate": 1.7886180740843383e-06,
      "loss": 0.1468,
      "step": 283
    },
    {
      "epoch": 0.4717607973421927,
      "grad_norm": 7.901587009429932,
      "learning_rate": 1.7869599374556438e-06,
      "loss": 0.1389,
      "step": 284
    },
    {
      "epoch": 0.473421926910299,
      "grad_norm": 8.337894439697266,
      "learning_rate": 1.785296097734721e-06,
      "loss": 0.1333,
      "step": 285
    },
    {
      "epoch": 0.4750830564784053,
      "grad_norm": 12.676334381103516,
      "learning_rate": 1.783626566979403e-06,
      "loss": 0.2014,
      "step": 286
    },
    {
      "epoch": 0.47674418604651164,
      "grad_norm": 7.533664703369141,
      "learning_rate": 1.7819513572887655e-06,
      "loss": 0.1238,
      "step": 287
    },
    {
      "epoch": 0.47840531561461797,
      "grad_norm": 8.285493850708008,
      "learning_rate": 1.7802704808030389e-06,
      "loss": 0.1078,
      "step": 288
    },
    {
      "epoch": 0.48006644518272423,
      "grad_norm": 8.483846664428711,
      "learning_rate": 1.778583949703522e-06,
      "loss": 0.0427,
      "step": 289
    },
    {
      "epoch": 0.48172757475083056,
      "grad_norm": 7.3853559494018555,
      "learning_rate": 1.7768917762124915e-06,
      "loss": 0.0716,
      "step": 290
    },
    {
      "epoch": 0.4833887043189369,
      "grad_norm": 18.682451248168945,
      "learning_rate": 1.7751939725931148e-06,
      "loss": 0.2179,
      "step": 291
    },
    {
      "epoch": 0.4850498338870432,
      "grad_norm": 14.185220718383789,
      "learning_rate": 1.773490551149361e-06,
      "loss": 0.185,
      "step": 292
    },
    {
      "epoch": 0.4867109634551495,
      "grad_norm": 15.155757904052734,
      "learning_rate": 1.7717815242259118e-06,
      "loss": 0.1625,
      "step": 293
    },
    {
      "epoch": 0.4883720930232558,
      "grad_norm": 15.565550804138184,
      "learning_rate": 1.7700669042080705e-06,
      "loss": 0.1547,
      "step": 294
    },
    {
      "epoch": 0.4900332225913621,
      "grad_norm": 12.508766174316406,
      "learning_rate": 1.7683467035216749e-06,
      "loss": 0.1153,
      "step": 295
    },
    {
      "epoch": 0.49169435215946844,
      "grad_norm": 14.286003112792969,
      "learning_rate": 1.7666209346330047e-06,
      "loss": 0.1617,
      "step": 296
    },
    {
      "epoch": 0.49335548172757476,
      "grad_norm": 12.081841468811035,
      "learning_rate": 1.7648896100486938e-06,
      "loss": 0.0842,
      "step": 297
    },
    {
      "epoch": 0.4950166112956811,
      "grad_norm": 10.268631935119629,
      "learning_rate": 1.7631527423156367e-06,
      "loss": 0.0907,
      "step": 298
    },
    {
      "epoch": 0.49667774086378735,
      "grad_norm": 5.768962383270264,
      "learning_rate": 1.7614103440209e-06,
      "loss": 0.077,
      "step": 299
    },
    {
      "epoch": 0.4983388704318937,
      "grad_norm": 11.941264152526855,
      "learning_rate": 1.7596624277916292e-06,
      "loss": 0.1292,
      "step": 300
    },
    {
      "epoch": 0.5,
      "grad_norm": 14.39963150024414,
      "learning_rate": 1.7579090062949597e-06,
      "loss": 0.1434,
      "step": 301
    },
    {
      "epoch": 0.5016611295681063,
      "grad_norm": 10.610041618347168,
      "learning_rate": 1.7561500922379225e-06,
      "loss": 0.1121,
      "step": 302
    },
    {
      "epoch": 0.5033222591362126,
      "grad_norm": 9.320536613464355,
      "learning_rate": 1.7543856983673528e-06,
      "loss": 0.0995,
      "step": 303
    },
    {
      "epoch": 0.5049833887043189,
      "grad_norm": 14.534180641174316,
      "learning_rate": 1.7526158374697997e-06,
      "loss": 0.1487,
      "step": 304
    },
    {
      "epoch": 0.5066445182724253,
      "grad_norm": 6.3102617263793945,
      "learning_rate": 1.7508405223714297e-06,
      "loss": 0.0614,
      "step": 305
    },
    {
      "epoch": 0.5083056478405316,
      "grad_norm": 13.029088973999023,
      "learning_rate": 1.7490597659379371e-06,
      "loss": 0.1273,
      "step": 306
    },
    {
      "epoch": 0.5099667774086378,
      "grad_norm": 11.671377182006836,
      "learning_rate": 1.7472735810744493e-06,
      "loss": 0.1749,
      "step": 307
    },
    {
      "epoch": 0.5116279069767442,
      "grad_norm": 11.286014556884766,
      "learning_rate": 1.745481980725433e-06,
      "loss": 0.1158,
      "step": 308
    },
    {
      "epoch": 0.5132890365448505,
      "grad_norm": 13.731078147888184,
      "learning_rate": 1.743684977874602e-06,
      "loss": 0.0995,
      "step": 309
    },
    {
      "epoch": 0.5149501661129569,
      "grad_norm": 13.560349464416504,
      "learning_rate": 1.7418825855448205e-06,
      "loss": 0.1754,
      "step": 310
    },
    {
      "epoch": 0.5166112956810631,
      "grad_norm": 15.380659103393555,
      "learning_rate": 1.7400748167980113e-06,
      "loss": 0.1893,
      "step": 311
    },
    {
      "epoch": 0.5182724252491694,
      "grad_norm": 17.06402015686035,
      "learning_rate": 1.7382616847350597e-06,
      "loss": 0.172,
      "step": 312
    },
    {
      "epoch": 0.5199335548172758,
      "grad_norm": 42.22692108154297,
      "learning_rate": 1.7364432024957193e-06,
      "loss": 0.1654,
      "step": 313
    },
    {
      "epoch": 0.521594684385382,
      "grad_norm": 17.076013565063477,
      "learning_rate": 1.7346193832585153e-06,
      "loss": 0.1461,
      "step": 314
    },
    {
      "epoch": 0.5232558139534884,
      "grad_norm": 13.314347267150879,
      "learning_rate": 1.7327902402406509e-06,
      "loss": 0.1855,
      "step": 315
    },
    {
      "epoch": 0.5249169435215947,
      "grad_norm": 14.270951271057129,
      "learning_rate": 1.7309557866979113e-06,
      "loss": 0.165,
      "step": 316
    },
    {
      "epoch": 0.526578073089701,
      "grad_norm": 8.411968231201172,
      "learning_rate": 1.7291160359245656e-06,
      "loss": 0.0846,
      "step": 317
    },
    {
      "epoch": 0.5282392026578073,
      "grad_norm": 8.805373191833496,
      "learning_rate": 1.727271001253273e-06,
      "loss": 0.1413,
      "step": 318
    },
    {
      "epoch": 0.5299003322259136,
      "grad_norm": 9.515275955200195,
      "learning_rate": 1.7254206960549842e-06,
      "loss": 0.1666,
      "step": 319
    },
    {
      "epoch": 0.53156146179402,
      "grad_norm": 9.840771675109863,
      "learning_rate": 1.7235651337388465e-06,
      "loss": 0.1051,
      "step": 320
    },
    {
      "epoch": 0.5332225913621262,
      "grad_norm": 10.479392051696777,
      "learning_rate": 1.721704327752104e-06,
      "loss": 0.1611,
      "step": 321
    },
    {
      "epoch": 0.5348837209302325,
      "grad_norm": 8.246603965759277,
      "learning_rate": 1.7198382915800032e-06,
      "loss": 0.1604,
      "step": 322
    },
    {
      "epoch": 0.5365448504983389,
      "grad_norm": 10.39277172088623,
      "learning_rate": 1.7179670387456926e-06,
      "loss": 0.1206,
      "step": 323
    },
    {
      "epoch": 0.5382059800664452,
      "grad_norm": 6.932436466217041,
      "learning_rate": 1.7160905828101261e-06,
      "loss": 0.1056,
      "step": 324
    },
    {
      "epoch": 0.5398671096345515,
      "grad_norm": 9.80469799041748,
      "learning_rate": 1.714208937371965e-06,
      "loss": 0.0961,
      "step": 325
    },
    {
      "epoch": 0.5415282392026578,
      "grad_norm": 10.003324508666992,
      "learning_rate": 1.7123221160674781e-06,
      "loss": 0.142,
      "step": 326
    },
    {
      "epoch": 0.5431893687707641,
      "grad_norm": 16.17133331298828,
      "learning_rate": 1.710430132570444e-06,
      "loss": 0.2312,
      "step": 327
    },
    {
      "epoch": 0.5448504983388704,
      "grad_norm": 19.575956344604492,
      "learning_rate": 1.7085330005920514e-06,
      "loss": 0.2264,
      "step": 328
    },
    {
      "epoch": 0.5465116279069767,
      "grad_norm": 17.07429313659668,
      "learning_rate": 1.7066307338808002e-06,
      "loss": 0.1616,
      "step": 329
    },
    {
      "epoch": 0.5481727574750831,
      "grad_norm": 6.755792140960693,
      "learning_rate": 1.7047233462224022e-06,
      "loss": 0.095,
      "step": 330
    },
    {
      "epoch": 0.5498338870431894,
      "grad_norm": 8.61792278289795,
      "learning_rate": 1.7028108514396797e-06,
      "loss": 0.1088,
      "step": 331
    },
    {
      "epoch": 0.5514950166112956,
      "grad_norm": 16.29265022277832,
      "learning_rate": 1.7008932633924671e-06,
      "loss": 0.2209,
      "step": 332
    },
    {
      "epoch": 0.553156146179402,
      "grad_norm": 6.72524356842041,
      "learning_rate": 1.6989705959775088e-06,
      "loss": 0.0891,
      "step": 333
    },
    {
      "epoch": 0.5548172757475083,
      "grad_norm": 7.47697114944458,
      "learning_rate": 1.69704286312836e-06,
      "loss": 0.0699,
      "step": 334
    },
    {
      "epoch": 0.5564784053156147,
      "grad_norm": 6.8748064041137695,
      "learning_rate": 1.695110078815285e-06,
      "loss": 0.1072,
      "step": 335
    },
    {
      "epoch": 0.5581395348837209,
      "grad_norm": 12.712291717529297,
      "learning_rate": 1.6931722570451562e-06,
      "loss": 0.1672,
      "step": 336
    },
    {
      "epoch": 0.5598006644518272,
      "grad_norm": 9.507533073425293,
      "learning_rate": 1.6912294118613515e-06,
      "loss": 0.0875,
      "step": 337
    },
    {
      "epoch": 0.5614617940199336,
      "grad_norm": 22.127843856811523,
      "learning_rate": 1.6892815573436546e-06,
      "loss": 0.1628,
      "step": 338
    },
    {
      "epoch": 0.5631229235880398,
      "grad_norm": 16.105968475341797,
      "learning_rate": 1.6873287076081506e-06,
      "loss": 0.168,
      "step": 339
    },
    {
      "epoch": 0.5647840531561462,
      "grad_norm": 17.4877986907959,
      "learning_rate": 1.6853708768071263e-06,
      "loss": 0.2376,
      "step": 340
    },
    {
      "epoch": 0.5664451827242525,
      "grad_norm": 18.06496238708496,
      "learning_rate": 1.683408079128965e-06,
      "loss": 0.1721,
      "step": 341
    },
    {
      "epoch": 0.5681063122923588,
      "grad_norm": 8.012746810913086,
      "learning_rate": 1.681440328798046e-06,
      "loss": 0.1208,
      "step": 342
    },
    {
      "epoch": 0.5697674418604651,
      "grad_norm": 9.295697212219238,
      "learning_rate": 1.6794676400746387e-06,
      "loss": 0.0754,
      "step": 343
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 15.045690536499023,
      "learning_rate": 1.6774900272548034e-06,
      "loss": 0.1164,
      "step": 344
    },
    {
      "epoch": 0.5730897009966778,
      "grad_norm": 6.664107322692871,
      "learning_rate": 1.6755075046702834e-06,
      "loss": 0.0714,
      "step": 345
    },
    {
      "epoch": 0.574750830564784,
      "grad_norm": 13.638888359069824,
      "learning_rate": 1.6735200866884035e-06,
      "loss": 0.1313,
      "step": 346
    },
    {
      "epoch": 0.5764119601328903,
      "grad_norm": 14.996500968933105,
      "learning_rate": 1.671527787711966e-06,
      "loss": 0.1683,
      "step": 347
    },
    {
      "epoch": 0.5780730897009967,
      "grad_norm": 6.490530014038086,
      "learning_rate": 1.669530622179145e-06,
      "loss": 0.0478,
      "step": 348
    },
    {
      "epoch": 0.579734219269103,
      "grad_norm": 12.507638931274414,
      "learning_rate": 1.6675286045633827e-06,
      "loss": 0.1066,
      "step": 349
    },
    {
      "epoch": 0.5813953488372093,
      "grad_norm": 13.5231294631958,
      "learning_rate": 1.6655217493732847e-06,
      "loss": 0.1507,
      "step": 350
    },
    {
      "epoch": 0.5830564784053156,
      "grad_norm": 10.306071281433105,
      "learning_rate": 1.663510071152514e-06,
      "loss": 0.0868,
      "step": 351
    },
    {
      "epoch": 0.584717607973422,
      "grad_norm": 13.600068092346191,
      "learning_rate": 1.6614935844796863e-06,
      "loss": 0.1112,
      "step": 352
    },
    {
      "epoch": 0.5863787375415282,
      "grad_norm": 7.447047710418701,
      "learning_rate": 1.6594723039682637e-06,
      "loss": 0.0685,
      "step": 353
    },
    {
      "epoch": 0.5880398671096345,
      "grad_norm": 17.658233642578125,
      "learning_rate": 1.65744624426645e-06,
      "loss": 0.1508,
      "step": 354
    },
    {
      "epoch": 0.5897009966777409,
      "grad_norm": 9.628240585327148,
      "learning_rate": 1.6554154200570825e-06,
      "loss": 0.1017,
      "step": 355
    },
    {
      "epoch": 0.5913621262458472,
      "grad_norm": 8.991092681884766,
      "learning_rate": 1.6533798460575284e-06,
      "loss": 0.1102,
      "step": 356
    },
    {
      "epoch": 0.5930232558139535,
      "grad_norm": 9.290343284606934,
      "learning_rate": 1.6513395370195755e-06,
      "loss": 0.0953,
      "step": 357
    },
    {
      "epoch": 0.5946843853820598,
      "grad_norm": 12.639936447143555,
      "learning_rate": 1.6492945077293268e-06,
      "loss": 0.1731,
      "step": 358
    },
    {
      "epoch": 0.5963455149501661,
      "grad_norm": 15.398722648620605,
      "learning_rate": 1.647244773007093e-06,
      "loss": 0.1704,
      "step": 359
    },
    {
      "epoch": 0.5980066445182725,
      "grad_norm": 4.541179656982422,
      "learning_rate": 1.6451903477072845e-06,
      "loss": 0.0519,
      "step": 360
    },
    {
      "epoch": 0.5996677740863787,
      "grad_norm": 7.931593418121338,
      "learning_rate": 1.6431312467183047e-06,
      "loss": 0.0743,
      "step": 361
    },
    {
      "epoch": 0.6013289036544851,
      "grad_norm": 10.063234329223633,
      "learning_rate": 1.6410674849624417e-06,
      "loss": 0.1202,
      "step": 362
    },
    {
      "epoch": 0.6029900332225914,
      "grad_norm": 12.936481475830078,
      "learning_rate": 1.6389990773957605e-06,
      "loss": 0.1219,
      "step": 363
    },
    {
      "epoch": 0.6046511627906976,
      "grad_norm": 20.578336715698242,
      "learning_rate": 1.636926039007993e-06,
      "loss": 0.1967,
      "step": 364
    },
    {
      "epoch": 0.606312292358804,
      "grad_norm": 7.607759475708008,
      "learning_rate": 1.634848384822432e-06,
      "loss": 0.0589,
      "step": 365
    },
    {
      "epoch": 0.6079734219269103,
      "grad_norm": 12.115143775939941,
      "learning_rate": 1.6327661298958203e-06,
      "loss": 0.1078,
