{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999804522056428,
  "eval_steps": 500,
  "global_step": 1918,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005212745161920897,
      "grad_norm": 5.11075803301647,
      "learning_rate": 5.208333333333333e-08,
      "loss": 0.4124,
      "step": 1
    },
    {
      "epoch": 0.0010425490323841794,
      "grad_norm": 5.17436422800092,
      "learning_rate": 1.0416666666666667e-07,
      "loss": 0.4261,
      "step": 2
    },
    {
      "epoch": 0.001563823548576269,
      "grad_norm": 5.345440991325531,
      "learning_rate": 1.5625e-07,
      "loss": 0.4092,
      "step": 3
    },
    {
      "epoch": 0.0020850980647683587,
      "grad_norm": 4.900834347782274,
      "learning_rate": 2.0833333333333333e-07,
      "loss": 0.3987,
      "step": 4
    },
    {
      "epoch": 0.002606372580960448,
      "grad_norm": 5.075426426426963,
      "learning_rate": 2.604166666666667e-07,
      "loss": 0.4086,
      "step": 5
    },
    {
      "epoch": 0.003127647097152538,
      "grad_norm": 4.851137328642362,
      "learning_rate": 3.125e-07,
      "loss": 0.3814,
      "step": 6
    },
    {
      "epoch": 0.0036489216133446275,
      "grad_norm": 4.849044652089594,
      "learning_rate": 3.6458333333333337e-07,
      "loss": 0.3795,
      "step": 7
    },
    {
      "epoch": 0.0041701961295367174,
      "grad_norm": 4.5784045386773755,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 0.4008,
      "step": 8
    },
    {
      "epoch": 0.004691470645728807,
      "grad_norm": 4.7682806495033745,
      "learning_rate": 4.6875000000000006e-07,
      "loss": 0.4106,
      "step": 9
    },
    {
      "epoch": 0.005212745161920896,
      "grad_norm": 4.328666361858364,
      "learning_rate": 5.208333333333334e-07,
      "loss": 0.3986,
      "step": 10
    },
    {
      "epoch": 0.005734019678112986,
      "grad_norm": 4.283851136233341,
      "learning_rate": 5.729166666666667e-07,
      "loss": 0.4031,
      "step": 11
    },
    {
      "epoch": 0.006255294194305076,
      "grad_norm": 4.131111662127315,
      "learning_rate": 6.25e-07,
      "loss": 0.3776,
      "step": 12
    },
    {
      "epoch": 0.006776568710497166,
      "grad_norm": 3.759641197331778,
      "learning_rate": 6.770833333333333e-07,
      "loss": 0.3865,
      "step": 13
    },
    {
      "epoch": 0.007297843226689255,
      "grad_norm": 3.451804643766274,
      "learning_rate": 7.291666666666667e-07,
      "loss": 0.3748,
      "step": 14
    },
    {
      "epoch": 0.007819117742881345,
      "grad_norm": 3.135418097302639,
      "learning_rate": 7.8125e-07,
      "loss": 0.3665,
      "step": 15
    },
    {
      "epoch": 0.008340392259073435,
      "grad_norm": 2.7047102856884093,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.3597,
      "step": 16
    },
    {
      "epoch": 0.008861666775265524,
      "grad_norm": 2.772951802260474,
      "learning_rate": 8.854166666666668e-07,
      "loss": 0.3735,
      "step": 17
    },
    {
      "epoch": 0.009382941291457614,
      "grad_norm": 2.3198899618016138,
      "learning_rate": 9.375000000000001e-07,
      "loss": 0.3503,
      "step": 18
    },
    {
      "epoch": 0.009904215807649703,
      "grad_norm": 2.2447520592031704,
      "learning_rate": 9.895833333333333e-07,
      "loss": 0.3331,
      "step": 19
    },
    {
      "epoch": 0.010425490323841793,
      "grad_norm": 2.400538942817267,
      "learning_rate": 1.0416666666666667e-06,
      "loss": 0.3194,
      "step": 20
    },
    {
      "epoch": 0.010946764840033882,
      "grad_norm": 2.4839418106829743,
      "learning_rate": 1.0937500000000001e-06,
      "loss": 0.3059,
      "step": 21
    },
    {
      "epoch": 0.011468039356225972,
      "grad_norm": 2.072907404146958,
      "learning_rate": 1.1458333333333333e-06,
      "loss": 0.3251,
      "step": 22
    },
    {
      "epoch": 0.011989313872418063,
      "grad_norm": 1.9341789006266392,
      "learning_rate": 1.197916666666667e-06,
      "loss": 0.3095,
      "step": 23
    },
    {
      "epoch": 0.012510588388610152,
      "grad_norm": 1.7307141646883002,
      "learning_rate": 1.25e-06,
      "loss": 0.3193,
      "step": 24
    },
    {
      "epoch": 0.013031862904802242,
      "grad_norm": 1.7615841916776682,
      "learning_rate": 1.3020833333333335e-06,
      "loss": 0.3319,
      "step": 25
    },
    {
      "epoch": 0.013553137420994331,
      "grad_norm": 1.7136648054606347,
      "learning_rate": 1.3541666666666667e-06,
      "loss": 0.2944,
      "step": 26
    },
    {
      "epoch": 0.01407441193718642,
      "grad_norm": 1.9987205714740135,
      "learning_rate": 1.40625e-06,
      "loss": 0.3041,
      "step": 27
    },
    {
      "epoch": 0.01459568645337851,
      "grad_norm": 2.012277299697316,
      "learning_rate": 1.4583333333333335e-06,
      "loss": 0.3063,
      "step": 28
    },
    {
      "epoch": 0.0151169609695706,
      "grad_norm": 2.0782151715380746,
      "learning_rate": 1.5104166666666667e-06,
      "loss": 0.3091,
      "step": 29
    },
    {
      "epoch": 0.01563823548576269,
      "grad_norm": 1.7474712574781637,
      "learning_rate": 1.5625e-06,
      "loss": 0.2879,
      "step": 30
    },
    {
      "epoch": 0.01615951000195478,
      "grad_norm": 1.8944893639350486,
      "learning_rate": 1.6145833333333335e-06,
      "loss": 0.3016,
      "step": 31
    },
    {
      "epoch": 0.01668078451814687,
      "grad_norm": 1.6876038268436602,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.3042,
      "step": 32
    },
    {
      "epoch": 0.017202059034338957,
      "grad_norm": 1.861475840222954,
      "learning_rate": 1.71875e-06,
      "loss": 0.3048,
      "step": 33
    },
    {
      "epoch": 0.01772333355053105,
      "grad_norm": 1.6018157378320321,
      "learning_rate": 1.7708333333333337e-06,
      "loss": 0.3043,
      "step": 34
    },
    {
      "epoch": 0.018244608066723136,