      "step": 366
    },
    {
      "epoch": 0.6096345514950167,
      "grad_norm": 10.23250675201416,
      "learning_rate": 1.630679289318242e-06,
      "loss": 0.0893,
      "step": 367
    },
    {
      "epoch": 0.6112956810631229,
      "grad_norm": 15.31701946258545,
      "learning_rate": 1.628587878213014e-06,
      "loss": 0.2084,
      "step": 368
    },
    {
      "epoch": 0.6129568106312292,
      "grad_norm": 13.3017578125,
      "learning_rate": 1.6264919117365756e-06,
      "loss": 0.1501,
      "step": 369
    },
    {
      "epoch": 0.6146179401993356,
      "grad_norm": 11.242172241210938,
      "learning_rate": 1.6243914050783782e-06,
      "loss": 0.1418,
      "step": 370
    },
    {
      "epoch": 0.6162790697674418,
      "grad_norm": 15.284845352172852,
      "learning_rate": 1.6222863734607767e-06,
      "loss": 0.2246,
      "step": 371
    },
    {
      "epoch": 0.6179401993355482,
      "grad_norm": 5.70803165435791,
      "learning_rate": 1.620176832138918e-06,
      "loss": 0.0742,
      "step": 372
    },
    {
      "epoch": 0.6196013289036545,
      "grad_norm": 11.78730297088623,
      "learning_rate": 1.618062796400631e-06,
      "loss": 0.1321,
      "step": 373
    },
    {
      "epoch": 0.6212624584717608,
      "grad_norm": 6.646214962005615,
      "learning_rate": 1.6159442815663149e-06,
      "loss": 0.1329,
      "step": 374
    },
    {
      "epoch": 0.6229235880398671,
      "grad_norm": 9.914974212646484,
      "learning_rate": 1.61382130298883e-06,
      "loss": 0.0823,
      "step": 375
    },
    {
      "epoch": 0.6245847176079734,
      "grad_norm": 7.075379848480225,
      "learning_rate": 1.6116938760533842e-06,
      "loss": 0.079,
      "step": 376
    },
    {
      "epoch": 0.6262458471760798,
      "grad_norm": 11.401320457458496,
      "learning_rate": 1.6095620161774232e-06,
      "loss": 0.1558,
      "step": 377
    },
    {
      "epoch": 0.627906976744186,
      "grad_norm": 8.446409225463867,
      "learning_rate": 1.6074257388105176e-06,
      "loss": 0.1214,
      "step": 378
    },
    {
      "epoch": 0.6295681063122923,
      "grad_norm": 21.146482467651367,
      "learning_rate": 1.6052850594342533e-06,
      "loss": 0.1599,
      "step": 379
    },
    {
      "epoch": 0.6312292358803987,
      "grad_norm": 8.168869018554688,
      "learning_rate": 1.6031399935621153e-06,
      "loss": 0.0863,
      "step": 380
    },
    {
      "epoch": 0.632890365448505,
      "grad_norm": 11.66989517211914,
      "learning_rate": 1.6009905567393783e-06,
      "loss": 0.1079,
      "step": 381
    },
    {
      "epoch": 0.6345514950166113,
      "grad_norm": 6.585104465484619,
      "learning_rate": 1.5988367645429936e-06,
      "loss": 0.0671,
      "step": 382
    },
    {
      "epoch": 0.6362126245847176,
      "grad_norm": 7.7533392906188965,
      "learning_rate": 1.5966786325814757e-06,
      "loss": 0.0791,
      "step": 383
    },
    {
      "epoch": 0.6378737541528239,
      "grad_norm": 19.80597496032715,
      "learning_rate": 1.5945161764947892e-06,
      "loss": 0.144,
      "step": 384
    },
    {
      "epoch": 0.6395348837209303,
      "grad_norm": 12.21318531036377,
      "learning_rate": 1.5923494119542357e-06,
      "loss": 0.1094,
      "step": 385
    },
    {
      "epoch": 0.6411960132890365,
      "grad_norm": 13.927626609802246,
      "learning_rate": 1.5901783546623402e-06,
      "loss": 0.1036,
      "step": 386
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 17.715267181396484,
      "learning_rate": 1.5880030203527367e-06,
      "loss": 0.1133,
      "step": 387
    },
    {
      "epoch": 0.6445182724252492,
      "grad_norm": 19.633594512939453,
      "learning_rate": 1.585823424790056e-06,
      "loss": 0.1864,
      "step": 388
    },
    {
      "epoch": 0.6461794019933554,
      "grad_norm": 13.235529899597168,
      "learning_rate": 1.5836395837698084e-06,
      "loss": 0.1207,
      "step": 389
    },
    {
      "epoch": 0.6478405315614618,
      "grad_norm": 10.060149192810059,
      "learning_rate": 1.5814515131182725e-06,
      "loss": 0.0723,
      "step": 390
    },
    {
      "epoch": 0.6495016611295681,
      "grad_norm": 16.433595657348633,
      "learning_rate": 1.5792592286923778e-06,
      "loss": 0.176,
      "step": 391
    },
    {
      "epoch": 0.6511627906976745,
      "grad_norm": 14.24596881866455,
      "learning_rate": 1.5770627463795924e-06,
      "loss": 0.1633,
      "step": 392
    },
    {
      "epoch": 0.6528239202657807,
      "grad_norm": 11.974753379821777,
      "learning_rate": 1.5748620820978055e-06,
      "loss": 0.1403,
      "step": 393
    },
    {
      "epoch": 0.654485049833887,
      "grad_norm": 25.53243637084961,
      "learning_rate": 1.5726572517952121e-06,
      "loss": 0.1529,
      "step": 394
    },
    {
      "epoch": 0.6561461794019934,
      "grad_norm": 11.890317916870117,
      "learning_rate": 1.5704482714502005e-06,
      "loss": 0.1004,
      "step": 395
    },
    {
      "epoch": 0.6578073089700996,
      "grad_norm": 13.248994827270508,
      "learning_rate": 1.568235157071233e-06,
      "loss": 0.1605,
      "step": 396
    },
    {
      "epoch": 0.659468438538206,
      "grad_norm": 10.647342681884766,
      "learning_rate": 1.5660179246967311e-06,
      "loss": 0.1145,
      "step": 397
    },
    {
      "epoch": 0.6611295681063123,
      "grad_norm": 9.527441024780273,
      "learning_rate": 1.5637965903949597e-06,
      "loss": 0.1327,
      "step": 398
    },
    {
      "epoch": 0.6627906976744186,
      "grad_norm": 10.44089412689209,
      "learning_rate": 1.5615711702639107e-06,
      "loss": 0.1212,
      "step": 399
    },
    {
      "epoch": 0.6644518272425249,
      "grad_norm": 7.151495456695557,
      "learning_rate": 1.559341680431185e-06,
      "loss": 0.1242,
      "step": 400
    },
    {
      "epoch": 0.6661129568106312,
      "grad_norm": 13.282041549682617,
      "learning_rate": 1.5571081370538775e-06,
      "loss": 0.2158,
      "step": 401
    },
    {
      "epoch": 0.6677740863787376,
      "grad_norm": 6.182162761688232,
      "learning_rate": 1.554870556318459e-06,
      "loss": 0.0821,
      "step": 402
    },
    {
      "epoch": 0.6694352159468439,
      "grad_norm": 6.321899890899658,
      "learning_rate": 1.5526289544406583e-06,
      "loss": 0.1002,
      "step": 403
    },
    {
      "epoch": 0.6710963455149501,
      "grad_norm": 13.514141082763672,
      "learning_rate": 1.5503833476653463e-06,
      "loss": 0.1969,
      "step": 404
    },
    {
      "epoch": 0.6727574750830565,
      "grad_norm": 9.189848899841309,
      "learning_rate": 1.5481337522664167e-06,
      "loss": 0.1388,
      "step": 405
    },
    {
      "epoch": 0.6744186046511628,
      "grad_norm": 7.638835430145264,
      "learning_rate": 1.545880184546669e-06,
      "loss": 0.1531,
      "step": 406
    },
    {
      "epoch": 0.6760797342192691,
      "grad_norm": 11.591959953308105,
      "learning_rate": 1.54362266083769e-06,
      "loss": 0.1512,
      "step": 407
    },
    {
      "epoch": 0.6777408637873754,
      "grad_norm": 11.338805198669434,
      "learning_rate": 1.5413611974997354e-06,
      "loss": 0.1792,
      "step": 408
    },
    {
      "epoch": 0.6794019933554817,
      "grad_norm": 9.469964981079102,
      "learning_rate": 1.539095810921612e-06,
      "loss": 0.1064,
      "step": 409
    },
    {
      "epoch": 0.6810631229235881,
      "grad_norm": 9.461012840270996,
      "learning_rate": 1.5368265175205572e-06,
      "loss": 0.145,
      "step": 410
    },
    {
      "epoch": 0.6827242524916943,
      "grad_norm": 6.2333903312683105,
      "learning_rate": 1.534553333742122e-06,
      "loss": 0.1042,
      "step": 411
    },
    {
      "epoch": 0.6843853820598007,
      "grad_norm": 9.561917304992676,
      "learning_rate": 1.5322762760600507e-06,
      "loss": 0.157,
      "step": 412
    },
    {
      "epoch": 0.686046511627907,
      "grad_norm": 17.19264030456543,
      "learning_rate": 1.5299953609761617e-06,
      "loss": 0.1578,
      "step": 413
    },
    {
      "epoch": 0.6877076411960132,
      "grad_norm": 13.552623748779297,
      "learning_rate": 1.527710605020228e-06,
      "loss": 0.1879,
      "step": 414
    },
    {
      "epoch": 0.6893687707641196,
      "grad_norm": 10.698836326599121,
      "learning_rate": 1.525422024749857e-06,
      "loss": 0.1388,
      "step": 415
    },
    {
      "epoch": 0.6910299003322259,
      "grad_norm": 8.805788040161133,
      "learning_rate": 1.5231296367503715e-06,
      "loss": 0.0887,
      "step": 416
    },
    {
      "epoch": 0.6926910299003323,
      "grad_norm": 9.94221019744873,
      "learning_rate": 1.520833457634688e-06,
      "loss": 0.1246,
      "step": 417
    },
    {
      "epoch": 0.6943521594684385,
      "grad_norm": 12.409692764282227,
      "learning_rate": 1.5185335040431988e-06,
      "loss": 0.1039,
      "step": 418
    },
    {
      "epoch": 0.6960132890365448,
      "grad_norm": 8.411705017089844,
      "learning_rate": 1.5162297926436481e-06,
      "loss": 0.0864,
      "step": 419
    },
    {
      "epoch": 0.6976744186046512,
      "grad_norm": 12.356369972229004,
      "learning_rate": 1.5139223401310135e-06,
      "loss": 0.1329,
      "step": 420
    },
    {
      "epoch": 0.6993355481727574,
      "grad_norm": 16.476016998291016,
      "learning_rate": 1.5116111632273847e-06,
      "loss": 0.1199,
      "step": 421
    },
    {
      "epoch": 0.7009966777408638,
      "grad_norm": 12.528667449951172,
      "learning_rate": 1.5092962786818418e-06,
      "loss": 0.1322,
      "step": 422
    },
    {
      "epoch": 0.7026578073089701,
      "grad_norm": 16.012834548950195,
      "learning_rate": 1.506977703270334e-06,
      "loss": 0.1461,
      "step": 423
    },
    {
      "epoch": 0.7043189368770764,
      "grad_norm": 8.67282772064209,
      "learning_rate": 1.5046554537955584e-06,
      "loss": 0.0871,
      "step": 424
    },
    {
      "epoch": 0.7059800664451827,
      "grad_norm": 9.403173446655273,
      "learning_rate": 1.502329547086838e-06,
      "loss": 0.1064,
      "step": 425
    },
    {
      "epoch": 0.707641196013289,
      "grad_norm": 10.017657279968262,
      "learning_rate": 1.5e-06,
      "loss": 0.0891,
      "step": 426
    },
    {
      "epoch": 0.7093023255813954,
      "grad_norm": 12.303718566894531,
      "learning_rate": 1.4976668294172526e-06,
      "loss": 0.1399,
      "step": 427
    },
    {
      "epoch": 0.7109634551495017,
      "grad_norm": 10.830999374389648,
      "learning_rate": 1.4953300522470637e-06,
      "loss": 0.1286,
      "step": 428
    },
    {
      "epoch": 0.7126245847176079,
      "grad_norm": 14.537104606628418,
      "learning_rate": 1.4929896854240394e-06,
      "loss": 0.1815,
      "step": 429
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 6.298677444458008,
      "learning_rate": 1.4906457459087975e-06,
      "loss": 0.0982,
      "step": 430
    },
    {
      "epoch": 0.7159468438538206,
      "grad_norm": 7.235755920410156,
      "learning_rate": 1.4882982506878498e-06,
      "loss": 0.0952,
      "step": 431
    },
    {
      "epoch": 0.717607973421927,
      "grad_norm": 12.166476249694824,
      "learning_rate": 1.4859472167734743e-06,
      "loss": 0.1752,
      "step": 432
    },
    {
      "epoch": 0.7192691029900332,
      "grad_norm": 7.841498374938965,
      "learning_rate": 1.4835926612035944e-06,
      "loss": 0.1136,
      "step": 433
    },
    {
      "epoch": 0.7209302325581395,
      "grad_norm": 9.74621868133545,
      "learning_rate": 1.4812346010416553e-06,
      "loss": 0.1213,
      "step": 434
    },
    {
      "epoch": 0.7225913621262459,
      "grad_norm": 6.355194568634033,
      "learning_rate": 1.4788730533765001e-06,
      "loss": 0.0953,
      "step": 435
    },
    {
      "epoch": 0.7242524916943521,
      "grad_norm": 13.128461837768555,
      "learning_rate": 1.4765080353222446e-06,
      "loss": 0.1959,
      "step": 436
    },
    {
      "epoch": 0.7259136212624585,
      "grad_norm": 9.213345527648926,
      "learning_rate": 1.4741395640181557e-06,
      "loss": 0.1339,
      "step": 437
    },
    {
      "epoch": 0.7275747508305648,
      "grad_norm": 8.842706680297852,
      "learning_rate": 1.4717676566285256e-06,
      "loss": 0.1312,
      "step": 438
    },
    {
      "epoch": 0.729235880398671,
      "grad_norm": 10.16006088256836,
      "learning_rate": 1.4693923303425477e-06,
      "loss": 0.1049,
      "step": 439
    },
    {
      "epoch": 0.7308970099667774,
      "grad_norm": 10.167031288146973,
      "learning_rate": 1.4670136023741924e-06,
      "loss": 0.139,
      "step": 440
    },
    {
      "epoch": 0.7325581395348837,
      "grad_norm": 7.213005542755127,
      "learning_rate": 1.464631489962081e-06,
      "loss": 0.087,
      "step": 441
    },
    {
      "epoch": 0.7342192691029901,
      "grad_norm": 5.736367702484131,
      "learning_rate": 1.4622460103693637e-06,
      "loss": 0.0763,
      "step": 442
    },
    {
      "epoch": 0.7358803986710963,
      "grad_norm": 7.528397560119629,
      "learning_rate": 1.459857180883591e-06,
      "loss": 0.0728,
      "step": 443
    },
    {
      "epoch": 0.7375415282392026,
      "grad_norm": 8.150105476379395,
      "learning_rate": 1.4574650188165908e-06,
      "loss": 0.1374,
      "step": 444
    },
    {
      "epoch": 0.739202657807309,
      "grad_norm": 11.9616060256958,
      "learning_rate": 1.4550695415043418e-06,
      "loss": 0.142,
      "step": 445
    },
    {
      "epoch": 0.7408637873754153,
      "grad_norm": 9.410012245178223,
      "learning_rate": 1.4526707663068488e-06,
      "loss": 0.1567,
      "step": 446
    },
    {
      "epoch": 0.7425249169435216,
      "grad_norm": 11.85716438293457,
      "learning_rate": 1.450268710608016e-06,
      "loss": 0.1245,
      "step": 447
    },
    {
      "epoch": 0.7441860465116279,
      "grad_norm": 9.234456062316895,
      "learning_rate": 1.4478633918155214e-06,
      "loss": 0.0959,
      "step": 448
    },
    {
      "epoch": 0.7458471760797342,
      "grad_norm": 22.62066078186035,
      "learning_rate": 1.4454548273606905e-06,
      "loss": 0.2426,
      "step": 449
    },
    {
      "epoch": 0.7475083056478405,
      "grad_norm": 11.37014389038086,