      "grad_norm": 1.7354154053640138,
      "learning_rate": 1.8229166666666666e-06,
      "loss": 0.2924,
      "step": 35
    },
    {
      "epoch": 0.018765882582915228,
      "grad_norm": 1.4987710469506477,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 0.2771,
      "step": 36
    },
    {
      "epoch": 0.01928715709910732,
      "grad_norm": 1.567127836898921,
      "learning_rate": 1.9270833333333334e-06,
      "loss": 0.2838,
      "step": 37
    },
    {
      "epoch": 0.019808431615299407,
      "grad_norm": 1.5064764171139846,
      "learning_rate": 1.9791666666666666e-06,
      "loss": 0.2941,
      "step": 38
    },
    {
      "epoch": 0.020329706131491498,
      "grad_norm": 1.340604325707909,
      "learning_rate": 2.0312500000000002e-06,
      "loss": 0.2918,
      "step": 39
    },
    {
      "epoch": 0.020850980647683585,
      "grad_norm": 1.3211877381742252,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 0.2762,
      "step": 40
    },
    {
      "epoch": 0.021372255163875677,
      "grad_norm": 1.4655683382115356,
      "learning_rate": 2.1354166666666666e-06,
      "loss": 0.2657,
      "step": 41
    },
    {
      "epoch": 0.021893529680067764,
      "grad_norm": 1.3423365858196985,
      "learning_rate": 2.1875000000000002e-06,
      "loss": 0.2897,
      "step": 42
    },
    {
      "epoch": 0.022414804196259856,
      "grad_norm": 1.3364845050339877,
      "learning_rate": 2.2395833333333334e-06,
      "loss": 0.2821,
      "step": 43
    },
    {
      "epoch": 0.022936078712451943,
      "grad_norm": 1.3593339435906422,
      "learning_rate": 2.2916666666666666e-06,
      "loss": 0.2599,
      "step": 44
    },
    {
      "epoch": 0.023457353228644034,
      "grad_norm": 1.3475417862019787,
      "learning_rate": 2.3437500000000002e-06,
      "loss": 0.2745,
      "step": 45
    },
    {
      "epoch": 0.023978627744836126,
      "grad_norm": 1.342587329740068,
      "learning_rate": 2.395833333333334e-06,
      "loss": 0.2626,
      "step": 46
    },
    {
      "epoch": 0.024499902261028213,
      "grad_norm": 1.2789760759653985,
      "learning_rate": 2.4479166666666666e-06,
      "loss": 0.2699,
      "step": 47
    },
    {
      "epoch": 0.025021176777220305,
      "grad_norm": 1.3631288905885437,
      "learning_rate": 2.5e-06,
      "loss": 0.2764,
      "step": 48
    },
    {
      "epoch": 0.025542451293412392,
      "grad_norm": 1.3230391864388509,
      "learning_rate": 2.5520833333333334e-06,
      "loss": 0.2774,
      "step": 49
    },
    {
      "epoch": 0.026063725809604484,
      "grad_norm": 1.21271476234125,
      "learning_rate": 2.604166666666667e-06,
      "loss": 0.2763,
      "step": 50
    },
    {
      "epoch": 0.02658500032579657,
      "grad_norm": 1.1932303504862625,
      "learning_rate": 2.65625e-06,
      "loss": 0.2586,
      "step": 51
    },
    {
      "epoch": 0.027106274841988662,
      "grad_norm": 1.2820424405047717,
      "learning_rate": 2.7083333333333334e-06,
      "loss": 0.273,
      "step": 52
    },
    {
      "epoch": 0.027627549358180754,
      "grad_norm": 1.2531931817054822,
      "learning_rate": 2.760416666666667e-06,
      "loss": 0.2633,
      "step": 53
    },
    {
      "epoch": 0.02814882387437284,
      "grad_norm": 1.4100197823383918,
      "learning_rate": 2.8125e-06,
      "loss": 0.2713,
      "step": 54
    },
    {
      "epoch": 0.028670098390564933,
      "grad_norm": 1.244230363037586,
      "learning_rate": 2.8645833333333334e-06,
      "loss": 0.2675,
      "step": 55
    },
    {
      "epoch": 0.02919137290675702,
      "grad_norm": 1.233721117946254,
      "learning_rate": 2.916666666666667e-06,
      "loss": 0.26,
      "step": 56
    },
    {
      "epoch": 0.02971264742294911,
      "grad_norm": 1.33074120780009,
      "learning_rate": 2.96875e-06,
      "loss": 0.2623,
      "step": 57
    },
    {
      "epoch": 0.0302339219391412,
      "grad_norm": 1.3279194320065162,
      "learning_rate": 3.0208333333333334e-06,
      "loss": 0.2709,
      "step": 58
    },
    {
      "epoch": 0.03075519645533329,
      "grad_norm": 1.4390908943253897,
      "learning_rate": 3.072916666666667e-06,
      "loss": 0.2519,
      "step": 59
    },
    {
      "epoch": 0.03127647097152538,
      "grad_norm": 1.4426086827285376,
      "learning_rate": 3.125e-06,
      "loss": 0.2683,
      "step": 60
    },
    {
      "epoch": 0.031797745487717466,
      "grad_norm": 1.2969321386639103,
      "learning_rate": 3.1770833333333333e-06,
      "loss": 0.2486,
      "step": 61
    },
    {
      "epoch": 0.03231902000390956,
      "grad_norm": 1.303884366809117,
      "learning_rate": 3.229166666666667e-06,
      "loss": 0.2589,
      "step": 62
    },
    {
      "epoch": 0.03284029452010165,
      "grad_norm": 1.4012507792386482,
      "learning_rate": 3.28125e-06,
      "loss": 0.2583,
      "step": 63
    },
    {
      "epoch": 0.03336156903629374,
      "grad_norm": 1.3811445006822614,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.2596,
      "step": 64
    },
    {
      "epoch": 0.03388284355248583,
      "grad_norm": 1.4475370235278413,
      "learning_rate": 3.385416666666667e-06,
      "loss": 0.2605,
      "step": 65
    },
    {
      "epoch": 0.034404118068677915,
      "grad_norm": 1.322688630617171,
      "learning_rate": 3.4375e-06,
      "loss": 0.2477,
      "step": 66
    },
    {
      "epoch": 0.034925392584870006,
      "grad_norm": 1.3992892078555939,
      "learning_rate": 3.4895833333333333e-06,
      "loss": 0.2446,
      "step": 67
    },
    {
      "epoch": 0.0354466671010621,
      "grad_norm": 1.2681725892431805,
      "learning_rate": 3.5416666666666673e-06,
      "loss": 0.2478,
      "step": 68
    },
    {
      "epoch": 0.03596794161725419,
      "grad_norm": 1.4436847621832125,
      "learning_rate": 3.59375e-06,
      "loss": 0.2619,
      "step": 69
    },
    {
      "epoch": 0.03648921613344627,