      "learning_rate": 1.4430430346983698e-06,
      "loss": 0.1089,
      "step": 450
    },
    {
      "epoch": 0.7491694352159468,
      "grad_norm": 15.76647663116455,
      "learning_rate": 1.4406280313068019e-06,
      "loss": 0.1549,
      "step": 451
    },
    {
      "epoch": 0.7508305647840532,
      "grad_norm": 8.85522174835205,
      "learning_rate": 1.4382098346874957e-06,
      "loss": 0.1208,
      "step": 452
    },
    {
      "epoch": 0.7524916943521595,
      "grad_norm": 9.161630630493164,
      "learning_rate": 1.4357884623651031e-06,
      "loss": 0.132,
      "step": 453
    },
    {
      "epoch": 0.7541528239202658,
      "grad_norm": 14.351627349853516,
      "learning_rate": 1.433363931887289e-06,
      "loss": 0.1408,
      "step": 454
    },
    {
      "epoch": 0.7558139534883721,
      "grad_norm": 7.650735378265381,
      "learning_rate": 1.430936260824606e-06,
      "loss": 0.1038,
      "step": 455
    },
    {
      "epoch": 0.7574750830564784,
      "grad_norm": 37.2721061706543,
      "learning_rate": 1.4285054667703666e-06,
      "loss": 0.1731,
      "step": 456
    },
    {
      "epoch": 0.7591362126245847,
      "grad_norm": 9.084954261779785,
      "learning_rate": 1.4260715673405154e-06,
      "loss": 0.0963,
      "step": 457
    },
    {
      "epoch": 0.760797342192691,
      "grad_norm": 9.70747184753418,
      "learning_rate": 1.4236345801735017e-06,
      "loss": 0.0959,
      "step": 458
    },
    {
      "epoch": 0.7624584717607974,
      "grad_norm": 13.642250061035156,
      "learning_rate": 1.421194522930151e-06,
      "loss": 0.1338,
      "step": 459
    },
    {
      "epoch": 0.7641196013289037,
      "grad_norm": 7.298130035400391,
      "learning_rate": 1.4187514132935392e-06,
      "loss": 0.0824,
      "step": 460
    },
    {
      "epoch": 0.7657807308970099,
      "grad_norm": 8.563011169433594,
      "learning_rate": 1.416305268968861e-06,
      "loss": 0.1036,
      "step": 461
    },
    {
      "epoch": 0.7674418604651163,
      "grad_norm": 5.615513801574707,
      "learning_rate": 1.4138561076833047e-06,
      "loss": 0.0747,
      "step": 462
    },
    {
      "epoch": 0.7691029900332226,
      "grad_norm": 11.83483600616455,
      "learning_rate": 1.411403947185922e-06,
      "loss": 0.1074,
      "step": 463
    },
    {
      "epoch": 0.770764119601329,
      "grad_norm": 10.225103378295898,
      "learning_rate": 1.4089488052474997e-06,
      "loss": 0.1229,
      "step": 464
    },
    {
      "epoch": 0.7724252491694352,
      "grad_norm": 11.923834800720215,
      "learning_rate": 1.406490699660432e-06,
      "loss": 0.1,
      "step": 465
    },
    {
      "epoch": 0.7740863787375415,
      "grad_norm": 8.985201835632324,
      "learning_rate": 1.404029648238589e-06,
      "loss": 0.0827,
      "step": 466
    },
    {
      "epoch": 0.7757475083056479,
      "grad_norm": 17.72941017150879,
      "learning_rate": 1.4015656688171916e-06,
      "loss": 0.1344,
      "step": 467
    },
    {
      "epoch": 0.7774086378737541,
      "grad_norm": 11.252090454101562,
      "learning_rate": 1.399098779252677e-06,
      "loss": 0.1579,
      "step": 468
    },
    {
      "epoch": 0.7790697674418605,
      "grad_norm": 12.625454902648926,
      "learning_rate": 1.396628997422575e-06,
      "loss": 0.1508,
      "step": 469
    },
    {
      "epoch": 0.7807308970099668,
      "grad_norm": 17.121810913085938,
      "learning_rate": 1.3941563412253729e-06,
      "loss": 0.1077,
      "step": 470
    },
    {
      "epoch": 0.782392026578073,
      "grad_norm": 9.032465934753418,
      "learning_rate": 1.3916808285803913e-06,
      "loss": 0.1013,
      "step": 471
    },
    {
      "epoch": 0.7840531561461794,
      "grad_norm": 9.71914005279541,
      "learning_rate": 1.3892024774276493e-06,
      "loss": 0.1332,
      "step": 472
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 21.92658233642578,
      "learning_rate": 1.3867213057277378e-06,
      "loss": 0.281,
      "step": 473
    },
    {
      "epoch": 0.7873754152823921,
      "grad_norm": 14.790511131286621,
      "learning_rate": 1.3842373314616878e-06,
      "loss": 0.1159,
      "step": 474
    },
    {
      "epoch": 0.7890365448504983,
      "grad_norm": 7.3284077644348145,
      "learning_rate": 1.38175057263084e-06,
      "loss": 0.1129,
      "step": 475
    },
    {
      "epoch": 0.7906976744186046,
      "grad_norm": 8.450250625610352,
      "learning_rate": 1.379261047256716e-06,
      "loss": 0.0967,
      "step": 476
    },
    {
      "epoch": 0.792358803986711,
      "grad_norm": 11.382468223571777,
      "learning_rate": 1.376768773380885e-06,
      "loss": 0.131,
      "step": 477
    },
    {
      "epoch": 0.7940199335548173,
      "grad_norm": 9.446375846862793,
      "learning_rate": 1.374273769064836e-06,
      "loss": 0.0924,
      "step": 478
    },
    {
      "epoch": 0.7956810631229236,
      "grad_norm": 15.360520362854004,
      "learning_rate": 1.3717760523898449e-06,
      "loss": 0.1401,
      "step": 479
    },
    {
      "epoch": 0.7973421926910299,
      "grad_norm": 8.478147506713867,
      "learning_rate": 1.369275641456844e-06,
      "loss": 0.1119,
      "step": 480
    },
    {
      "epoch": 0.7990033222591362,
      "grad_norm": 7.132209777832031,
      "learning_rate": 1.3667725543862904e-06,
      "loss": 0.0967,
      "step": 481
    },
    {
      "epoch": 0.8006644518272426,
      "grad_norm": 7.364678382873535,
      "learning_rate": 1.364266809318036e-06,
      "loss": 0.1483,
      "step": 482
    },
    {
      "epoch": 0.8023255813953488,
      "grad_norm": 6.641203880310059,
      "learning_rate": 1.3617584244111947e-06,
      "loss": 0.1094,
      "step": 483
    },
    {
      "epoch": 0.8039867109634552,
      "grad_norm": 9.827098846435547,
      "learning_rate": 1.3592474178440114e-06,
      "loss": 0.1196,
      "step": 484
    },
    {
      "epoch": 0.8056478405315615,
      "grad_norm": 7.959251880645752,
      "learning_rate": 1.3567338078137304e-06,
      "loss": 0.1112,
      "step": 485
    },
    {
      "epoch": 0.8073089700996677,
      "grad_norm": 10.041277885437012,
      "learning_rate": 1.3542176125364624e-06,
      "loss": 0.125,
      "step": 486
    },
    {
      "epoch": 0.8089700996677741,
      "grad_norm": 7.443721294403076,
      "learning_rate": 1.3516988502470547e-06,
      "loss": 0.0793,
      "step": 487
    },
    {
      "epoch": 0.8106312292358804,
      "grad_norm": 4.981016635894775,
      "learning_rate": 1.3491775391989567e-06,
      "loss": 0.0578,
      "step": 488
    },
    {
      "epoch": 0.8122923588039868,
      "grad_norm": 8.452247619628906,
      "learning_rate": 1.3466536976640885e-06,
      "loss": 0.0756,
      "step": 489
    },
    {
      "epoch": 0.813953488372093,
      "grad_norm": 11.617294311523438,
      "learning_rate": 1.3441273439327099e-06,
      "loss": 0.1108,
      "step": 490
    },
    {
      "epoch": 0.8156146179401993,
      "grad_norm": 13.18224048614502,
      "learning_rate": 1.3415984963132851e-06,
      "loss": 0.1833,
      "step": 491
    },
    {
      "epoch": 0.8172757475083057,
      "grad_norm": 8.125739097595215,
      "learning_rate": 1.3390671731323526e-06,
      "loss": 0.0788,
      "step": 492
    },
    {
      "epoch": 0.8189368770764119,
      "grad_norm": 9.55408763885498,
      "learning_rate": 1.3365333927343905e-06,
      "loss": 0.0969,
      "step": 493
    },
    {
      "epoch": 0.8205980066445183,
      "grad_norm": 14.843066215515137,
      "learning_rate": 1.3339971734816842e-06,
      "loss": 0.1476,
      "step": 494
    },
    {
      "epoch": 0.8222591362126246,
      "grad_norm": 12.41232681274414,
      "learning_rate": 1.3314585337541947e-06,
      "loss": 0.1056,
      "step": 495
    },
    {
      "epoch": 0.8239202657807309,
      "grad_norm": 18.931259155273438,
      "learning_rate": 1.3289174919494226e-06,
      "loss": 0.1967,
      "step": 496
    },
    {
      "epoch": 0.8255813953488372,
      "grad_norm": 19.2910213470459,
      "learning_rate": 1.3263740664822773e-06,
      "loss": 0.1126,
      "step": 497
    },
    {
      "epoch": 0.8272425249169435,
      "grad_norm": 10.538687705993652,
      "learning_rate": 1.3238282757849428e-06,
      "loss": 0.1148,
      "step": 498
    },
    {
      "epoch": 0.8289036544850499,
      "grad_norm": 9.582296371459961,
      "learning_rate": 1.321280138306743e-06,
      "loss": 0.0927,
      "step": 499
    },
    {
      "epoch": 0.8305647840531561,
      "grad_norm": 7.91030740737915,
      "learning_rate": 1.318729672514009e-06,
      "loss": 0.0851,
      "step": 500
    },
    {
      "epoch": 0.8322259136212624,
      "grad_norm": 13.952978134155273,
      "learning_rate": 1.3161768968899466e-06,
      "loss": 0.0748,
      "step": 501
    },
    {
      "epoch": 0.8338870431893688,
      "grad_norm": 6.64433479309082,
      "learning_rate": 1.3136218299344992e-06,
      "loss": 0.0811,
      "step": 502
    },
    {
      "epoch": 0.8355481727574751,
      "grad_norm": 11.787875175476074,
      "learning_rate": 1.3110644901642162e-06,
      "loss": 0.1176,
      "step": 503
    },
    {
      "epoch": 0.8372093023255814,
      "grad_norm": 8.039713859558105,
      "learning_rate": 1.3085048961121175e-06,
      "loss": 0.0978,
      "step": 504
    },
    {
      "epoch": 0.8388704318936877,
      "grad_norm": 21.696443557739258,
      "learning_rate": 1.3059430663275608e-06,
      "loss": 0.1458,
      "step": 505
    },
    {
      "epoch": 0.840531561461794,
      "grad_norm": 7.948482990264893,
      "learning_rate": 1.3033790193761058e-06,
      "loss": 0.0772,
      "step": 506
    },
    {
      "epoch": 0.8421926910299004,
      "grad_norm": 13.564935684204102,
      "learning_rate": 1.3008127738393793e-06,
      "loss": 0.1865,
      "step": 507
    },
    {
      "epoch": 0.8438538205980066,
      "grad_norm": 13.352168083190918,
      "learning_rate": 1.2982443483149422e-06,
      "loss": 0.1602,
      "step": 508
    },
    {
      "epoch": 0.845514950166113,
      "grad_norm": 13.05396842956543,
      "learning_rate": 1.2956737614161525e-06,
      "loss": 0.1097,
      "step": 509
    },
    {
      "epoch": 0.8471760797342193,
      "grad_norm": 11.9678373336792,
      "learning_rate": 1.293101031772033e-06,
      "loss": 0.1681,
      "step": 510
    },
    {
      "epoch": 0.8488372093023255,
      "grad_norm": 9.613672256469727,
      "learning_rate": 1.2905261780271343e-06,
      "loss": 0.1004,
      "step": 511
    },
    {
      "epoch": 0.8504983388704319,
      "grad_norm": 9.589925765991211,
      "learning_rate": 1.2879492188414008e-06,
      "loss": 0.0901,
      "step": 512
    },
    {
      "epoch": 0.8521594684385382,
      "grad_norm": 11.159270286560059,
      "learning_rate": 1.2853701728900337e-06,
      "loss": 0.1538,
      "step": 513
    },
    {
      "epoch": 0.8538205980066446,
      "grad_norm": 11.989233016967773,
      "learning_rate": 1.2827890588633588e-06,
      "loss": 0.1468,
      "step": 514
    },
    {
      "epoch": 0.8554817275747508,
      "grad_norm": 8.866310119628906,
      "learning_rate": 1.2802058954666873e-06,
      "loss": 0.1249,
      "step": 515
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 5.853137493133545,
      "learning_rate": 1.2776207014201843e-06,
      "loss": 0.0761,
      "step": 516
    },
    {
      "epoch": 0.8588039867109635,
      "grad_norm": 10.618156433105469,
      "learning_rate": 1.2750334954587297e-06,
      "loss": 0.1087,
      "step": 517
    },
    {
      "epoch": 0.8604651162790697,
      "grad_norm": 10.3444242477417,
      "learning_rate": 1.2724442963317835e-06,
      "loss": 0.1751,
      "step": 518
    },
    {
      "epoch": 0.8621262458471761,
      "grad_norm": 12.944339752197266,
      "learning_rate": 1.2698531228032508e-06,
      "loss": 0.2003,
      "step": 519
    },
    {
      "epoch": 0.8637873754152824,
      "grad_norm": 9.638875007629395,
      "learning_rate": 1.2672599936513448e-06,
      "loss": 0.1065,
      "step": 520
    },
    {
      "epoch": 0.8654485049833887,
      "grad_norm": 7.887294292449951,
      "learning_rate": 1.2646649276684514e-06,
      "loss": 0.1576,
      "step": 521
    },
    {
      "epoch": 0.867109634551495,
      "grad_norm": 9.495591163635254,
      "learning_rate": 1.262067943660993e-06,
      "loss": 0.1919,
      "step": 522
    },
    {
      "epoch": 0.8687707641196013,
      "grad_norm": 7.008487224578857,
      "learning_rate": 1.2594690604492904e-06,
      "loss": 0.1603,
      "step": 523
    },
    {
      "epoch": 0.8704318936877077,
      "grad_norm": 5.752983093261719,
      "learning_rate": 1.25686829686743e-06,
      "loss": 0.116,
      "step": 524
    },
    {
      "epoch": 0.872093023255814,
      "grad_norm": 6.817920207977295,
      "learning_rate": 1.2542656717631236e-06,
      "loss": 0.1433,
      "step": 525
    },
    {
      "epoch": 0.8737541528239202,
      "grad_norm": 9.408466339111328,
      "learning_rate": 1.2516612039975742e-06,
      "loss": 0.1372,
      "step": 526
    },
    {
      "epoch": 0.8754152823920266,
      "grad_norm": 12.023665428161621,
      "learning_rate": 1.2490549124453385e-06,
      "loss": 0.1777,
      "step": 527
    },
    {
      "epoch": 0.8770764119601329,
      "grad_norm": 7.048364639282227,
      "learning_rate": 1.2464468159941898e-06,
      "loss": 0.1261,
      "step": 528
    },
    {
      "epoch": 0.8787375415282392,
      "grad_norm": 6.778327941894531,
      "learning_rate": 1.2438369335449822e-06,
      "loss": 0.1319,
      "step": 529
    },
    {
      "epoch": 0.8803986710963455,
      "grad_norm": 7.284000873565674,
      "learning_rate": 1.241225284011512e-06,
      "loss": 0.1324,
      "step": 530
    },
    {
      "epoch": 0.8820598006644518,
      "grad_norm": 11.69240951538086,
      "learning_rate": 1.2386118863203818e-06,
      "loss": 0.1602,
      "step": 531
    },
    {
      "epoch": 0.8837209302325582,
      "grad_norm": 9.516483306884766,
      "learning_rate": 1.2359967594108642e-06,
      "loss": 0.1768,
      "step": 532
    },
    {
      "epoch": 0.8853820598006644,
      "grad_norm": 5.771677494049072,
      "learning_rate": 1.2333799222347617e-06,
      "loss": 0.1187,