      "grad_norm": 1.4010222407978898,
      "learning_rate": 3.6458333333333333e-06,
      "loss": 0.2458,
      "step": 70
    },
    {
      "epoch": 0.037010490649638364,
      "grad_norm": 1.1847301699214483,
      "learning_rate": 3.6979166666666673e-06,
      "loss": 0.2457,
      "step": 71
    },
    {
      "epoch": 0.037531765165830455,
      "grad_norm": 1.2656606340585645,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.2496,
      "step": 72
    },
    {
      "epoch": 0.038053039682022546,
      "grad_norm": 1.4606373768036418,
      "learning_rate": 3.8020833333333333e-06,
      "loss": 0.2512,
      "step": 73
    },
    {
      "epoch": 0.03857431419821464,
      "grad_norm": 1.2166433050839274,
      "learning_rate": 3.854166666666667e-06,
      "loss": 0.2219,
      "step": 74
    },
    {
      "epoch": 0.03909558871440672,
      "grad_norm": 1.3292968759151849,
      "learning_rate": 3.90625e-06,
      "loss": 0.2579,
      "step": 75
    },
    {
      "epoch": 0.03961686323059881,
      "grad_norm": 1.4951315237751113,
      "learning_rate": 3.958333333333333e-06,
      "loss": 0.2447,
      "step": 76
    },
    {
      "epoch": 0.040138137746790904,
      "grad_norm": 1.3489871277817018,
      "learning_rate": 4.010416666666667e-06,
      "loss": 0.2575,
      "step": 77
    },
    {
      "epoch": 0.040659412262982995,
      "grad_norm": 1.3802154019982127,
      "learning_rate": 4.0625000000000005e-06,
      "loss": 0.2524,
      "step": 78
    },
    {
      "epoch": 0.04118068677917508,
      "grad_norm": 1.2213764719182099,
      "learning_rate": 4.114583333333334e-06,
      "loss": 0.2534,
      "step": 79
    },
    {
      "epoch": 0.04170196129536717,
      "grad_norm": 1.2310142856144946,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.2434,
      "step": 80
    },
    {
      "epoch": 0.04222323581155926,
      "grad_norm": 1.2489940345168304,
      "learning_rate": 4.21875e-06,
      "loss": 0.2224,
      "step": 81
    },
    {
      "epoch": 0.04274451032775135,
      "grad_norm": 1.5335069492862015,
      "learning_rate": 4.270833333333333e-06,
      "loss": 0.2646,
      "step": 82
    },
    {
      "epoch": 0.043265784843943444,
      "grad_norm": 1.448069380566151,
      "learning_rate": 4.322916666666667e-06,
      "loss": 0.2613,
      "step": 83
    },
    {
      "epoch": 0.04378705936013553,
      "grad_norm": 1.6356379211230665,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 0.2616,
      "step": 84
    },
    {
      "epoch": 0.04430833387632762,
      "grad_norm": 1.7155523706338194,
      "learning_rate": 4.427083333333334e-06,
      "loss": 0.241,
      "step": 85
    },
    {
      "epoch": 0.04482960839251971,
      "grad_norm": 1.3286348110703599,
      "learning_rate": 4.479166666666667e-06,
      "loss": 0.2664,
      "step": 86
    },
    {
      "epoch": 0.0453508829087118,
      "grad_norm": 1.4003238573470502,
      "learning_rate": 4.53125e-06,
      "loss": 0.2364,
      "step": 87
    },
    {
      "epoch": 0.04587215742490389,
      "grad_norm": 1.3658849183921546,
      "learning_rate": 4.583333333333333e-06,
      "loss": 0.235,
      "step": 88
    },
    {
      "epoch": 0.04639343194109598,
      "grad_norm": 1.3866176657796205,
      "learning_rate": 4.635416666666667e-06,
      "loss": 0.2502,
      "step": 89
    },
    {
      "epoch": 0.04691470645728807,
      "grad_norm": 1.4329731217987849,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.2591,
      "step": 90
    },
    {
      "epoch": 0.04743598097348016,
      "grad_norm": 1.258561927924062,
      "learning_rate": 4.739583333333334e-06,
      "loss": 0.2503,
      "step": 91
    },
    {
      "epoch": 0.04795725548967225,
      "grad_norm": 1.3076546779001206,
      "learning_rate": 4.791666666666668e-06,
      "loss": 0.2345,
      "step": 92
    },
    {
      "epoch": 0.048478530005864336,
      "grad_norm": 1.2960823334662572,
      "learning_rate": 4.84375e-06,
      "loss": 0.2258,
      "step": 93
    },
    {
      "epoch": 0.04899980452205643,
      "grad_norm": 1.210681753840828,
      "learning_rate": 4.895833333333333e-06,
      "loss": 0.2469,
      "step": 94
    },
    {
      "epoch": 0.04952107903824852,
      "grad_norm": 1.3727031474669664,
      "learning_rate": 4.947916666666667e-06,
      "loss": 0.2595,
      "step": 95
    },
    {
      "epoch": 0.05004235355444061,
      "grad_norm": 1.385506278655309,
      "learning_rate": 5e-06,
      "loss": 0.2357,
      "step": 96
    },
    {
      "epoch": 0.050563628070632693,
      "grad_norm": 1.3815222257014061,
      "learning_rate": 4.999996283681687e-06,
      "loss": 0.2617,
      "step": 97
    },
    {
      "epoch": 0.051084902586824785,
      "grad_norm": 1.327004917894176,
      "learning_rate": 4.9999851347377946e-06,
      "loss": 0.2476,
      "step": 98
    },
    {
      "epoch": 0.051606177103016876,
      "grad_norm": 1.3696537124053607,
      "learning_rate": 4.99996655320147e-06,
      "loss": 0.2392,
      "step": 99
    },
    {
      "epoch": 0.05212745161920897,
      "grad_norm": 1.3657512033991304,
      "learning_rate": 4.999940539127958e-06,
      "loss": 0.2608,
      "step": 100
    },
    {
      "epoch": 0.05264872613540106,
      "grad_norm": 1.4194965343026547,
      "learning_rate": 4.999907092594598e-06,
      "loss": 0.2442,
      "step": 101
    },
    {
      "epoch": 0.05317000065159314,
      "grad_norm": 1.2749089132199982,
      "learning_rate": 4.99986621370083e-06,
      "loss": 0.2226,
      "step": 102
    },
    {
      "epoch": 0.053691275167785234,
      "grad_norm": 1.1887113141245471,
      "learning_rate": 4.99981790256819e-06,
      "loss": 0.2432,
      "step": 103
    },
    {
      "epoch": 0.054212549683977325,
      "grad_norm": 1.1938219400507954,
      "learning_rate": 4.999762159340305e-06,
      "loss": 0.2283,
      "step": 104
    },
    {
      "epoch": 0.054733824200169416,
      "grad_norm": 1.2840348614013515,
      "learning_rate": 4.999698984182909e-06,
      "loss": 0.218,
      "step": 105
    },
    {
      "epoch": 0.05525509871636151,
      "grad_norm": 1.2803746740205806,
      "learning_rate": 4.999628377283821e-06,
      "loss": 0.2508,
      "step": 106
    },
    {
      "epoch": 0.05577637323255359,
      "grad_norm": 1.4168047008355111,
      "learning_rate": 4.99955033885296e-06,
      "loss": 0.227,
      "step": 107
    },
    {
      "epoch": 0.05629764774874568,
      "grad_norm": 1.3985330741489335,
      "learning_rate": 4.999464869122339e-06,
      "loss": 0.2654,
      "step": 108
    },
    {
      "epoch": 0.056818922264937774,
      "grad_norm": 1.3276357405175772,
      "learning_rate": 4.999371968346064e-06,
      "loss": 0.2559,
      "step": 109
    },
    {
      "epoch": 0.057340196781129865,
      "grad_norm": 1.3466763643818087,
      "learning_rate": 4.999271636800334e-06,
      "loss": 0.2453,
      "step": 110
    },
    {
      "epoch": 0.05786147129732195,
      "grad_norm": 1.306537355090405,
      "learning_rate": 4.999163874783441e-06,
      "loss": 0.2546,
      "step": 111
    },
    {
      "epoch": 0.05838274581351404,
      "grad_norm": 1.3217941323578062,
      "learning_rate": 4.999048682615766e-06,
      "loss": 0.2555,
      "step": 112
    },
    {
      "epoch": 0.05890402032970613,
      "grad_norm": 1.2438940840815147,
      "learning_rate": 4.9989260606397816e-06,
      "loss": 0.2449,
      "step": 113
    },
    {
      "epoch": 0.05942529484589822,
      "grad_norm": 1.4326367838877314,
      "learning_rate": 4.998796009220051e-06,
      "loss": 0.255,
      "step": 114
    },
    {
      "epoch": 0.059946569362090314,
      "grad_norm": 1.269352153923466,
      "learning_rate": 4.9986585287432236e-06,
      "loss": 0.2372,
      "step": 115
    },
    {
      "epoch": 0.0604678438782824,
      "grad_norm": 1.3071274325044602,
      "learning_rate": 4.998513619618036e-06,
      "loss": 0.2442,
      "step": 116
    },
    {
      "epoch": 0.06098911839447449,
      "grad_norm": 1.3637289863199187,
      "learning_rate": 4.998361282275311e-06,
      "loss": 0.2281,
      "step": 117
    },
    {
      "epoch": 0.06151039291066658,
      "grad_norm": 1.4354154013144573,
      "learning_rate": 4.998201517167956e-06,
      "loss": 0.2554,
      "step": 118
    },
    {
      "epoch": 0.06203166742685867,
      "grad_norm": 1.2686877262427592,
      "learning_rate": 4.998034324770962e-06,
      "loss": 0.2334,
      "step": 119
    },
    {
      "epoch": 0.06255294194305076,
      "grad_norm": 1.4652927159239066,
      "learning_rate": 4.997859705581399e-06,
      "loss": 0.2407,
      "step": 120
    },
    {
      "epoch": 0.06307421645924285,
      "grad_norm": 1.382408681054862,
      "learning_rate": 4.997677660118423e-06,
      "loss": 0.23,
      "step": 121
    },
    {
      "epoch": 0.06359549097543493,
      "grad_norm": 1.3083101827135726,
      "learning_rate": 4.997488188923262e-06,
      "loss": 0.2369,
      "step": 122
    },
    {
      "epoch": 0.06411676549162702,
      "grad_norm": 1.2680968491770188,
      "learning_rate": 4.9972912925592245e-06,
      "loss": 0.2322,
      "step": 123
    },
    {
      "epoch": 0.06463804000781911,
      "grad_norm": 1.3728476777387406,
      "learning_rate": 4.997086971611696e-06,
      "loss": 0.2254,
      "step": 124
    },
    {
      "epoch": 0.0651593145240112,
      "grad_norm": 1.2130340794808803,
      "learning_rate": 4.996875226688133e-06,
      "loss": 0.2322,
      "step": 125
    },
    {
      "epoch": 0.0656805890402033,
      "grad_norm": 1.1725670998162612,
      "learning_rate": 4.996656058418064e-06,
      "loss": 0.2511,
      "step": 126
    },
    {
      "epoch": 0.06620186355639539,
      "grad_norm": 1.2975213169142519,
      "learning_rate": 4.996429467453088e-06,
      "loss": 0.2359,
      "step": 127
    },
    {
      "epoch": 0.06672313807258748,
      "grad_norm": 1.2983892428971382,
      "learning_rate": 4.996195454466873e-06,
      "loss": 0.2176,
      "step": 128
    },
    {
      "epoch": 0.06724441258877957,
      "grad_norm": 1.3409928375397802,
      "learning_rate": 4.995954020155153e-06,
      "loss": 0.2409,
      "step": 129
    },
    {
      "epoch": 0.06776568710497166,
      "grad_norm": 1.2708811055521516,
      "learning_rate": 4.995705165235726e-06,
      "loss": 0.2472,
      "step": 130
    },
    {
      "epoch": 0.06828696162116374,
      "grad_norm": 1.2719045463078689,
      "learning_rate": 4.995448890448449e-06,
      "loss": 0.2446,
      "step": 131
    },
    {
      "epoch": 0.06880823613735583,
      "grad_norm": 1.2812681207614192,
      "learning_rate": 4.995185196555242e-06,
      "loss": 0.2364,
      "step": 132
    },
    {
      "epoch": 0.06932951065354792,
      "grad_norm": 1.2101435618357532,
      "learning_rate": 4.994914084340082e-06,
      "loss": 0.2469,
      "step": 133
    },
    {
      "epoch": 0.06985078516974001,
      "grad_norm": 1.233938197016408,
      "learning_rate": 4.994635554608999e-06,
      "loss": 0.2286,
      "step": 134
    },
    {
      "epoch": 0.0703720596859321,
      "grad_norm": 1.2730808235698976,
      "learning_rate": 4.994349608190079e-06,
      "loss": 0.24,
      "step": 135
    },
    {
      "epoch": 0.0708933342021242,
      "grad_norm": 1.384333360859674,
      "learning_rate": 4.994056245933454e-06,
      "loss": 0.2393,
      "step": 136
    },
    {
      "epoch": 0.07141460871831629,
      "grad_norm": 1.2385359011388384,
      "learning_rate": 4.993755468711308e-06,
      "loss": 0.2446,
      "step": 137
    },
    {
      "epoch": 0.07193588323450838,
      "grad_norm": 1.2478572413164555,
      "learning_rate": 4.993447277417867e-06,
      "loss": 0.2269,
      "step": 138
    },
    {
      "epoch": 0.07245715775070047,
      "grad_norm": 1.2105094944153947,
      "learning_rate": 4.993131672969402e-06,