      "step": 533
    },
    {
      "epoch": 0.8870431893687708,
      "grad_norm": 6.525841236114502,
      "learning_rate": 1.230761393756272e-06,
      "loss": 0.1169,
      "step": 534
    },
    {
      "epoch": 0.8887043189368771,
      "grad_norm": 8.253816604614258,
      "learning_rate": 1.22814119295185e-06,
      "loss": 0.1622,
      "step": 535
    },
    {
      "epoch": 0.8903654485049833,
      "grad_norm": 9.743227005004883,
      "learning_rate": 1.225519338810069e-06,
      "loss": 0.1169,
      "step": 536
    },
    {
      "epoch": 0.8920265780730897,
      "grad_norm": 8.275440216064453,
      "learning_rate": 1.222895850331485e-06,
      "loss": 0.1149,
      "step": 537
    },
    {
      "epoch": 0.893687707641196,
      "grad_norm": 7.478592872619629,
      "learning_rate": 1.220270746528497e-06,
      "loss": 0.1055,
      "step": 538
    },
    {
      "epoch": 0.8953488372093024,
      "grad_norm": 9.597691535949707,
      "learning_rate": 1.217644046425211e-06,
      "loss": 0.1186,
      "step": 539
    },
    {
      "epoch": 0.8970099667774086,
      "grad_norm": 12.220060348510742,
      "learning_rate": 1.2150157690573017e-06,
      "loss": 0.105,
      "step": 540
    },
    {
      "epoch": 0.8986710963455149,
      "grad_norm": 10.828849792480469,
      "learning_rate": 1.212385933471873e-06,
      "loss": 0.1782,
      "step": 541
    },
    {
      "epoch": 0.9003322259136213,
      "grad_norm": 7.446874141693115,
      "learning_rate": 1.209754558727322e-06,
      "loss": 0.1279,
      "step": 542
    },
    {
      "epoch": 0.9019933554817275,
      "grad_norm": 10.593635559082031,
      "learning_rate": 1.2071216638932e-06,
      "loss": 0.1389,
      "step": 543
    },
    {
      "epoch": 0.9036544850498339,
      "grad_norm": 6.1521148681640625,
      "learning_rate": 1.204487268050074e-06,
      "loss": 0.08,
      "step": 544
    },
    {
      "epoch": 0.9053156146179402,
      "grad_norm": 7.610264301300049,
      "learning_rate": 1.20185139028939e-06,
      "loss": 0.0942,
      "step": 545
    },
    {
      "epoch": 0.9069767441860465,
      "grad_norm": 4.965451240539551,
      "learning_rate": 1.199214049713332e-06,
      "loss": 0.0551,
      "step": 546
    },
    {
      "epoch": 0.9086378737541528,
      "grad_norm": 8.87406063079834,
      "learning_rate": 1.1965752654346853e-06,
      "loss": 0.1101,
      "step": 547
    },
    {
      "epoch": 0.9102990033222591,
      "grad_norm": 12.057513236999512,
      "learning_rate": 1.1939350565766986e-06,
      "loss": 0.1266,
      "step": 548
    },
    {
      "epoch": 0.9119601328903655,
      "grad_norm": 6.295923233032227,
      "learning_rate": 1.1912934422729433e-06,
      "loss": 0.0802,
      "step": 549
    },
    {
      "epoch": 0.9136212624584718,
      "grad_norm": 8.021347999572754,
      "learning_rate": 1.1886504416671768e-06,
      "loss": 0.1245,
      "step": 550
    },
    {
      "epoch": 0.915282392026578,
      "grad_norm": 19.368101119995117,
      "learning_rate": 1.1860060739132034e-06,
      "loss": 0.1785,
      "step": 551
    },
    {
      "epoch": 0.9169435215946844,
      "grad_norm": 6.273396968841553,
      "learning_rate": 1.1833603581747337e-06,
      "loss": 0.0842,
      "step": 552
    },
    {
      "epoch": 0.9186046511627907,
      "grad_norm": 11.780827522277832,
      "learning_rate": 1.180713313625249e-06,
      "loss": 0.1468,
      "step": 553
    },
    {
      "epoch": 0.920265780730897,
      "grad_norm": 11.886472702026367,
      "learning_rate": 1.1780649594478588e-06,
      "loss": 0.1338,
      "step": 554
    },
    {
      "epoch": 0.9219269102990033,
      "grad_norm": 15.677685737609863,
      "learning_rate": 1.1754153148351638e-06,
      "loss": 0.0981,
      "step": 555
    },
    {
      "epoch": 0.9235880398671097,
      "grad_norm": 15.455180168151855,
      "learning_rate": 1.172764398989118e-06,
      "loss": 0.1265,
      "step": 556
    },
    {
      "epoch": 0.925249169435216,
      "grad_norm": 11.938597679138184,
      "learning_rate": 1.1701122311208858e-06,
      "loss": 0.1704,
      "step": 557
    },
    {
      "epoch": 0.9269102990033222,
      "grad_norm": 6.007061004638672,
      "learning_rate": 1.1674588304507066e-06,
      "loss": 0.0721,
      "step": 558
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 6.777620792388916,
      "learning_rate": 1.1648042162077534e-06,
      "loss": 0.1089,
      "step": 559
    },
    {
      "epoch": 0.9302325581395349,
      "grad_norm": 6.482328414916992,
      "learning_rate": 1.1621484076299933e-06,
      "loss": 0.0737,
      "step": 560
    },
    {
      "epoch": 0.9318936877076412,
      "grad_norm": 13.06513786315918,
      "learning_rate": 1.15949142396405e-06,
      "loss": 0.1332,
      "step": 561
    },
    {
      "epoch": 0.9335548172757475,
      "grad_norm": 15.244009971618652,
      "learning_rate": 1.156833284465062e-06,
      "loss": 0.1236,
      "step": 562
    },
    {
      "epoch": 0.9352159468438538,
      "grad_norm": 9.853615760803223,
      "learning_rate": 1.1541740083965445e-06,
      "loss": 0.0926,
      "step": 563
    },
    {
      "epoch": 0.9368770764119602,
      "grad_norm": 17.81363868713379,
      "learning_rate": 1.1515136150302497e-06,
      "loss": 0.1233,
      "step": 564
    },
    {
      "epoch": 0.9385382059800664,
      "grad_norm": 17.04298973083496,
      "learning_rate": 1.1488521236460265e-06,
      "loss": 0.1688,
      "step": 565
    },
    {
      "epoch": 0.9401993355481728,
      "grad_norm": 17.051122665405273,
      "learning_rate": 1.1461895535316807e-06,
      "loss": 0.1891,
      "step": 566
    },
    {
      "epoch": 0.9418604651162791,
      "grad_norm": 7.387003421783447,
      "learning_rate": 1.143525923982837e-06,
      "loss": 0.1123,
      "step": 567
    },
    {
      "epoch": 0.9435215946843853,
      "grad_norm": 12.299041748046875,
      "learning_rate": 1.1408612543027963e-06,
      "loss": 0.1341,
      "step": 568
    },
    {
      "epoch": 0.9451827242524917,
      "grad_norm": 10.729867935180664,
      "learning_rate": 1.138195563802398e-06,
      "loss": 0.0759,
      "step": 569
    },
    {
      "epoch": 0.946843853820598,
      "grad_norm": 10.060506820678711,
      "learning_rate": 1.1355288717998797e-06,
      "loss": 0.1343,
      "step": 570
    },
    {
      "epoch": 0.9485049833887044,
      "grad_norm": 5.130414009094238,
      "learning_rate": 1.1328611976207355e-06,
      "loss": 0.0635,
      "step": 571
    },
    {
      "epoch": 0.9501661129568106,
      "grad_norm": 8.048657417297363,
      "learning_rate": 1.13019256059758e-06,
      "loss": 0.093,
      "step": 572
    },
    {
      "epoch": 0.9518272425249169,
      "grad_norm": 11.608697891235352,
      "learning_rate": 1.1275229800700028e-06,
      "loss": 0.1367,
      "step": 573
    },
    {
      "epoch": 0.9534883720930233,
      "grad_norm": 9.293169021606445,
      "learning_rate": 1.1248524753844324e-06,
      "loss": 0.1262,
      "step": 574
    },
    {
      "epoch": 0.9551495016611296,
      "grad_norm": 8.593324661254883,
      "learning_rate": 1.1221810658939946e-06,
      "loss": 0.0909,
      "step": 575
    },
    {
      "epoch": 0.9568106312292359,
      "grad_norm": 7.67872953414917,
      "learning_rate": 1.1195087709583713e-06,
      "loss": 0.0845,
      "step": 576
    },
    {
      "epoch": 0.9584717607973422,
      "grad_norm": 5.514711856842041,
      "learning_rate": 1.1168356099436628e-06,
      "loss": 0.0699,
      "step": 577
    },
    {
      "epoch": 0.9601328903654485,
      "grad_norm": 8.198745727539062,
      "learning_rate": 1.1141616022222453e-06,
      "loss": 0.1107,
      "step": 578
    },
    {
      "epoch": 0.9617940199335548,
      "grad_norm": 10.59040355682373,
      "learning_rate": 1.1114867671726302e-06,
      "loss": 0.1206,
      "step": 579
    },
    {
      "epoch": 0.9634551495016611,
      "grad_norm": 10.234724998474121,
      "learning_rate": 1.1088111241793258e-06,
      "loss": 0.0797,
      "step": 580
    },
    {
      "epoch": 0.9651162790697675,
      "grad_norm": 4.5755486488342285,
      "learning_rate": 1.1061346926326942e-06,
      "loss": 0.0451,
      "step": 581
    },
    {
      "epoch": 0.9667774086378738,
      "grad_norm": 12.668401718139648,
      "learning_rate": 1.1034574919288137e-06,
      "loss": 0.0975,
      "step": 582
    },
    {
      "epoch": 0.96843853820598,
      "grad_norm": 4.996627330780029,
      "learning_rate": 1.100779541469336e-06,
      "loss": 0.0908,
      "step": 583
    },
    {
      "epoch": 0.9700996677740864,
      "grad_norm": 7.567203998565674,
      "learning_rate": 1.0981008606613454e-06,
      "loss": 0.0926,
      "step": 584
    },
    {
      "epoch": 0.9717607973421927,
      "grad_norm": 7.270530700683594,
      "learning_rate": 1.0954214689172203e-06,
      "loss": 0.0851,
      "step": 585
    },
    {
      "epoch": 0.973421926910299,
      "grad_norm": 11.763467788696289,
      "learning_rate": 1.0927413856544905e-06,
      "loss": 0.1197,
      "step": 586
    },
    {
      "epoch": 0.9750830564784053,
      "grad_norm": 10.839402198791504,
      "learning_rate": 1.0900606302956978e-06,
      "loss": 0.0805,
      "step": 587
    },
    {
      "epoch": 0.9767441860465116,
      "grad_norm": 8.311976432800293,
      "learning_rate": 1.087379222268254e-06,
      "loss": 0.1222,
      "step": 588
    },
    {
      "epoch": 0.978405315614618,
      "grad_norm": 16.371187210083008,
      "learning_rate": 1.0846971810043015e-06,
      "loss": 0.1755,
      "step": 589
    },
    {
      "epoch": 0.9800664451827242,
      "grad_norm": 8.993597030639648,
      "learning_rate": 1.0820145259405713e-06,
      "loss": 0.1057,
      "step": 590
    },
    {
      "epoch": 0.9817275747508306,
      "grad_norm": 8.658944129943848,
      "learning_rate": 1.0793312765182426e-06,
      "loss": 0.085,
      "step": 591
    },
    {
      "epoch": 0.9833887043189369,
      "grad_norm": 17.43745994567871,
      "learning_rate": 1.076647452182802e-06,
      "loss": 0.1772,
      "step": 592
    },
    {
      "epoch": 0.9850498338870431,
      "grad_norm": 12.07256031036377,
      "learning_rate": 1.0739630723839021e-06,
      "loss": 0.1592,
      "step": 593
    },
    {
      "epoch": 0.9867109634551495,
      "grad_norm": 14.670401573181152,
      "learning_rate": 1.0712781565752222e-06,
      "loss": 0.1656,
      "step": 594
    },
    {
      "epoch": 0.9883720930232558,
      "grad_norm": 18.60042953491211,
      "learning_rate": 1.0685927242143246e-06,
      "loss": 0.142,
      "step": 595
    },
    {
      "epoch": 0.9900332225913622,
      "grad_norm": 7.408260345458984,
      "learning_rate": 1.0659067947625161e-06,
      "loss": 0.1022,
      "step": 596
    },
    {
      "epoch": 0.9916943521594684,
      "grad_norm": 7.936831951141357,
      "learning_rate": 1.0632203876847048e-06,
      "loss": 0.0884,
      "step": 597
    },
    {
      "epoch": 0.9933554817275747,
      "grad_norm": 11.372370719909668,
      "learning_rate": 1.0605335224492615e-06,
      "loss": 0.1498,
      "step": 598
    },
    {
      "epoch": 0.9950166112956811,
      "grad_norm": 7.789324760437012,
      "learning_rate": 1.057846218527876e-06,
      "loss": 0.0774,
      "step": 599
    },
    {
      "epoch": 0.9966777408637874,
      "grad_norm": 11.030444145202637,
      "learning_rate": 1.0551584953954182e-06,
      "loss": 0.1359,
      "step": 600
    },
    {
      "epoch": 0.9983388704318937,
      "grad_norm": 7.336810111999512,
      "learning_rate": 1.0524703725297956e-06,
      "loss": 0.0632,
      "step": 601
    },
    {
      "epoch": 1.0,
      "grad_norm": 10.567761421203613,
      "learning_rate": 1.0497818694118122e-06,
      "loss": 0.1714,
      "step": 602
    },
    {
      "epoch": 1.0016611295681064,
      "grad_norm": 2.572457790374756,
      "learning_rate": 1.0470930055250292e-06,
      "loss": 0.0303,
      "step": 603
    },
    {
      "epoch": 1.0033222591362125,
      "grad_norm": 2.8890881538391113,
      "learning_rate": 1.04440380035562e-06,
      "loss": 0.0319,
      "step": 604
    },
    {
      "epoch": 1.004983388704319,
      "grad_norm": 2.415273904800415,
      "learning_rate": 1.041714273392233e-06,
      "loss": 0.0177,
      "step": 605
    },
    {
      "epoch": 1.0066445182724253,
      "grad_norm": 2.018181800842285,
      "learning_rate": 1.0390244441258478e-06,
      "loss": 0.0181,
      "step": 606
    },
    {
      "epoch": 1.0083056478405317,
      "grad_norm": 1.576615571975708,
      "learning_rate": 1.0363343320496357e-06,
      "loss": 0.0164,
      "step": 607
    },
    {
      "epoch": 1.0099667774086378,
      "grad_norm": 3.9152066707611084,
      "learning_rate": 1.033643956658816e-06,
      "loss": 0.0152,
      "step": 608
    },
    {
      "epoch": 1.0116279069767442,
      "grad_norm": 9.031929969787598,
      "learning_rate": 1.0309533374505182e-06,
      "loss": 0.0461,
      "step": 609
    },
    {
      "epoch": 1.0132890365448506,
      "grad_norm": 3.230257272720337,
      "learning_rate": 1.0282624939236367e-06,
      "loss": 0.017,
      "step": 610
    },
    {
      "epoch": 1.0149501661129567,
      "grad_norm": 7.940582752227783,
      "learning_rate": 1.025571445578693e-06,
      "loss": 0.0407,
      "step": 611
    },
    {
      "epoch": 1.0166112956810631,
      "grad_norm": 5.057779788970947,
      "learning_rate": 1.0228802119176927e-06,
      "loss": 0.0452,
      "step": 612
    },
    {
      "epoch": 1.0182724252491695,
      "grad_norm": 18.299306869506836,
      "learning_rate": 1.0201888124439834e-06,
      "loss": 0.0196,
      "step": 613
    },
    {
      "epoch": 1.0199335548172757,
      "grad_norm": 5.936789035797119,
      "learning_rate": 1.017497266662116e-06,
      "loss": 0.0333,
      "step": 614
    },
    {
      "epoch": 1.021594684385382,
      "grad_norm": 3.513627767562866,
      "learning_rate": 1.0148055940776998e-06,
      "loss": 0.0084,
      "step": 615
    },
    {
      "epoch": 1.0232558139534884,
      "grad_norm": 0.37478646636009216,
      "learning_rate": 1.0121138141972648e-06,
      "loss": 0.0011,
      "step": 616
    },
    {
      "epoch": 1.0249169435215948,
      "grad_norm": 2.6802878379821777,
      "learning_rate": 1.0094219465281172e-06,
      "loss": 0.0082,
      "step": 617
    },
    {
      "epoch": 1.026578073089701,
      "grad_norm": 4.141627311706543,
      "learning_rate": 1.0067300105782002e-06,
      "loss": 0.0133,
      "step": 618
    },
    {
      "epoch": 1.0282392026578073,
      "grad_norm": 1.266526460647583,
      "learning_rate": 1.0040380258559518e-06,
      "loss": 0.0041,
      "step": 619
    },
    {
      "epoch": 1.0299003322259137,
      "grad_norm": 7.685907363891602,
      "learning_rate": 1.001346011870163e-06,
      "loss": 0.0261,
      "step": 620
    },
    {
      "epoch": 1.0315614617940199,
      "grad_norm": 6.315763473510742,
      "learning_rate": 9.986539881298368e-07,
      "loss": 0.0089,
      "step": 621
    },
    {
      "epoch": 1.0332225913621262,
      "grad_norm": 2.4348950386047363,
      "learning_rate": 9.959619741440483e-07,
      "loss": 0.0061,
      "step": 622
    },
    {
      "epoch": 1.0348837209302326,
      "grad_norm": 12.046463012695312,
      "learning_rate": 9.932699894217997e-07,
      "loss": 0.0241,
      "step": 623
    },
    {
      "epoch": 1.0365448504983388,
      "grad_norm": 7.545411586761475,
      "learning_rate": 9.905780534718825e-07,
      "loss": 0.0335,
      "step": 624
    },
    {
      "epoch": 1.0382059800664452,
      "grad_norm": 0.5908049941062927,
      "learning_rate": 9.878861858027353e-07,
      "loss": 0.0015,
      "step": 625
    },
    {
      "epoch": 1.0398671096345515,
      "grad_norm": 0.7220317125320435,
      "learning_rate": 9.851944059223001e-07,
      "loss": 0.0015,
      "step": 626
    },
    {
      "epoch": 1.041528239202658,
      "grad_norm": 0.47070062160491943,
      "learning_rate": 9.825027333378842e-07,
      "loss": 0.001,
      "step": 627
    },
    {
      "epoch": 1.043189368770764,
      "grad_norm": 1.697333574295044,
      "learning_rate": 9.798111875560165e-07,
      "loss": 0.0033,
      "step": 628
    },
    {
      "epoch": 1.0448504983388704,
      "grad_norm": 0.6291197538375854,
      "learning_rate": 9.771197880823072e-07,
      "loss": 0.0008,
      "step": 629
    },
    {
      "epoch": 1.0465116279069768,
      "grad_norm": 56.27286148071289,
      "learning_rate": 9.74428554421307e-07,
      "loss": 0.1389,
      "step": 630
    },
    {
      "epoch": 1.048172757475083,
      "grad_norm": 9.490167617797852,
      "learning_rate": 9.717375060763635e-07,
      "loss": 0.0348,
      "step": 631
    },
    {
      "epoch": 1.0498338870431894,
      "grad_norm": 12.414752006530762,
      "learning_rate": 9.690466625494822e-07,
      "loss": 0.0156,
      "step": 632
    },
    {
      "epoch": 1.0514950166112957,
      "grad_norm": 11.833617210388184,
      "learning_rate": 9.663560433411839e-07,
      "loss": 0.0161,
      "step": 633
    },
    {
      "epoch": 1.053156146179402,
      "grad_norm": 12.894364356994629,
      "learning_rate": 9.636656679503647e-07,
      "loss": 0.0187,
      "step": 634
    },
    {
      "epoch": 1.0548172757475083,
      "grad_norm": 17.071706771850586,
      "learning_rate": 9.609755558741523e-07,
      "loss": 0.096,
      "step": 635
    },
    {
      "epoch": 1.0564784053156147,
      "grad_norm": 0.9374960064888,
      "learning_rate": 9.58285726607767e-07,
      "loss": 0.0029,
      "step": 636
    },
    {
      "epoch": 1.058139534883721,
      "grad_norm": 20.89844512939453,
      "learning_rate": 9.555961996443802e-07,
      "loss": 0.1398,
      "step": 637
    },
    {
      "epoch": 1.0598006644518272,
      "grad_norm": 14.290300369262695,
      "learning_rate": 9.52906994474971e-07,
      "loss": 0.0158,
      "step": 638
    },
    {
      "epoch": 1.0614617940199336,
      "grad_norm": 38.73701477050781,
      "learning_rate": 9.502181305881876e-07,
      "loss": 0.0567,
      "step": 639
    },
    {
      "epoch": 1.06312292358804,
      "grad_norm": 14.144023895263672,
      "learning_rate": 9.475296274702043e-07,
      "loss": 0.0168,
      "step": 640
    },
    {
      "epoch": 1.064784053156146,
      "grad_norm": 16.016897201538086,
      "learning_rate": 9.448415046045817e-07,
      "loss": 0.0747,
      "step": 641
    },
    {
      "epoch": 1.0664451827242525,
      "grad_norm": 1.7351895570755005,
      "learning_rate": 9.42153781472124e-07,
      "loss": 0.0036,
      "step": 642
    },
    {
      "epoch": 1.0681063122923589,
      "grad_norm": 0.4755818545818329,
      "learning_rate": 9.394664775507385e-07,
      "loss": 0.0007,
      "step": 643
    },
    {
      "epoch": 1.069767441860465,
      "grad_norm": 2.340486526489258,
      "learning_rate": 9.367796123152951e-07,
      "loss": 0.0104,
      "step": 644
    },
    {
      "epoch": 1.0714285714285714,
      "grad_norm": 5.121689796447754,
      "learning_rate": 9.340932052374839e-07,
      "loss": 0.006,
      "step": 645
    },
    {
      "epoch": 1.0730897009966778,
      "grad_norm": 9.980019569396973,
      "learning_rate": 9.314072757856752e-07,
      "loss": 0.024,
      "step": 646
    },
    {
      "epoch": 1.0747508305647842,
      "grad_norm": 12.077081680297852,
      "learning_rate": 9.287218434247776e-07,
      "loss": 0.0776,
      "step": 647
    },
    {
      "epoch": 1.0764119601328903,
      "grad_norm": 10.7062349319458,
      "learning_rate": 9.260369276160976e-07,
      "loss": 0.0303,
      "step": 648
    },
    {
      "epoch": 1.0780730897009967,
      "grad_norm": 7.534442901611328,
      "learning_rate": 9.233525478171984e-07,
      "loss": 0.0474,
      "step": 649
    },
    {
      "epoch": 1.079734219269103,
      "grad_norm": 1.34402334690094,
      "learning_rate": 9.206687234817574e-07,
      "loss": 0.0015,
      "step": 650
    },
    {
      "epoch": 1.0813953488372092,
      "grad_norm": 2.5344579219818115,
      "learning_rate": 9.179854740594285e-07,
      "loss": 0.0074,
      "step": 651
    },
    {
      "epoch": 1.0830564784053156,
      "grad_norm": 4.098090648651123,
      "learning_rate": 9.153028189956985e-07,
      "loss": 0.0106,
      "step": 652
    },
    {
      "epoch": 1.084717607973422,
      "grad_norm": 8.03905200958252,
      "learning_rate": 9.126207777317457e-07,
      "loss": 0.0158,
      "step": 653
    },
    {
      "epoch": 1.0863787375415281,
      "grad_norm": 11.426288604736328,
      "learning_rate": 9.099393697043023e-07,
      "loss": 0.024,
      "step": 654
    },
    {
      "epoch": 1.0880398671096345,
      "grad_norm": 18.223331451416016,
      "learning_rate": 9.072586143455095e-07,
      "loss": 0.079,
      "step": 655
    },
    {
      "epoch": 1.089700996677741,
      "grad_norm": 6.2648844718933105,
      "learning_rate": 9.045785310827799e-07,
      "loss": 0.0076,
      "step": 656
    },
    {
      "epoch": 1.0913621262458473,
      "grad_norm": 22.87317657470703,
      "learning_rate": 9.018991393386548e-07,
      "loss": 0.0205,
      "step": 657
    },
    {
      "epoch": 1.0930232558139534,
      "grad_norm": 9.388856887817383,
      "learning_rate": 8.99220458530664e-07,
      "loss": 0.0443,
      "step": 658
    },
    {
      "epoch": 1.0946843853820598,
      "grad_norm": 4.490157127380371,
      "learning_rate": 8.965425080711863e-07,
      "loss": 0.014,
      "step": 659
    },
    {
      "epoch": 1.0963455149501662,
      "grad_norm": 1.3594952821731567,
      "learning_rate": 8.938653073673057e-07,
      "loss": 0.0053,
      "step": 660
    },
    {
      "epoch": 1.0980066445182723,
      "grad_norm": 8.04047679901123,
      "learning_rate": 8.911888758206746e-07,
      "loss": 0.036,
      "step": 661
    },
    {
      "epoch": 1.0996677740863787,
      "grad_norm": 11.46448040008545,
      "learning_rate": 8.885132328273699e-07,
      "loss": 0.0622,
      "step": 662
    },
    {
      "epoch": 1.101328903654485,
      "grad_norm": 9.573933601379395,
      "learning_rate": 8.858383977777549e-07,
      "loss": 0.052,
      "step": 663
    },
    {
      "epoch": 1.1029900332225913,
      "grad_norm": 13.688606262207031,
      "learning_rate": 8.831643900563372e-07,
      "loss": 0.0361,
      "step": 664
    },
    {
      "epoch": 1.1046511627906976,
      "grad_norm": 2.0683696269989014,
      "learning_rate": 8.804912290416286e-07,
      "loss": 0.0057,
      "step": 665
    },
    {
      "epoch": 1.106312292358804,
      "grad_norm": 2.5601813793182373,
      "learning_rate": 8.778189341060058e-07,
      "loss": 0.0032,
      "step": 666
    },
    {
      "epoch": 1.1079734219269104,
      "grad_norm": 14.53157901763916,
      "learning_rate": 8.751475246155677e-07,
      "loss": 0.0506,
      "step": 667
    },
    {
      "epoch": 1.1096345514950166,
      "grad_norm": 4.818484783172607,
      "learning_rate": 8.724770199299972e-07,
      "loss": 0.018,
      "step": 668
    },
    {
      "epoch": 1.111295681063123,
      "grad_norm": 25.98501205444336,
      "learning_rate": 8.6980743940242e-07,
      "loss": 0.188,
      "step": 669
    },
    {
      "epoch": 1.1129568106312293,
      "grad_norm": 29.606630325317383,
      "learning_rate": 8.67138802379264e-07,
      "loss": 0.0453,
      "step": 670
    },
    {
      "epoch": 1.1146179401993355,
      "grad_norm": 9.709136962890625,
      "learning_rate": 8.644711282001207e-07,
      "loss": 0.0377,
      "step": 671
    },
    {
      "epoch": 1.1162790697674418,
      "grad_norm": 8.595226287841797,
      "learning_rate": 8.61804436197602e-07,
      "loss": 0.0385,
      "step": 672
    },
    {
      "epoch": 1.1179401993355482,
      "grad_norm": 2.2285189628601074,
      "learning_rate": 8.591387456972039e-07,
      "loss": 0.0076,
      "step": 673
    },
    {
      "epoch": 1.1196013289036544,
      "grad_norm": 7.5171990394592285,
      "learning_rate": 8.564740760171631e-07,
      "loss": 0.0803,
      "step": 674
    },
    {
      "epoch": 1.1212624584717608,
      "grad_norm": 12.671175956726074,
      "learning_rate": 8.53810446468319e-07,
      "loss": 0.0193,
      "step": 675
    },
    {
      "epoch": 1.1229235880398671,
      "grad_norm": 6.085766315460205,
      "learning_rate": 8.511478763539736e-07,
      "loss": 0.0109,
      "step": 676
    },
    {
      "epoch": 1.1245847176079735,
      "grad_norm": 8.109288215637207,
      "learning_rate": 8.484863849697501e-07,
      "loss": 0.0229,
      "step": 677
    },
    {
      "epoch": 1.1262458471760797,
      "grad_norm": 11.192323684692383,
      "learning_rate": 8.458259916034556e-07,
      "loss": 0.0367,
      "step": 678
    },
    {
      "epoch": 1.127906976744186,
      "grad_norm": 6.7384796142578125,
      "learning_rate": 8.43166715534938e-07,
      "loss": 0.0242,
      "step": 679
    },
    {
      "epoch": 1.1295681063122924,
      "grad_norm": 6.001007080078125,
      "learning_rate": 8.405085760359499e-07,
      "loss": 0.0387,
      "step": 680
    },
    {
      "epoch": 1.1312292358803986,
      "grad_norm": 4.371544361114502,
      "learning_rate": 8.378515923700069e-07,
      "loss": 0.0132,
      "step": 681
    },
    {
      "epoch": 1.132890365448505,
      "grad_norm": 10.13823413848877,
      "learning_rate": 8.351957837922467e-07,
      "loss": 0.0236,
      "step": 682
    },
    {
      "epoch": 1.1345514950166113,
      "grad_norm": 6.677521228790283,
      "learning_rate": 8.325411695492936e-07,
      "loss": 0.0451,
      "step": 683
    },
    {
      "epoch": 1.1362126245847177,
      "grad_norm": 9.64169692993164,
      "learning_rate": 8.298877688791143e-07,
      "loss": 0.0266,
      "step": 684
    },
    {
      "epoch": 1.1378737541528239,
      "grad_norm": 15.662538528442383,
      "learning_rate": 8.27235601010882e-07,
      "loss": 0.0218,
      "step": 685
    },
    {
      "epoch": 1.1395348837209303,
      "grad_norm": 8.98953628540039,
      "learning_rate": 8.245846851648363e-07,
      "loss": 0.031,
      "step": 686
    },
    {
      "epoch": 1.1411960132890366,
      "grad_norm": 2.669698476791382,
      "learning_rate": 8.219350405521414e-07,
      "loss": 0.0055,
      "step": 687
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 4.246596813201904,
      "learning_rate": 8.192866863747514e-07,
      "loss": 0.0153,
      "step": 688
    },
    {
      "epoch": 1.1445182724252492,
      "grad_norm": 8.687833786010742,
      "learning_rate": 8.166396418252663e-07,
      "loss": 0.0387,
      "step": 689
    },
    {
      "epoch": 1.1461794019933556,
      "grad_norm": 10.32325267791748,
      "learning_rate": 8.139939260867965e-07,
      "loss": 0.0539,
      "step": 690
    },
    {
      "epoch": 1.1478405315614617,
      "grad_norm": 3.5581681728363037,
      "learning_rate": 8.113495583328232e-07,
      "loss": 0.0145,
      "step": 691
    },
    {
      "epoch": 1.149501661129568,
      "grad_norm": 0.5473489165306091,
      "learning_rate": 8.087065577270569e-07,
      "loss": 0.0009,
      "step": 692
    },
    {
      "epoch": 1.1511627906976745,
      "grad_norm": 4.436628818511963,
      "learning_rate": 8.060649434233016e-07,
      "loss": 0.0103,
      "step": 693
    },
    {
      "epoch": 1.1528239202657806,
      "grad_norm": 12.929503440856934,
      "learning_rate": 8.034247345653147e-07,
      "loss": 0.0396,
      "step": 694
    },
    {
      "epoch": 1.154485049833887,
      "grad_norm": 15.860251426696777,
      "learning_rate": 8.007859502866683e-07,
      "loss": 0.0702,
      "step": 695
    },
    {
      "epoch": 1.1561461794019934,
      "grad_norm": 1.6955807209014893,
      "learning_rate": 7.9814860971061e-07,
      "loss": 0.0069,
      "step": 696
    },
    {
      "epoch": 1.1578073089700998,
      "grad_norm": 3.191066265106201,
      "learning_rate": 7.955127319499257e-07,
      "loss": 0.0143,
      "step": 697
    },
    {
      "epoch": 1.159468438538206,
      "grad_norm": 1.4074821472167969,
      "learning_rate": 7.928783361068001e-07,
      "loss": 0.0024,
      "step": 698
    },
    {
      "epoch": 1.1611295681063123,
      "grad_norm": 2.3482460975646973,
      "learning_rate": 7.90245441272678e-07,
      "loss": 0.01,
      "step": 699
    },
    {
      "epoch": 1.1627906976744187,
      "grad_norm": 14.107357025146484,
      "learning_rate": 7.876140665281271e-07,
      "loss": 0.0377,
      "step": 700
    },
    {