      "loss": 0.2484,
      "step": 139
    },
    {
      "epoch": 0.07297843226689255,
      "grad_norm": 1.3486869348259192,
      "learning_rate": 4.992808656304221e-06,
      "loss": 0.238,
      "step": 140
    },
    {
      "epoch": 0.07349970678308464,
      "grad_norm": 1.20349798497798,
      "learning_rate": 4.99247822838267e-06,
      "loss": 0.231,
      "step": 141
    },
    {
      "epoch": 0.07402098129927673,
      "grad_norm": 1.1821840497420153,
      "learning_rate": 4.99214039018713e-06,
      "loss": 0.2373,
      "step": 142
    },
    {
      "epoch": 0.07454225581546882,
      "grad_norm": 1.452457587792453,
      "learning_rate": 4.991795142722012e-06,
      "loss": 0.2395,
      "step": 143
    },
    {
      "epoch": 0.07506353033166091,
      "grad_norm": 1.239932267685592,
      "learning_rate": 4.9914424870137565e-06,
      "loss": 0.235,
      "step": 144
    },
    {
      "epoch": 0.075584804847853,
      "grad_norm": 1.22718860947722,
      "learning_rate": 4.991082424110826e-06,
      "loss": 0.2427,
      "step": 145
    },
    {
      "epoch": 0.07610607936404509,
      "grad_norm": 1.146867050689835,
      "learning_rate": 4.990714955083709e-06,
      "loss": 0.2437,
      "step": 146
    },
    {
      "epoch": 0.07662735388023718,
      "grad_norm": 1.1165066351844564,
      "learning_rate": 4.9903400810249116e-06,
      "loss": 0.2237,
      "step": 147
    },
    {
      "epoch": 0.07714862839642928,
      "grad_norm": 1.22634913619575,
      "learning_rate": 4.9899578030489534e-06,
      "loss": 0.2277,
      "step": 148
    },
    {
      "epoch": 0.07766990291262135,
      "grad_norm": 1.1756420690402456,
      "learning_rate": 4.9895681222923685e-06,
      "loss": 0.2349,
      "step": 149
    },
    {
      "epoch": 0.07819117742881344,
      "grad_norm": 1.2768853313777127,
      "learning_rate": 4.989171039913698e-06,
      "loss": 0.2378,
      "step": 150
    },
    {
      "epoch": 0.07871245194500553,
      "grad_norm": 1.298030043050591,
      "learning_rate": 4.9887665570934905e-06,
      "loss": 0.2309,
      "step": 151
    },
    {
      "epoch": 0.07923372646119763,
      "grad_norm": 1.1324424296737856,
      "learning_rate": 4.988354675034296e-06,
      "loss": 0.2242,
      "step": 152
    },
    {
      "epoch": 0.07975500097738972,
      "grad_norm": 1.289394446194592,
      "learning_rate": 4.987935394960661e-06,
      "loss": 0.2285,
      "step": 153
    },
    {
      "epoch": 0.08027627549358181,
      "grad_norm": 1.2538484453836192,
      "learning_rate": 4.98750871811913e-06,
      "loss": 0.2545,
      "step": 154
    },
    {
      "epoch": 0.0807975500097739,
      "grad_norm": 1.2220953206167016,
      "learning_rate": 4.987074645778234e-06,
      "loss": 0.2277,
      "step": 155
    },
    {
      "epoch": 0.08131882452596599,
      "grad_norm": 1.187543706345677,
      "learning_rate": 4.986633179228495e-06,
      "loss": 0.2327,
      "step": 156
    },
    {
      "epoch": 0.08184009904215808,
      "grad_norm": 1.2934140935539535,
      "learning_rate": 4.986184319782418e-06,
      "loss": 0.2424,
      "step": 157
    },
    {
      "epoch": 0.08236137355835016,
      "grad_norm": 1.2219190172913152,
      "learning_rate": 4.9857280687744856e-06,
      "loss": 0.2508,
      "step": 158
    },
    {
      "epoch": 0.08288264807454225,
      "grad_norm": 1.2427749225301898,
      "learning_rate": 4.985264427561158e-06,
      "loss": 0.2297,
      "step": 159
    },
    {
      "epoch": 0.08340392259073434,
      "grad_norm": 1.2357587940866266,
      "learning_rate": 4.984793397520865e-06,
      "loss": 0.2307,
      "step": 160
    },
    {
      "epoch": 0.08392519710692643,
      "grad_norm": 1.2364383187414076,
      "learning_rate": 4.984314980054005e-06,
      "loss": 0.2323,
      "step": 161
    },
    {
      "epoch": 0.08444647162311852,
      "grad_norm": 1.2393212896003536,
      "learning_rate": 4.983829176582939e-06,
      "loss": 0.2473,
      "step": 162
    },
    {
      "epoch": 0.08496774613931062,
      "grad_norm": 1.3103587969557084,
      "learning_rate": 4.983335988551986e-06,
      "loss": 0.2344,
      "step": 163
    },
    {
      "epoch": 0.0854890206555027,
      "grad_norm": 1.2950039813259882,
      "learning_rate": 4.982835417427424e-06,
      "loss": 0.2447,
      "step": 164
    },
    {
      "epoch": 0.0860102951716948,
      "grad_norm": 1.1823175382955655,
      "learning_rate": 4.982327464697476e-06,
      "loss": 0.2339,
      "step": 165
    },
    {
      "epoch": 0.08653156968788689,
      "grad_norm": 1.3397014918648167,
      "learning_rate": 4.981812131872315e-06,
      "loss": 0.2421,
      "step": 166
    },
    {
      "epoch": 0.08705284420407897,
      "grad_norm": 1.1880479785297025,
      "learning_rate": 4.981289420484051e-06,
      "loss": 0.2063,
      "step": 167
    },
    {
      "epoch": 0.08757411872027106,
      "grad_norm": 1.1844610096859758,
      "learning_rate": 4.980759332086736e-06,
      "loss": 0.2223,
      "step": 168
    },
    {
      "epoch": 0.08809539323646315,
      "grad_norm": 1.1743257421756592,
      "learning_rate": 4.980221868256351e-06,
      "loss": 0.2518,
      "step": 169
    },
    {
      "epoch": 0.08861666775265524,
      "grad_norm": 1.2816501589273177,
      "learning_rate": 4.9796770305908045e-06,
      "loss": 0.2244,
      "step": 170
    },
    {
      "epoch": 0.08913794226884733,
      "grad_norm": 1.3332187743906725,
      "learning_rate": 4.979124820709931e-06,
      "loss": 0.2315,
      "step": 171
    },
    {
      "epoch": 0.08965921678503942,
      "grad_norm": 1.1756963489546968,
      "learning_rate": 4.978565240255477e-06,
      "loss": 0.226,
      "step": 172
    },
    {
      "epoch": 0.09018049130123151,
      "grad_norm": 1.4972378793543208,
      "learning_rate": 4.977998290891109e-06,
      "loss": 0.2314,
      "step": 173
    },
    {