      "epoch": 1.1644518272425248,
      "grad_norm": 19.679927825927734,
      "learning_rate": 7.849842309426983e-07,
      "loss": 0.0264,
      "step": 701
    },
    {
      "epoch": 1.1661129568106312,
      "grad_norm": 32.07101058959961,
      "learning_rate": 7.823559535747885e-07,
      "loss": 0.1096,
      "step": 702
    },
    {
      "epoch": 1.1677740863787376,
      "grad_norm": 14.08077621459961,
      "learning_rate": 7.79729253471503e-07,
      "loss": 0.1361,
      "step": 703
    },
    {
      "epoch": 1.169435215946844,
      "grad_norm": 1.9143798351287842,
      "learning_rate": 7.771041496685149e-07,
      "loss": 0.0043,
      "step": 704
    },
    {
      "epoch": 1.1710963455149501,
      "grad_norm": 6.996606349945068,
      "learning_rate": 7.744806611899309e-07,
      "loss": 0.0085,
      "step": 705
    },
    {
      "epoch": 1.1727574750830565,
      "grad_norm": 31.96712875366211,
      "learning_rate": 7.7185880704815e-07,
      "loss": 0.0564,
      "step": 706
    },
    {
      "epoch": 1.1744186046511629,
      "grad_norm": 5.491878986358643,
      "learning_rate": 7.692386062437279e-07,
      "loss": 0.0083,
      "step": 707
    },
    {
      "epoch": 1.176079734219269,
      "grad_norm": 7.473092079162598,
      "learning_rate": 7.666200777652384e-07,
      "loss": 0.0191,
      "step": 708
    },
    {
      "epoch": 1.1777408637873754,
      "grad_norm": 22.00745964050293,
      "learning_rate": 7.640032405891359e-07,
      "loss": 0.0485,
      "step": 709
    },
    {
      "epoch": 1.1794019933554818,
      "grad_norm": 7.070401668548584,
      "learning_rate": 7.613881136796183e-07,
      "loss": 0.0273,
      "step": 710
    },
    {
      "epoch": 1.181063122923588,
      "grad_norm": 6.378243923187256,
      "learning_rate": 7.587747159884881e-07,
      "loss": 0.0227,
      "step": 711
    },
    {
      "epoch": 1.1827242524916943,
      "grad_norm": 3.251873254776001,
      "learning_rate": 7.561630664550179e-07,
      "loss": 0.0059,
      "step": 712
    },
    {
      "epoch": 1.1843853820598007,
      "grad_norm": 0.48984941840171814,
      "learning_rate": 7.535531840058101e-07,
      "loss": 0.0006,
      "step": 713
    },
    {
      "epoch": 1.1860465116279069,
      "grad_norm": 14.683696746826172,
      "learning_rate": 7.509450875546615e-07,
      "loss": 0.0288,
      "step": 714
    },
    {
      "epoch": 1.1877076411960132,
      "grad_norm": 5.808125972747803,
      "learning_rate": 7.48338796002426e-07,
      "loss": 0.0081,
      "step": 715
    },
    {
      "epoch": 1.1893687707641196,
      "grad_norm": 0.9389947056770325,
      "learning_rate": 7.457343282368763e-07,
      "loss": 0.0014,
      "step": 716
    },
    {
      "epoch": 1.191029900332226,
      "grad_norm": 18.770702362060547,
      "learning_rate": 7.431317031325703e-07,
      "loss": 0.0237,
      "step": 717
    },
    {
      "epoch": 1.1926910299003322,
      "grad_norm": 4.809552192687988,
      "learning_rate": 7.405309395507096e-07,
      "loss": 0.0111,
      "step": 718
    },
    {
      "epoch": 1.1943521594684385,
      "grad_norm": 14.019580841064453,
      "learning_rate": 7.37932056339007e-07,
      "loss": 0.0242,
      "step": 719
    },
    {
      "epoch": 1.196013289036545,
      "grad_norm": 1.6724600791931152,
      "learning_rate": 7.353350723315485e-07,
      "loss": 0.0013,
      "step": 720
    },
    {
      "epoch": 1.197674418604651,
      "grad_norm": 24.528261184692383,
      "learning_rate": 7.327400063486553e-07,
      "loss": 0.0404,
      "step": 721
    },
    {
      "epoch": 1.1993355481727574,
      "grad_norm": 9.839111328125,
      "learning_rate": 7.301468771967495e-07,
      "loss": 0.022,
      "step": 722
    },
    {
      "epoch": 1.2009966777408638,
      "grad_norm": 169.1666259765625,
      "learning_rate": 7.275557036682167e-07,
      "loss": 0.0955,
      "step": 723
    },
    {
      "epoch": 1.2026578073089702,
      "grad_norm": 3.6535372734069824,
      "learning_rate": 7.249665045412702e-07,
      "loss": 0.0075,
      "step": 724
    },
    {
      "epoch": 1.2043189368770764,
      "grad_norm": 16.60807228088379,
      "learning_rate": 7.223792985798156e-07,
      "loss": 0.0245,
      "step": 725
    },
    {
      "epoch": 1.2059800664451827,
      "grad_norm": 6.681334018707275,
      "learning_rate": 7.197941045333125e-07,
      "loss": 0.0332,
      "step": 726
    },
    {
      "epoch": 1.2076411960132891,
      "grad_norm": 7.528415203094482,
      "learning_rate": 7.172109411366416e-07,
      "loss": 0.0486,
      "step": 727
    },
    {
      "epoch": 1.2093023255813953,
      "grad_norm": 7.378742694854736,
      "learning_rate": 7.146298271099665e-07,
      "loss": 0.0142,
      "step": 728
    },
    {
      "epoch": 1.2109634551495017,
      "grad_norm": 0.4893551468849182,
      "learning_rate": 7.120507811585994e-07,
      "loss": 0.0008,
      "step": 729
    },
    {
      "epoch": 1.212624584717608,
      "grad_norm": 12.16398811340332,
      "learning_rate": 7.094738219728656e-07,
      "loss": 0.0503,
      "step": 730
    },
    {
      "epoch": 1.2142857142857142,
      "grad_norm": 0.8328701257705688,
      "learning_rate": 7.068989682279669e-07,
      "loss": 0.0011,
      "step": 731
    },
    {
      "epoch": 1.2159468438538206,
      "grad_norm": 5.240982532501221,
      "learning_rate": 7.043262385838475e-07,
      "loss": 0.0211,
      "step": 732
    },
    {
      "epoch": 1.217607973421927,
      "grad_norm": 3.157712936401367,
      "learning_rate": 7.01755651685058e-07,
      "loss": 0.0129,
      "step": 733
    },
    {
      "epoch": 1.219269102990033,
      "grad_norm": 5.299394607543945,
      "learning_rate": 6.991872261606206e-07,
      "loss": 0.0166,
      "step": 734
    },
    {
      "epoch": 1.2209302325581395,
      "grad_norm": 5.783647060394287,
      "learning_rate": 6.966209806238942e-07,
      "loss": 0.024,
      "step": 735
    },
    {
      "epoch": 1.2225913621262459,
      "grad_norm": 29.13331413269043,
      "learning_rate": 6.940569336724389e-07,
      "loss": 0.0335,
      "step": 736
    },
    {
      "epoch": 1.2242524916943522,
      "grad_norm": 8.021717071533203,
      "learning_rate": 6.914951038878826e-07,
      "loss": 0.0254,
      "step": 737
    },
    {
      "epoch": 1.2259136212624584,
      "grad_norm": 18.33582305908203,
      "learning_rate": 6.889355098357841e-07,
      "loss": 0.0814,
      "step": 738
    },
    {
      "epoch": 1.2275747508305648,
      "grad_norm": 3.572312116622925,
      "learning_rate": 6.863781700655012e-07,
      "loss": 0.0147,
      "step": 739
    },
    {
      "epoch": 1.2292358803986712,
      "grad_norm": 2.5561416149139404,
      "learning_rate": 6.838231031100532e-07,
      "loss": 0.0034,
      "step": 740
    },
    {
      "epoch": 1.2308970099667773,
      "grad_norm": 3.5589852333068848,
      "learning_rate": 6.812703274859905e-07,
      "loss": 0.0022,
      "step": 741
    },
    {
      "epoch": 1.2325581395348837,
      "grad_norm": 3.630929946899414,
      "learning_rate": 6.78719861693257e-07,
      "loss": 0.0095,
      "step": 742
    },
    {
      "epoch": 1.23421926910299,
      "grad_norm": 16.794384002685547,
      "learning_rate": 6.761717242150571e-07,
      "loss": 0.0761,
      "step": 743
    },
    {
      "epoch": 1.2358803986710964,
      "grad_norm": 4.896450042724609,
      "learning_rate": 6.736259335177228e-07,
      "loss": 0.0075,
      "step": 744
    },
    {
      "epoch": 1.2375415282392026,
      "grad_norm": 6.048307418823242,
      "learning_rate": 6.710825080505774e-07,
      "loss": 0.0125,
      "step": 745
    },
    {
      "epoch": 1.239202657807309,
      "grad_norm": 12.081902503967285,
      "learning_rate": 6.685414662458054e-07,
      "loss": 0.0468,
      "step": 746
    },
    {
      "epoch": 1.2408637873754154,
      "grad_norm": 30.948566436767578,
      "learning_rate": 6.660028265183159e-07,
      "loss": 0.0549,
      "step": 747
    },
    {
      "epoch": 1.2425249169435215,
      "grad_norm": 5.927905082702637,
      "learning_rate": 6.634666072656097e-07,
      "loss": 0.0345,
      "step": 748
    },
    {
      "epoch": 1.244186046511628,
      "grad_norm": 15.502345085144043,
      "learning_rate": 6.609328268676476e-07,
      "loss": 0.0928,
      "step": 749
    },
    {
      "epoch": 1.2458471760797343,
      "grad_norm": 6.187903881072998,
      "learning_rate": 6.584015036867149e-07,
      "loss": 0.0139,
      "step": 750
    },
    {
      "epoch": 1.2475083056478407,
      "grad_norm": 2.137287139892578,
      "learning_rate": 6.558726560672898e-07,
      "loss": 0.0027,
      "step": 751
    },
    {
      "epoch": 1.2491694352159468,
      "grad_norm": 11.991990089416504,
      "learning_rate": 6.533463023359114e-07,
      "loss": 0.0323,
      "step": 752
    },
    {
      "epoch": 1.2508305647840532,
      "grad_norm": 4.784742832183838,
      "learning_rate": 6.508224608010435e-07,
      "loss": 0.0105,
      "step": 753
    },
    {
      "epoch": 1.2524916943521593,
      "grad_norm": 11.65071964263916,
      "learning_rate": 6.483011497529456e-07,
      "loss": 0.0182,
      "step": 754
    },
    {
      "epoch": 1.2541528239202657,
      "grad_norm": 18.266075134277344,
      "learning_rate": 6.457823874635376e-07,
      "loss": 0.0502,
      "step": 755
    },
    {
      "epoch": 1.255813953488372,
      "grad_norm": 8.90103530883789,
      "learning_rate": 6.432661921862699e-07,
      "loss": 0.0364,
      "step": 756
    },
    {
      "epoch": 1.2574750830564785,
      "grad_norm": 0.34128686785697937,
      "learning_rate": 6.407525821559888e-07,
      "loss": 0.001,
      "step": 757
    },
    {
      "epoch": 1.2591362126245846,
      "grad_norm": 8.536415100097656,
      "learning_rate": 6.382415755888053e-07,
      "loss": 0.0258,
      "step": 758
    },
    {
      "epoch": 1.260797342192691,
      "grad_norm": 0.3508094549179077,
      "learning_rate": 6.35733190681964e-07,
      "loss": 0.006,
      "step": 759
    },
    {
      "epoch": 1.2624584717607974,
      "grad_norm": 5.0073628425598145,
      "learning_rate": 6.332274456137096e-07,
      "loss": 0.0098,
      "step": 760
    },
    {
      "epoch": 1.2641196013289036,
      "grad_norm": 6.868859767913818,
      "learning_rate": 6.307243585431562e-07,
      "loss": 0.03,
      "step": 761
    },
    {
      "epoch": 1.26578073089701,
      "grad_norm": 5.491968154907227,
      "learning_rate": 6.282239476101549e-07,
      "loss": 0.004,
      "step": 762
    },
    {
      "epoch": 1.2674418604651163,
      "grad_norm": 16.57515525817871,
      "learning_rate": 6.257262309351636e-07,
      "loss": 0.0318,
      "step": 763
    },
    {
      "epoch": 1.2691029900332227,
      "grad_norm": 10.209383010864258,
      "learning_rate": 6.232312266191149e-07,
      "loss": 0.0263,
      "step": 764
    },
    {
      "epoch": 1.2707641196013288,
      "grad_norm": 10.865987777709961,
      "learning_rate": 6.20738952743284e-07,
      "loss": 0.0335,
      "step": 765
    },
    {
      "epoch": 1.2724252491694352,
      "grad_norm": 5.695135593414307,
      "learning_rate": 6.1824942736916e-07,
      "loss": 0.0256,
      "step": 766
    },
    {
      "epoch": 1.2740863787375416,
      "grad_norm": 2.8923935890197754,
      "learning_rate": 6.157626685383123e-07,
      "loss": 0.0047,
      "step": 767
    },
    {
      "epoch": 1.2757475083056478,
      "grad_norm": 4.0880842208862305,
      "learning_rate": 6.13278694272262e-07,
      "loss": 0.0114,
      "step": 768
    },
    {
      "epoch": 1.2774086378737541,
      "grad_norm": 22.687896728515625,
      "learning_rate": 6.107975225723505e-07,
      "loss": 0.0596,
      "step": 769
    },
    {
      "epoch": 1.2790697674418605,
      "grad_norm": 0.6194286346435547,
      "learning_rate": 6.083191714196085e-07,
      "loss": 0.0012,
      "step": 770
    },
    {
      "epoch": 1.280730897009967,
      "grad_norm": 21.863033294677734,
      "learning_rate": 6.058436587746269e-07,
      "loss": 0.0596,
      "step": 771
    },
    {
      "epoch": 1.282392026578073,
      "grad_norm": 1.8822462558746338,
      "learning_rate": 6.033710025774253e-07,
      "loss": 0.0043,
      "step": 772
    },
    {
      "epoch": 1.2840531561461794,
      "grad_norm": 6.4909515380859375,
      "learning_rate": 6.009012207473228e-07,
      "loss": 0.0152,
      "step": 773
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 1.9386980533599854,
      "learning_rate": 5.984343311828086e-07,
      "loss": 0.0061,
      "step": 774
    },
    {
      "epoch": 1.287375415282392,
      "grad_norm": 8.743645668029785,
      "learning_rate": 5.959703517614106e-07,
      "loss": 0.0134,
      "step": 775
    },
    {
      "epoch": 1.2890365448504983,
      "grad_norm": 23.125574111938477,
      "learning_rate": 5.935093003395684e-07,
      "loss": 0.0258,
      "step": 776
    },
    {
      "epoch": 1.2906976744186047,
      "grad_norm": 2.1917898654937744,
      "learning_rate": 5.910511947525001e-07,
      "loss": 0.0058,
      "step": 777
    },
    {
      "epoch": 1.292358803986711,
      "grad_norm": 4.327230930328369,
      "learning_rate": 5.885960528140783e-07,
      "loss": 0.0111,
      "step": 778
    },
    {
      "epoch": 1.2940199335548173,
      "grad_norm": 9.585992813110352,
      "learning_rate": 5.861438923166955e-07,
      "loss": 0.0214,
      "step": 779
    },
    {
      "epoch": 1.2956810631229236,
      "grad_norm": 16.866348266601562,
      "learning_rate": 5.836947310311388e-07,
      "loss": 0.0145,
      "step": 780
    },
    {
      "epoch": 1.2973421926910298,
      "grad_norm": 29.587238311767578,
      "learning_rate": 5.812485867064607e-07,
      "loss": 0.0262,
      "step": 781
    },
    {
      "epoch": 1.2990033222591362,
      "grad_norm": 3.231372117996216,
      "learning_rate": 5.788054770698487e-07,
      "loss": 0.0076,
      "step": 782
    },
    {
      "epoch": 1.3006644518272426,
      "grad_norm": 8.976773262023926,
      "learning_rate": 5.763654198264985e-07,
      "loss": 0.0182,
      "step": 783
    },
    {
      "epoch": 1.302325581395349,
      "grad_norm": 9.699551582336426,
      "learning_rate": 5.739284326594844e-07,
      "loss": 0.0214,
      "step": 784
    },
    {
      "epoch": 1.303986710963455,
      "grad_norm": 25.53984832763672,
      "learning_rate": 5.714945332296333e-07,
      "loss": 0.0154,
      "step": 785
    },
    {
      "epoch": 1.3056478405315615,
      "grad_norm": 1.9964032173156738,
      "learning_rate": 5.690637391753941e-07,
      "loss": 0.0043,
      "step": 786
    },
    {
      "epoch": 1.3073089700996678,
      "grad_norm": 28.76060676574707,
      "learning_rate": 5.666360681127108e-07,
      "loss": 0.0258,
      "step": 787
    },
    {
      "epoch": 1.308970099667774,
      "grad_norm": 3.6327102184295654,
      "learning_rate": 5.642115376348972e-07,
      "loss": 0.0049,
      "step": 788
    },
    {
      "epoch": 1.3106312292358804,
      "grad_norm": 1.5518959760665894,
      "learning_rate": 5.617901653125042e-07,
      "loss": 0.0064,
      "step": 789
    },
    {
      "epoch": 1.3122923588039868,
      "grad_norm": 5.097245216369629,
      "learning_rate": 5.59371968693198e-07,
      "loss": 0.0039,
      "step": 790
    },
    {
      "epoch": 1.3139534883720931,
      "grad_norm": 1.6257424354553223,
      "learning_rate": 5.569569653016299e-07,
      "loss": 0.002,
      "step": 791
    },
    {
      "epoch": 1.3156146179401993,
      "grad_norm": 1.6841596364974976,
      "learning_rate": 5.545451726393097e-07,
      "loss": 0.0031,
      "step": 792
    },
    {
      "epoch": 1.3172757475083057,
      "grad_norm": 33.802764892578125,
      "learning_rate": 5.521366081844787e-07,
      "loss": 0.0603,
      "step": 793
    },
    {
      "epoch": 1.3189368770764118,
      "grad_norm": 6.568824291229248,
      "learning_rate": 5.497312893919837e-07,
      "loss": 0.0082,
      "step": 794
    },
    {
      "epoch": 1.3205980066445182,
      "grad_norm": 2.7456605434417725,
      "learning_rate": 5.47329233693151e-07,
      "loss": 0.0081,
      "step": 795
    },
    {
      "epoch": 1.3222591362126246,
      "grad_norm": 27.489286422729492,
      "learning_rate": 5.449304584956581e-07,
      "loss": 0.1081,
      "step": 796
    },
    {
      "epoch": 1.323920265780731,
      "grad_norm": 4.846098899841309,
      "learning_rate": 5.42534981183409e-07,
      "loss": 0.0064,
      "step": 797
    },
    {
      "epoch": 1.3255813953488373,
      "grad_norm": 9.961216926574707,
      "learning_rate": 5.401428191164092e-07,
      "loss": 0.0215,
      "step": 798
    },
    {
      "epoch": 1.3272425249169435,
      "grad_norm": 10.51478099822998,
      "learning_rate": 5.377539896306363e-07,
      "loss": 0.0214,
      "step": 799
    },
    {
      "epoch": 1.3289036544850499,
      "grad_norm": 23.659414291381836,
      "learning_rate": 5.353685100379189e-07,
      "loss": 0.0265,
      "step": 800
    },
    {
      "epoch": 1.330564784053156,
      "grad_norm": 6.710888862609863,
      "learning_rate": 5.32986397625808e-07,
      "loss": 0.0251,
      "step": 801
    },
    {
      "epoch": 1.3322259136212624,
      "grad_norm": 8.738547325134277,
      "learning_rate": 5.306076696574522e-07,
      "loss": 0.012,
      "step": 802
    },
    {
      "epoch": 1.3338870431893688,
      "grad_norm": 16.569536209106445,
      "learning_rate": 5.282323433714743e-07,
      "loss": 0.0281,
      "step": 803
    },
    {
      "epoch": 1.3355481727574752,
      "grad_norm": 15.599034309387207,
      "learning_rate": 5.258604359818443e-07,
      "loss": 0.0395,
      "step": 804
    },
    {
      "epoch": 1.3372093023255813,
      "grad_norm": 11.257906913757324,
      "learning_rate": 5.234919646777555e-07,
      "loss": 0.0456,
      "step": 805
    },
    {
      "epoch": 1.3388704318936877,
      "grad_norm": 9.306239128112793,
      "learning_rate": 5.211269466235e-07,
      "loss": 0.0675,
      "step": 806
    },
    {
      "epoch": 1.340531561461794,
      "grad_norm": 1.2154159545898438,
      "learning_rate": 5.187653989583445e-07,
      "loss": 0.0017,
      "step": 807
    },
    {
      "epoch": 1.3421926910299002,
      "grad_norm": 0.9059697389602661,
      "learning_rate": 5.164073387964057e-07,
      "loss": 0.0016,
      "step": 808
    },
    {
      "epoch": 1.3438538205980066,
      "grad_norm": 0.11658035218715668,
      "learning_rate": 5.140527832265255e-07,
      "loss": 0.0002,
      "step": 809
    },
    {
      "epoch": 1.345514950166113,
      "grad_norm": 22.50667953491211,
      "learning_rate": 5.117017493121501e-07,
      "loss": 0.06,
      "step": 810
    },
    {
      "epoch": 1.3471760797342194,
      "grad_norm": 20.302230834960938,
      "learning_rate": 5.093542540912024e-07,
      "loss": 0.0738,
      "step": 811
    },
    {
      "epoch": 1.3488372093023255,
      "grad_norm": 9.267167091369629,
      "learning_rate": 5.070103145759605e-07,
      "loss": 0.011,
      "step": 812
    },
    {
      "epoch": 1.350498338870432,
      "grad_norm": 13.021448135375977,
      "learning_rate": 5.046699477529359e-07,
      "loss": 0.0318,
      "step": 813
    },
    {
      "epoch": 1.352159468438538,
      "grad_norm": 3.4434173107147217,
      "learning_rate": 5.023331705827476e-07,
      "loss": 0.0078,
      "step": 814
    },
    {
      "epoch": 1.3538205980066444,
      "grad_norm": 3.8246960639953613,
      "learning_rate": 5.000000000000002e-07,
      "loss": 0.0071,
      "step": 815
    },
    {
      "epoch": 1.3554817275747508,
      "grad_norm": 22.36045265197754,
      "learning_rate": 4.976704529131616e-07,
      "loss": 0.0593,
      "step": 816
    },
    {
      "epoch": 1.3571428571428572,
      "grad_norm": 7.905923366546631,
      "learning_rate": 4.953445462044414e-07,
      "loss": 0.02,
      "step": 817
    },
    {
      "epoch": 1.3588039867109636,
      "grad_norm": 13.977715492248535,
      "learning_rate": 4.930222967296661e-07,
      "loss": 0.059,
      "step": 818
    },
    {
      "epoch": 1.3604651162790697,
      "grad_norm": 0.6750720739364624,
      "learning_rate": 4.90703721318158e-07,
      "loss": 0.0012,
      "step": 819
    },
    {
      "epoch": 1.3621262458471761,
      "grad_norm": 2.0304207801818848,
      "learning_rate": 4.883888367726152e-07,
      "loss": 0.0036,
      "step": 820
    },
    {
      "epoch": 1.3637873754152823,
      "grad_norm": 9.32971477508545,
      "learning_rate": 4.860776598689865e-07,
      "loss": 0.0328,
      "step": 821
    },
    {
      "epoch": 1.3654485049833887,
      "grad_norm": 11.44428825378418,
      "learning_rate": 4.837702073563521e-07,
      "loss": 0.0399,
      "step": 822
    },
    {
      "epoch": 1.367109634551495,
      "grad_norm": 15.22606086730957,
      "learning_rate": 4.81466495956801e-07,
      "loss": 0.0632,
      "step": 823
    },
    {
      "epoch": 1.3687707641196014,
      "grad_norm": 5.094400882720947,
      "learning_rate": 4.791665423653117e-07,
      "loss": 0.0126,
      "step": 824
    },
    {
      "epoch": 1.3704318936877076,
      "grad_norm": 11.821290016174316,
      "learning_rate": 4.768703632496287e-07,
      "loss": 0.0196,
      "step": 825
    },
    {
      "epoch": 1.372093023255814,
      "grad_norm": 4.006501197814941,
      "learning_rate": 4.745779752501429e-07,
      "loss": 0.0044,
      "step": 826
    },
    {
      "epoch": 1.3737541528239203,
      "grad_norm": 10.801963806152344,
      "learning_rate": 4.722893949797724e-07,
      "loss": 0.023,
      "step": 827
    },
    {
      "epoch": 1.3754152823920265,
      "grad_norm": 3.6390554904937744,
      "learning_rate": 4.7000463902383825e-07,
      "loss": 0.0036,
      "step": 828
    },
    {
      "epoch": 1.3770764119601329,
      "grad_norm": 1.589859962463379,
      "learning_rate": 4.677237239399489e-07,
      "loss": 0.0026,
      "step": 829
    },
    {
      "epoch": 1.3787375415282392,
      "grad_norm": 8.354605674743652,
      "learning_rate": 4.654466662578781e-07,
      "loss": 0.0206,
      "step": 830
    },
    {
      "epoch": 1.3803986710963456,
      "grad_norm": 6.640425205230713,
      "learning_rate": 4.631734824794428e-07,
      "loss": 0.0141,
      "step": 831
    },
    {
      "epoch": 1.3820598006644518,
      "grad_norm": 13.431829452514648,
      "learning_rate": 4.6090418907838814e-07,
      "loss": 0.0374,
      "step": 832
    },
    {
      "epoch": 1.3837209302325582,
      "grad_norm": 12.409584999084473,
      "learning_rate": 4.586388025002647e-07,
      "loss": 0.0198,
      "step": 833
    },
    {
      "epoch": 1.3853820598006645,
      "grad_norm": 1.4373682737350464,
      "learning_rate": 4.5637733916231004e-07,
      "loss": 0.0023,
      "step": 834
    },
    {
      "epoch": 1.3870431893687707,
      "grad_norm": 12.105127334594727,
      "learning_rate": 4.541198154533311e-07,
      "loss": 0.0145,
      "step": 835
    },
    {
      "epoch": 1.388704318936877,
      "grad_norm": 1.378914475440979,
      "learning_rate": 4.518662477335834e-07,
      "loss": 0.0017,
      "step": 836
    },
    {
      "epoch": 1.3903654485049834,
      "grad_norm": 2.0379269123077393,
      "learning_rate": 4.4961665233465393e-07,
      "loss": 0.0026,
      "step": 837
    },
    {
      "epoch": 1.3920265780730898,
      "grad_norm": 4.553691864013672,
      "learning_rate": 4.473710455593416e-07,
      "loss": 0.011,
      "step": 838
    },
    {
      "epoch": 1.393687707641196,
      "grad_norm": 9.32184886932373,
      "learning_rate": 4.4512944368154114e-07,
      "loss": 0.0311,
      "step": 839
    },
    {
      "epoch": 1.3953488372093024,
      "grad_norm": 3.257469892501831,
      "learning_rate": 4.4289186294612256e-07,
      "loss": 0.0049,
      "step": 840
    },
    {
      "epoch": 1.3970099667774085,
      "grad_norm": 4.196468830108643,
      "learning_rate": 4.4065831956881494e-07,
      "loss": 0.0092,
      "step": 841
    },
    {
      "epoch": 1.398671096345515,
      "grad_norm": 2.6750576496124268,
      "learning_rate": 4.384288297360894e-07,
      "loss": 0.0065,
      "step": 842
    },
    {
      "epoch": 1.4003322259136213,
      "grad_norm": 0.7593079805374146,
      "learning_rate": 4.362034096050403e-07,
      "loss": 0.0007,
      "step": 843
    },
    {
      "epoch": 1.4019933554817277,
      "grad_norm": 0.7584846019744873,
      "learning_rate": 4.339820753032691e-07,
      "loss": 0.0016,
      "step": 844
    },
    {
      "epoch": 1.4036544850498338,
      "grad_norm": 24.843107223510742,
      "learning_rate": 4.31764842928767e-07,
      "loss": 0.0216,
      "step": 845
    },
    {
      "epoch": 1.4053156146179402,
      "grad_norm": 7.665549278259277,
      "learning_rate": 4.295517285497995e-07,
      "loss": 0.0099,
      "step": 846
    },
    {
      "epoch": 1.4069767441860466,
      "grad_norm": 16.56751251220703,
      "learning_rate": 4.27342748204788e-07,
      "loss": 0.0097,
      "step": 847
    },
    {
      "epoch": 1.4086378737541527,
      "grad_norm": 8.430388450622559,
      "learning_rate": 4.2513791790219467e-07,
      "loss": 0.0162,
      "step": 848
    },
    {
      "epoch": 1.410299003322259,
      "grad_norm": 6.689608097076416,
      "learning_rate": 4.229372536204074e-07,
      "loss": 0.018,
      "step": 849
    },
    {
      "epoch": 1.4119601328903655,
      "grad_norm": 13.74167537689209,
      "learning_rate": 4.207407713076221e-07,
      "loss": 0.0113,
      "step": 850
    },
    {
      "epoch": 1.4136212624584719,
      "grad_norm": 13.198443412780762,
      "learning_rate": 4.185484868817274e-07,
      "loss": 0.0628,
      "step": 851
    },
    {
      "epoch": 1.415282392026578,
      "grad_norm": 17.86145782470703,
      "learning_rate": 4.1636041623019135e-07,
      "loss": 0.0521,
      "step": 852
    },
    {
      "epoch": 1.4169435215946844,
      "grad_norm": 25.738597869873047,
      "learning_rate": 4.14176575209944e-07,
      "loss": 0.1025,
      "step": 853
    },
    {
      "epoch": 1.4186046511627908,
      "grad_norm": 0.3076265752315521,
      "learning_rate": 4.119969796472631e-07,
      "loss": 0.0003,
      "step": 854
    },
    {
      "epoch": 1.420265780730897,
      "grad_norm": 17.335969924926758,
      "learning_rate": 4.098216453376596e-07,
      "loss": 0.0617,
      "step": 855
    },
    {
      "epoch": 1.4219269102990033,
      "grad_norm": 6.2664899826049805,
      "learning_rate": 4.076505880457641e-07,
      "loss": 0.0253,
      "step": 856
    },
    {
      "epoch": 1.4235880398671097,
      "grad_norm": 5.2418060302734375,
      "learning_rate": 4.0548382350521084e-07,
      "loss": 0.0074,
      "step": 857
    },
    {
      "epoch": 1.425249169435216,
      "grad_norm": 3.4297940731048584,
      "learning_rate": 4.033213674185241e-07,
      "loss": 0.0053,
      "step": 858
    },
    {
      "epoch": 1.4269102990033222,
      "grad_norm": 9.643595695495605,
      "learning_rate": 4.011632354570067e-07,
      "loss": 0.076,
      "step": 859
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 4.146191596984863,
      "learning_rate": 3.990094432606218e-07,
      "loss": 0.0027,
      "step": 860
    },
    {
      "epoch": 1.4302325581395348,
      "grad_norm": 0.19965682923793793,
      "learning_rate": 3.96860006437885e-07,
      "loss": 0.0003,
      "step": 861
    },
    {
      "epoch": 1.4318936877076411,
      "grad_norm": 19.833518981933594,
      "learning_rate": 3.9471494056574685e-07,
      "loss": 0.0622,
      "step": 862
    },
    {
      "epoch": 1.4335548172757475,
      "grad_norm": 0.17571118474006653,
      "learning_rate": 3.92574261189482e-07,
      "loss": 0.0003,
      "step": 863
    },
    {
      "epoch": 1.435215946843854,
      "grad_norm": 7.566018581390381,
      "learning_rate": 3.9043798382257697e-07,
      "loss": 0.0078,
      "step": 864
    },
    {
      "epoch": 1.43687707641196,
      "grad_norm": 11.348194122314453,
      "learning_rate": 3.88306123946616e-07,
      "loss": 0.0235,
      "step": 865
    },
    {
      "epoch": 1.4385382059800664,
      "grad_norm": 10.6011381149292,
      "learning_rate": 3.861786970111702e-07,
      "loss": 0.0313,
      "step": 866
    },
    {
      "epoch": 1.4401993355481728,
      "grad_norm": 0.5557393431663513,
      "learning_rate": 3.840557184336849e-07,
      "loss": 0.0006,
      "step": 867
    },
    {
      "epoch": 1.441860465116279,
      "grad_norm": 25.956459045410156,