      "epoch": 0.0907017658174236,
      "grad_norm": 1.3196899591714728,
      "learning_rate": 4.9774239743023975e-06,
      "loss": 0.2359,
      "step": 174
    },
    {
      "epoch": 0.0912230403336157,
      "grad_norm": 1.4208086181110644,
      "learning_rate": 4.976842292196817e-06,
      "loss": 0.2371,
      "step": 175
    },
    {
      "epoch": 0.09174431484980777,
      "grad_norm": 1.5874713119749235,
      "learning_rate": 4.9762532463037385e-06,
      "loss": 0.2346,
      "step": 176
    },
    {
      "epoch": 0.09226558936599986,
      "grad_norm": 1.3081464000461915,
      "learning_rate": 4.97565683837443e-06,
      "loss": 0.2377,
      "step": 177
    },
    {
      "epoch": 0.09278686388219196,
      "grad_norm": 1.4793193226923158,
      "learning_rate": 4.9750530701820446e-06,
      "loss": 0.2301,
      "step": 178
    },
    {
      "epoch": 0.09330813839838405,
      "grad_norm": 1.45104441094533,
      "learning_rate": 4.974441943521616e-06,
      "loss": 0.246,
      "step": 179
    },
    {
      "epoch": 0.09382941291457614,
      "grad_norm": 1.1712951871890278,
      "learning_rate": 4.9738234602100605e-06,
      "loss": 0.2255,
      "step": 180
    },
    {
      "epoch": 0.09435068743076823,
      "grad_norm": 1.4773502084469639,
      "learning_rate": 4.97319762208616e-06,
      "loss": 0.2326,
      "step": 181
    },
    {
      "epoch": 0.09487196194696032,
      "grad_norm": 1.306772468267686,
      "learning_rate": 4.972564431010567e-06,
      "loss": 0.245,
      "step": 182
    },
    {
      "epoch": 0.09539323646315241,
      "grad_norm": 1.209641545426326,
      "learning_rate": 4.971923888865792e-06,
      "loss": 0.2247,
      "step": 183
    },
    {
      "epoch": 0.0959145109793445,
      "grad_norm": 1.200834610292657,
      "learning_rate": 4.971275997556203e-06,
      "loss": 0.2073,
      "step": 184
    },
    {
      "epoch": 0.09643578549553658,
      "grad_norm": 1.266596376115279,
      "learning_rate": 4.970620759008015e-06,
      "loss": 0.2284,
      "step": 185
    },
    {
      "epoch": 0.09695706001172867,
      "grad_norm": 1.4118237346104745,
      "learning_rate": 4.969958175169291e-06,
      "loss": 0.2483,
      "step": 186
    },
    {
      "epoch": 0.09747833452792076,
      "grad_norm": 1.182726741808869,
      "learning_rate": 4.969288248009924e-06,
      "loss": 0.2184,
      "step": 187
    },
    {
      "epoch": 0.09799960904411285,
      "grad_norm": 1.2332451914842804,
      "learning_rate": 4.968610979521647e-06,
      "loss": 0.2229,
      "step": 188
    },
    {
      "epoch": 0.09852088356030494,
      "grad_norm": 1.296696985788193,
      "learning_rate": 4.967926371718017e-06,
      "loss": 0.2456,
      "step": 189
    },
    {
      "epoch": 0.09904215807649704,
      "grad_norm": 1.2471025773185285,
      "learning_rate": 4.96723442663441e-06,
      "loss": 0.2358,
      "step": 190
    },
    {
      "epoch": 0.09956343259268913,
      "grad_norm": 1.2081020978259593,
      "learning_rate": 4.966535146328014e-06,
      "loss": 0.2259,
      "step": 191
    },
    {
      "epoch": 0.10008470710888122,
      "grad_norm": 1.1747127101005468,
      "learning_rate": 4.965828532877831e-06,
      "loss": 0.2262,
      "step": 192
    },
    {
      "epoch": 0.10060598162507331,
      "grad_norm": 1.239720059272851,
      "learning_rate": 4.96511458838466e-06,
      "loss": 0.2475,
      "step": 193
    },
    {
      "epoch": 0.10112725614126539,
      "grad_norm": 1.3048056513019626,
      "learning_rate": 4.964393314971096e-06,
      "loss": 0.2311,
      "step": 194
    },
    {
      "epoch": 0.10164853065745748,
      "grad_norm": 1.183469548460636,
      "learning_rate": 4.963664714781525e-06,
      "loss": 0.2395,
      "step": 195
    },
    {
      "epoch": 0.10216980517364957,
      "grad_norm": 1.333533223581565,
      "learning_rate": 4.962928789982117e-06,
      "loss": 0.2191,
      "step": 196
    },
    {
      "epoch": 0.10269107968984166,
      "grad_norm": 1.196909258791872,
      "learning_rate": 4.9621855427608134e-06,
      "loss": 0.2208,
      "step": 197
    },
    {
      "epoch": 0.10321235420603375,
      "grad_norm": 1.2853727381550513,
      "learning_rate": 4.961434975327331e-06,
      "loss": 0.2459,
      "step": 198
    },
    {
      "epoch": 0.10373362872222584,
      "grad_norm": 1.1160859614220253,
      "learning_rate": 4.960677089913146e-06,
      "loss": 0.2328,
      "step": 199
    },
    {
      "epoch": 0.10425490323841793,
      "grad_norm": 1.1166526080209556,
      "learning_rate": 4.959911888771496e-06,
      "loss": 0.2312,
      "step": 200
    },
    {
      "epoch": 0.10477617775461003,
      "grad_norm": 1.0880873416190306,
      "learning_rate": 4.959139374177364e-06,
      "loss": 0.2074,
      "step": 201
    },
    {
      "epoch": 0.10529745227080212,
      "grad_norm": 1.2040328431798148,
      "learning_rate": 4.958359548427478e-06,
      "loss": 0.2357,
      "step": 202
    },
    {
      "epoch": 0.10581872678699421,
      "grad_norm": 1.0518301757015505,
      "learning_rate": 4.957572413840303e-06,
      "loss": 0.2182,
      "step": 203
    },
    {
      "epoch": 0.10634000130318629,
      "grad_norm": 1.154934427237305,
      "learning_rate": 4.956777972756033e-06,
      "loss": 0.2237,
      "step": 204
    },
    {
      "epoch": 0.10686127581937838,
      "grad_norm": 1.1276890679621758,
      "learning_rate": 4.955976227536584e-06,
      "loss": 0.2105,
      "step": 205
    },
    {
      "epoch": 0.10738255033557047,
      "grad_norm": 1.1058755543209657,
      "learning_rate": 4.95516718056559e-06,
      "loss": 0.2273,
      "step": 206
    },
    {
      "epoch": 0.10790382485176256,
      "grad_norm": 1.2126591248862537,
      "learning_rate": 4.95435083424839e-06,
      "loss": 0.239,
      "step": 207
    },
    {
      "epoch": 0.10842509936795465,
      "grad_norm": 1.2627590952539536,