      "learning_rate": 3.81937203599369e-07,
      "loss": 0.0449,
      "step": 868
    },
    {
      "epoch": 1.4435215946843853,
      "grad_norm": 11.637957572937012,
      "learning_rate": 3.798231678610819e-07,
      "loss": 0.0237,
      "step": 869
    },
    {
      "epoch": 1.4451827242524917,
      "grad_norm": 5.680612087249756,
      "learning_rate": 3.777136265392231e-07,
      "loss": 0.0055,
      "step": 870
    },
    {
      "epoch": 1.446843853820598,
      "grad_norm": 4.692670822143555,
      "learning_rate": 3.7560859492162176e-07,
      "loss": 0.0038,
      "step": 871
    },
    {
      "epoch": 1.4485049833887043,
      "grad_norm": 7.798187732696533,
      "learning_rate": 3.735080882634246e-07,
      "loss": 0.0065,
      "step": 872
    },
    {
      "epoch": 1.4501661129568106,
      "grad_norm": 14.402698516845703,
      "learning_rate": 3.7141212178698576e-07,
      "loss": 0.0611,
      "step": 873
    },
    {
      "epoch": 1.451827242524917,
      "grad_norm": 14.016066551208496,
      "learning_rate": 3.69320710681758e-07,
      "loss": 0.0834,
      "step": 874
    },
    {
      "epoch": 1.4534883720930232,
      "grad_norm": 7.999434947967529,
      "learning_rate": 3.672338701041798e-07,
      "loss": 0.0183,
      "step": 875
    },
    {
      "epoch": 1.4551495016611296,
      "grad_norm": 10.387938499450684,
      "learning_rate": 3.651516151775681e-07,
      "loss": 0.0428,
      "step": 876
    },
    {
      "epoch": 1.456810631229236,
      "grad_norm": 15.999258041381836,
      "learning_rate": 3.630739609920068e-07,
      "loss": 0.0825,
      "step": 877
    },
    {
      "epoch": 1.4584717607973423,
      "grad_norm": 7.274929523468018,
      "learning_rate": 3.6100092260423955e-07,
      "loss": 0.0211,
      "step": 878
    },
    {
      "epoch": 1.4601328903654485,
      "grad_norm": 13.323637962341309,
      "learning_rate": 3.5893251503755815e-07,
      "loss": 0.0177,
      "step": 879
    },
    {
      "epoch": 1.4617940199335548,
      "grad_norm": 4.893885135650635,
      "learning_rate": 3.5686875328169507e-07,
      "loss": 0.0122,
      "step": 880
    },
    {
      "epoch": 1.463455149501661,
      "grad_norm": 5.68451452255249,
      "learning_rate": 3.548096522927154e-07,
      "loss": 0.0199,
      "step": 881
    },
    {
      "epoch": 1.4651162790697674,
      "grad_norm": 5.512393951416016,
      "learning_rate": 3.5275522699290704e-07,
      "loss": 0.0164,
      "step": 882
    },
    {
      "epoch": 1.4667774086378738,
      "grad_norm": 18.978796005249023,
      "learning_rate": 3.5070549227067324e-07,
      "loss": 0.0334,
      "step": 883
    },
    {
      "epoch": 1.4684385382059801,
      "grad_norm": 0.4384339153766632,
      "learning_rate": 3.4866046298042426e-07,
      "loss": 0.0008,
      "step": 884
    },
    {
      "epoch": 1.4700996677740865,
      "grad_norm": 3.6679792404174805,
      "learning_rate": 3.466201539424716e-07,
      "loss": 0.0035,
      "step": 885
    },
    {
      "epoch": 1.4717607973421927,
      "grad_norm": 8.465070724487305,
      "learning_rate": 3.4458457994291757e-07,
      "loss": 0.0178,
      "step": 886
    },
    {
      "epoch": 1.473421926910299,
      "grad_norm": 9.443462371826172,
      "learning_rate": 3.4255375573355014e-07,
      "loss": 0.0077,
      "step": 887
    },
    {
      "epoch": 1.4750830564784052,
      "grad_norm": 8.275284767150879,
      "learning_rate": 3.405276960317366e-07,
      "loss": 0.0075,
      "step": 888
    },
    {
      "epoch": 1.4767441860465116,
      "grad_norm": 4.26287317276001,
      "learning_rate": 3.385064155203138e-07,
      "loss": 0.0107,
      "step": 889
    },
    {
      "epoch": 1.478405315614618,
      "grad_norm": 17.598369598388672,
      "learning_rate": 3.364899288474857e-07,
      "loss": 0.0305,
      "step": 890
    },
    {
      "epoch": 1.4800664451827243,
      "grad_norm": 25.065013885498047,
      "learning_rate": 3.344782506267154e-07,
      "loss": 0.0558,
      "step": 891
    },
    {
      "epoch": 1.4817275747508305,
      "grad_norm": 8.171377182006836,
      "learning_rate": 3.3247139543661705e-07,
      "loss": 0.0585,
      "step": 892
    },
    {
      "epoch": 1.4833887043189369,
      "grad_norm": 13.189953804016113,
      "learning_rate": 3.3046937782085506e-07,
      "loss": 0.036,
      "step": 893
    },
    {
      "epoch": 1.4850498338870433,
      "grad_norm": 6.294521331787109,
      "learning_rate": 3.2847221228803414e-07,
      "loss": 0.0275,
      "step": 894
    },
    {
      "epoch": 1.4867109634551494,
      "grad_norm": 2.937936305999756,
      "learning_rate": 3.264799133115963e-07,
      "loss": 0.003,
      "step": 895
    },
    {
      "epoch": 1.4883720930232558,
      "grad_norm": 8.645936012268066,
      "learning_rate": 3.244924953297167e-07,
      "loss": 0.0213,
      "step": 896
    },
    {
      "epoch": 1.4900332225913622,
      "grad_norm": 19.482379913330078,
      "learning_rate": 3.2250997274519666e-07,
      "loss": 0.0985,
      "step": 897
    },
    {
      "epoch": 1.4916943521594686,
      "grad_norm": 2.1434147357940674,
      "learning_rate": 3.205323599253613e-07,
      "loss": 0.0075,
      "step": 898
    },
    {
      "epoch": 1.4933554817275747,
      "grad_norm": 1.688423991203308,
      "learning_rate": 3.1855967120195413e-07,
      "loss": 0.0037,
      "step": 899
    },
    {
      "epoch": 1.495016611295681,
      "grad_norm": 4.679425239562988,
      "learning_rate": 3.165919208710348e-07,
      "loss": 0.0176,
      "step": 900
    },
    {
      "epoch": 1.4966777408637872,
      "grad_norm": 25.044464111328125,
      "learning_rate": 3.146291231928737e-07,
      "loss": 0.0764,
      "step": 901
    },
    {
      "epoch": 1.4983388704318936,
      "grad_norm": 4.9685893058776855,
      "learning_rate": 3.1267129239184907e-07,
      "loss": 0.0123,
      "step": 902
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.012315273284912,
      "learning_rate": 3.107184426563455e-07,
      "loss": 0.0124,
      "step": 903
    },
    {
      "epoch": 1.5016611295681064,
      "grad_norm": 13.312503814697266,
      "learning_rate": 3.0877058813864854e-07,
      "loss": 0.0925,
      "step": 904
    },
    {
      "epoch": 1.5033222591362128,
      "grad_norm": 14.907830238342285,
      "learning_rate": 3.0682774295484406e-07,
      "loss": 0.0279,
      "step": 905
    },
    {
      "epoch": 1.504983388704319,
      "grad_norm": 3.517716646194458,
      "learning_rate": 3.048899211847148e-07,
      "loss": 0.0094,
      "step": 906
    },
    {
      "epoch": 1.5066445182724253,
      "grad_norm": 5.761756896972656,
      "learning_rate": 3.0295713687164004e-07,
      "loss": 0.0078,
      "step": 907
    },
    {
      "epoch": 1.5083056478405314,
      "grad_norm": 8.124967575073242,
      "learning_rate": 3.010294040224914e-07,
      "loss": 0.0134,
      "step": 908
    },
    {
      "epoch": 1.5099667774086378,
      "grad_norm": 14.692148208618164,
      "learning_rate": 2.9910673660753295e-07,
      "loss": 0.0122,
      "step": 909
    },
    {
      "epoch": 1.5116279069767442,
      "grad_norm": 3.8068079948425293,
      "learning_rate": 2.971891485603203e-07,
      "loss": 0.0062,
      "step": 910
    },
    {
      "epoch": 1.5132890365448506,
      "grad_norm": 7.373571395874023,
      "learning_rate": 2.952766537775978e-07,
      "loss": 0.0153,
      "step": 911
    },
    {
      "epoch": 1.514950166112957,
      "grad_norm": 6.404834270477295,
      "learning_rate": 2.9336926611919953e-07,
      "loss": 0.0157,
      "step": 912
    },
    {
      "epoch": 1.5166112956810631,
      "grad_norm": 19.058998107910156,
      "learning_rate": 2.914669994079486e-07,
      "loss": 0.0377,
      "step": 913
    },
    {
      "epoch": 1.5182724252491693,
      "grad_norm": 1.187256097793579,
      "learning_rate": 2.8956986742955604e-07,
      "loss": 0.0013,
      "step": 914
    },
    {
      "epoch": 1.5199335548172757,
      "grad_norm": 19.175989151000977,
      "learning_rate": 2.876778839325219e-07,
      "loss": 0.0542,
      "step": 915
    },
    {
      "epoch": 1.521594684385382,
      "grad_norm": 12.16926097869873,
      "learning_rate": 2.8579106262803465e-07,
      "loss": 0.0622,
      "step": 916
    },
    {
      "epoch": 1.5232558139534884,
      "grad_norm": 10.201384544372559,
      "learning_rate": 2.839094171898736e-07,
      "loss": 0.0154,
      "step": 917
    },
    {
      "epoch": 1.5249169435215948,
      "grad_norm": 17.753768920898438,
      "learning_rate": 2.8203296125430744e-07,
      "loss": 0.0435,
      "step": 918
    },
    {
      "epoch": 1.526578073089701,
      "grad_norm": 9.888166427612305,
      "learning_rate": 2.801617084199966e-07,
      "loss": 0.0331,
      "step": 919
    },
    {
      "epoch": 1.5282392026578073,
      "grad_norm": 1.097863793373108,
      "learning_rate": 2.782956722478962e-07,
      "loss": 0.001,
      "step": 920
    },
    {
      "epoch": 1.5299003322259135,
      "grad_norm": 1.8991848230361938,
      "learning_rate": 2.764348662611538e-07,
      "loss": 0.003,
      "step": 921
    },
    {
      "epoch": 1.5315614617940199,
      "grad_norm": 20.583703994750977,
      "learning_rate": 2.7457930394501563e-07,
      "loss": 0.0439,
      "step": 922
    },
    {
      "epoch": 1.5332225913621262,
      "grad_norm": 6.2648539543151855,
      "learning_rate": 2.727289987467274e-07,
      "loss": 0.0226,
      "step": 923
    },
    {
      "epoch": 1.5348837209302326,
      "grad_norm": 1.496519923210144,
      "learning_rate": 2.7088396407543435e-07,
      "loss": 0.0018,
      "step": 924
    },
    {
      "epoch": 1.536544850498339,
      "grad_norm": 6.547482013702393,
      "learning_rate": 2.690442133020888e-07,
      "loss": 0.0212,
      "step": 925
    },
    {
      "epoch": 1.5382059800664452,
      "grad_norm": 17.5012264251709,
      "learning_rate": 2.67209759759349e-07,
      "loss": 0.0462,
      "step": 926
    },
    {
      "epoch": 1.5398671096345515,
      "grad_norm": 0.6141988635063171,
      "learning_rate": 2.6538061674148504e-07,
      "loss": 0.0013,
      "step": 927
    },
    {
      "epoch": 1.5415282392026577,
      "grad_norm": 1.5886684656143188,
      "learning_rate": 2.6355679750428084e-07,
      "loss": 0.0029,
      "step": 928
    },
    {
      "epoch": 1.543189368770764,
      "grad_norm": 27.549413681030273,
      "learning_rate": 2.617383152649403e-07,
      "loss": 0.0891,
      "step": 929
    },
    {
      "epoch": 1.5448504983388704,
      "grad_norm": 10.47612476348877,
      "learning_rate": 2.5992518320198874e-07,
      "loss": 0.0201,
      "step": 930
    },
    {
      "epoch": 1.5465116279069768,
      "grad_norm": 10.489360809326172,
      "learning_rate": 2.5811741445517944e-07,
      "loss": 0.0269,
      "step": 931
    },
    {
      "epoch": 1.5481727574750832,
      "grad_norm": 3.5623435974121094,
      "learning_rate": 2.5631502212539815e-07,
      "loss": 0.0069,
      "step": 932
    },
    {
      "epoch": 1.5498338870431894,
      "grad_norm": 6.290643215179443,
      "learning_rate": 2.5451801927456694e-07,
      "loss": 0.0272,
      "step": 933
    },
    {
      "epoch": 1.5514950166112955,
      "grad_norm": 11.861459732055664,
      "learning_rate": 2.5272641892555067e-07,
      "loss": 0.0365,
      "step": 934
    },
    {
      "epoch": 1.553156146179402,
      "grad_norm": 1.0134330987930298,
      "learning_rate": 2.509402340620629e-07,
      "loss": 0.0022,
      "step": 935
    },
    {
      "epoch": 1.5548172757475083,
      "grad_norm": 2.671765089035034,
      "learning_rate": 2.491594776285704e-07,
      "loss": 0.0024,
      "step": 936
    },
    {
      "epoch": 1.5564784053156147,
      "grad_norm": 4.3023223876953125,
      "learning_rate": 2.473841625302006e-07,
      "loss": 0.0068,
      "step": 937
    },
    {
      "epoch": 1.558139534883721,
      "grad_norm": 3.784409999847412,
      "learning_rate": 2.45614301632647e-07,
      "loss": 0.0088,
      "step": 938
    },
    {
      "epoch": 1.5598006644518272,
      "grad_norm": 11.982508659362793,
      "learning_rate": 2.438499077620777e-07,
      "loss": 0.0185,
      "step": 939
    },
    {
      "epoch": 1.5614617940199336,
      "grad_norm": 6.5332818031311035,
      "learning_rate": 2.420909937050405e-07,
      "loss": 0.0146,
      "step": 940
    },
    {
      "epoch": 1.5631229235880397,
      "grad_norm": 14.1775484085083,
      "learning_rate": 2.403375722083707e-07,
      "loss": 0.0693,
      "step": 941
    },
    {
      "epoch": 1.564784053156146,
      "grad_norm": 8.041645050048828,
      "learning_rate": 2.385896559791002e-07,
      "loss": 0.0071,
      "step": 942
    },
    {
      "epoch": 1.5664451827242525,
      "grad_norm": 1.4674506187438965,
      "learning_rate": 2.3684725768436332e-07,
      "loss": 0.0026,
      "step": 943
    },
    {
      "epoch": 1.5681063122923589,
      "grad_norm": 10.562248229980469,
      "learning_rate": 2.3511038995130628e-07,
      "loss": 0.0332,
      "step": 944
    },
    {
      "epoch": 1.5697674418604652,
      "grad_norm": 7.5686187744140625,
      "learning_rate": 2.3337906536699504e-07,
      "loss": 0.0138,
      "step": 945
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 6.138749122619629,
      "learning_rate": 2.316532964783252e-07,
      "loss": 0.0173,
      "step": 946
    },
    {
      "epoch": 1.5730897009966778,
      "grad_norm": 9.5950927734375,
      "learning_rate": 2.2993309579192964e-07,
      "loss": 0.0328,
      "step": 947
    },
    {
      "epoch": 1.574750830564784,
      "grad_norm": 8.108460426330566,
      "learning_rate": 2.2821847577408815e-07,
      "loss": 0.041,
      "step": 948
    },
    {
      "epoch": 1.5764119601328903,
      "grad_norm": 43.922672271728516,
      "learning_rate": 2.2650944885063894e-07,
      "loss": 0.0735,
      "step": 949
    },
    {
      "epoch": 1.5780730897009967,
      "grad_norm": 2.7781217098236084,
      "learning_rate": 2.2480602740688514e-07,
      "loss": 0.0084,
      "step": 950
    }
  ],
  "logging_steps": 1,
  "max_steps": 1204,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0459192604691005e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}