      "learning_rate": 4.953527191012029e-06,
      "loss": 0.2417,
      "step": 208
    },
    {
      "epoch": 0.10894637388414674,
      "grad_norm": 1.2188135211116549,
      "learning_rate": 4.95269625330524e-06,
      "loss": 0.2282,
      "step": 209
    },
    {
      "epoch": 0.10946764840033883,
      "grad_norm": 1.087249434059975,
      "learning_rate": 4.951858023598448e-06,
      "loss": 0.2013,
      "step": 210
    },
    {
      "epoch": 0.10998892291653092,
      "grad_norm": 1.1192373478037494,
      "learning_rate": 4.951012504383756e-06,
      "loss": 0.2271,
      "step": 211
    },
    {
      "epoch": 0.11051019743272301,
      "grad_norm": 1.1846883599390468,
      "learning_rate": 4.9501596981749375e-06,
      "loss": 0.2393,
      "step": 212
    },
    {
      "epoch": 0.11103147194891509,
      "grad_norm": 1.2820136837325322,
      "learning_rate": 4.949299607507434e-06,
      "loss": 0.2311,
      "step": 213
    },
    {
      "epoch": 0.11155274646510718,
      "grad_norm": 1.0711936772195123,
      "learning_rate": 4.94843223493834e-06,
      "loss": 0.2029,
      "step": 214
    },
    {
      "epoch": 0.11207402098129927,
      "grad_norm": 1.1864803079209343,
      "learning_rate": 4.947557583046403e-06,
      "loss": 0.2342,
      "step": 215
    },
    {
      "epoch": 0.11259529549749137,
      "grad_norm": 1.1855853986108895,
      "learning_rate": 4.94667565443201e-06,
      "loss": 0.2188,
      "step": 216
    },
    {
      "epoch": 0.11311657001368346,
      "grad_norm": 1.2164131878100903,
      "learning_rate": 4.945786451717183e-06,
      "loss": 0.2376,
      "step": 217
    },
    {
      "epoch": 0.11363784452987555,
      "grad_norm": 1.2725613700670313,
      "learning_rate": 4.944889977545571e-06,
      "loss": 0.2353,
      "step": 218
    },
    {
      "epoch": 0.11415911904606764,
      "grad_norm": 1.2700871402224212,
      "learning_rate": 4.94398623458244e-06,
      "loss": 0.249,
      "step": 219
    },
    {
      "epoch": 0.11468039356225973,
      "grad_norm": 1.0477852266713497,
      "learning_rate": 4.943075225514667e-06,
      "loss": 0.2212,
      "step": 220
    },
    {
      "epoch": 0.11520166807845182,
      "grad_norm": 1.1949076765095417,
      "learning_rate": 4.942156953050733e-06,
      "loss": 0.2427,
      "step": 221
    },
    {
      "epoch": 0.1157229425946439,
      "grad_norm": 1.1910711165686847,
      "learning_rate": 4.94123141992071e-06,
      "loss": 0.2315,
      "step": 222
    },
    {
      "epoch": 0.11624421711083599,
      "grad_norm": 1.1171795697179367,
      "learning_rate": 4.940298628876261e-06,
      "loss": 0.2323,
      "step": 223
    },
    {
      "epoch": 0.11676549162702808,
      "grad_norm": 1.119061233540857,
      "learning_rate": 4.9393585826906245e-06,
      "loss": 0.2208,
      "step": 224
    },
    {
      "epoch": 0.11728676614322017,
      "grad_norm": 1.1501275918061766,
      "learning_rate": 4.938411284158608e-06,
      "loss": 0.219,
      "step": 225
    },
    {
      "epoch": 0.11780804065941226,
      "grad_norm": 1.2954253968552865,
      "learning_rate": 4.937456736096581e-06,
      "loss": 0.2482,
      "step": 226
    },
    {
      "epoch": 0.11832931517560435,
      "grad_norm": 1.2008217826774177,
      "learning_rate": 4.936494941342469e-06,
      "loss": 0.2373,
      "step": 227
    },
    {
      "epoch": 0.11885058969179645,
      "grad_norm": 1.2630056268710672,
      "learning_rate": 4.93552590275574e-06,
      "loss": 0.2323,
      "step": 228
    },
    {
      "epoch": 0.11937186420798854,
      "grad_norm": 1.1943668197100574,
      "learning_rate": 4.934549623217399e-06,
      "loss": 0.2266,
      "step": 229
    },
    {
      "epoch": 0.11989313872418063,
      "grad_norm": 1.2274448290359987,
      "learning_rate": 4.9335661056299755e-06,
      "loss": 0.2325,
      "step": 230
    },
    {
      "epoch": 0.1204144132403727,
      "grad_norm": 1.1933170165235585,
      "learning_rate": 4.932575352917524e-06,
      "loss": 0.2224,
      "step": 231
    },
    {
      "epoch": 0.1209356877565648,
      "grad_norm": 1.1015086989558849,
      "learning_rate": 4.931577368025607e-06,
      "loss": 0.2159,
      "step": 232
    },
    {
      "epoch": 0.12145696227275689,
      "grad_norm": 1.323236281530893,
      "learning_rate": 4.930572153921287e-06,
      "loss": 0.2418,
      "step": 233
    },
    {
      "epoch": 0.12197823678894898,
      "grad_norm": 1.328150537223152,
      "learning_rate": 4.92955971359312e-06,
      "loss": 0.2296,
      "step": 234
    },
    {
      "epoch": 0.12249951130514107,
      "grad_norm": 1.1592367964351038,
      "learning_rate": 4.928540050051146e-06,
      "loss": 0.211,
      "step": 235
    },
    {
      "epoch": 0.12302078582133316,
      "grad_norm": 1.1266688403160974,
      "learning_rate": 4.927513166326881e-06,
      "loss": 0.2147,
      "step": 236
    },
    {
      "epoch": 0.12354206033752525,
      "grad_norm": 1.3104950047973025,
      "learning_rate": 4.9264790654733076e-06,
      "loss": 0.2274,
      "step": 237
    },
    {
      "epoch": 0.12406333485371734,
      "grad_norm": 1.3857089010320038,
      "learning_rate": 4.925437750564863e-06,
      "loss": 0.235,
      "step": 238
    },
    {
      "epoch": 0.12458460936990944,
      "grad_norm": 1.0833873266614178,
      "learning_rate": 4.924389224697433e-06,
      "loss": 0.2272,
      "step": 239
    },
    {
      "epoch": 0.12510588388610153,
      "grad_norm": 1.2529871243213786,
      "learning_rate": 4.923333490988343e-06,
      "loss": 0.2412,
      "step": 240
    },
    {
      "epoch": 0.1256271584022936,
      "grad_norm": 1.118800053266488,
      "learning_rate": 4.922270552576347e-06,
      "loss": 0.2168,
      "step": 241
    },
    {
      "epoch": 0.1261484329184857,
      "grad_norm": 1.153865139835547,
      "learning_rate": 4.921200412621619e-06,
      "loss": 0.2133,
      "step": 242
    },
    {
      "epoch": 0.1266697074346778,
      "grad_norm": 1.1260326719408678,
      "learning_rate": 4.920123074305743e-06,
      "loss": 0.2246,
      "step": 243
    },
    {
      "epoch": 0.12719098195086986,
      "grad_norm": 1.2169366957395464,
      "learning_rate": 4.919038540831705e-06,
      "loss": 0.2182,
      "step": 244
    },
    {
      "epoch": 0.12771225646706197,
      "grad_norm": 1.106220335006384,
      "learning_rate": 4.917946815423883e-06,
      "loss": 0.2119,
      "step": 245
    },
    {
      "epoch": 0.12823353098325405,
      "grad_norm": 1.128901025595511,
      "learning_rate": 4.916847901328035e-06,
      "loss": 0.231,
      "step": 246
    },
    {
      "epoch": 0.12875480549944615,
      "grad_norm": 1.1605446687661631,
      "learning_rate": 4.915741801811294e-06,
      "loss": 0.2362,
      "step": 247
    },
    {
      "epoch": 0.12927608001563823,
      "grad_norm": 1.282977790360365,
      "learning_rate": 4.914628520162154e-06,
      "loss": 0.2375,
      "step": 248
    },
    {
      "epoch": 0.12979735453183033,
      "grad_norm": 1.0118063392484975,
      "learning_rate": 4.913508059690461e-06,
      "loss": 0.1983,
      "step": 249
    },
    {
      "epoch": 0.1303186290480224,
      "grad_norm": 1.1850969192532463,
      "learning_rate": 4.912380423727405e-06,
      "loss": 0.2208,
      "step": 250
    },
    {
      "epoch": 0.13083990356421452,
      "grad_norm": 1.1567774979095986,
      "learning_rate": 4.911245615625512e-06,
      "loss": 0.2163,
      "step": 251
    },
    {
      "epoch": 0.1313611780804066,
      "grad_norm": 1.2330915782132366,
      "learning_rate": 4.910103638758627e-06,
      "loss": 0.2379,
      "step": 252
    },
    {
      "epoch": 0.13188245259659867,
      "grad_norm": 1.1057712455940338,
      "learning_rate": 4.9089544965219095e-06,
      "loss": 0.224,
      "step": 253
    },
    {
      "epoch": 0.13240372711279078,
      "grad_norm": 1.1137495249197693,
      "learning_rate": 4.907798192331821e-06,
      "loss": 0.2088,
      "step": 254
    },
    {
      "epoch": 0.13292500162898285,
      "grad_norm": 1.183848047830925,
      "learning_rate": 4.90663472962612e-06,
      "loss": 0.2211,
      "step": 255
    },
    {
      "epoch": 0.13344627614517496,
      "grad_norm": 1.210251233263837,
      "learning_rate": 4.905464111863841e-06,
      "loss": 0.2287,
      "step": 256
    },
    {
      "epoch": 0.13396755066136704,
      "grad_norm": 1.2156314491143745,
      "learning_rate": 4.904286342525298e-06,
      "loss": 0.2339,
      "step": 257
    },
    {
      "epoch": 0.13448882517755914,
      "grad_norm": 1.1335883548375618,
      "learning_rate": 4.903101425112062e-06,
      "loss": 0.2327,
      "step": 258
    },
    {
      "epoch": 0.13501009969375122,
      "grad_norm": 1.1196076296130062,
      "learning_rate": 4.9019093631469575e-06,
      "loss": 0.2253,
      "step": 259
    },
    {
      "epoch": 0.13553137420994332,
      "grad_norm": 1.2260470187534245,
      "learning_rate": 4.900710160174048e-06,
      "loss": 0.248,
      "step": 260
    },
    {
      "epoch": 0.1360526487261354,
      "grad_norm": 1.1196565222861217,
      "learning_rate": 4.899503819758633e-06,
      "loss": 0.2174,
      "step": 261
    },
    {
      "epoch": 0.13657392324232748,
      "grad_norm": 1.0864136858861488,
      "learning_rate": 4.898290345487226e-06,
      "loss": 0.2219,
      "step": 262
    },
    {
      "epoch": 0.13709519775851958,
      "grad_norm": 1.0137672217351832,
      "learning_rate": 4.897069740967554e-06,
      "loss": 0.2131,
      "step": 263
    },
    {
      "epoch": 0.13761647227471166,
      "grad_norm": 1.1037519976554437,
      "learning_rate": 4.89584200982854e-06,
      "loss": 0.2284,
      "step": 264
    },
    {
      "epoch": 0.13813774679090376,
      "grad_norm": 1.1303347452809993,
      "learning_rate": 4.894607155720294e-06,
      "loss": 0.2418,
      "step": 265
    },
    {
      "epoch": 0.13865902130709584,
      "grad_norm": 1.1657801655573423,
      "learning_rate": 4.893365182314108e-06,
      "loss": 0.2385,
      "step": 266
    },
    {
      "epoch": 0.13918029582328795,
      "grad_norm": 1.1427107191789623,
      "learning_rate": 4.892116093302436e-06,
      "loss": 0.2229,
      "step": 267
    },
    {
      "epoch": 0.13970157033948002,
      "grad_norm": 1.1330210325415613,
      "learning_rate": 4.890859892398886e-06,
      "loss": 0.2355,
      "step": 268
    },
    {
      "epoch": 0.14022284485567213,
      "grad_norm": 1.105052560447426,
      "learning_rate": 4.889596583338213e-06,
      "loss": 0.2344,
      "step": 269
    },
    {
      "epoch": 0.1407441193718642,
      "grad_norm": 1.0965253134452662,
      "learning_rate": 4.8883261698763045e-06,
      "loss": 0.2188,
      "step": 270
    },
    {
      "epoch": 0.14126539388805628,
      "grad_norm": 1.147653006818557,
      "learning_rate": 4.887048655790169e-06,
      "loss": 0.2274,
      "step": 271
    },
    {
      "epoch": 0.1417866684042484,
      "grad_norm": 1.0895631770032759,
      "learning_rate": 4.8857640448779246e-06,
      "loss": 0.2398,
      "step": 272
    },
    {
      "epoch": 0.14230794292044047,
      "grad_norm": 1.0681929006676545,
      "learning_rate": 4.884472340958791e-06,
      "loss": 0.2221,
      "step": 273
    },
    {
      "epoch": 0.14282921743663257,
      "grad_norm": 1.1493300322422102,
      "learning_rate": 4.883173547873073e-06,
      "loss": 0.2318,
      "step": 274
    },
    {
      "epoch": 0.14335049195282465,
      "grad_norm": 1.178680600036612,
      "learning_rate": 4.881867669482157e-06,
      "loss": 0.2313,
      "step": 275
    },
    {
      "epoch": 0.14387176646901675,
      "grad_norm": 1.104816541922132,
      "learning_rate": 4.880554709668486e-06,
      "loss": 0.2332,
      "step": 276
    },
    {
      "epoch": 0.14439304098520883,
      "grad_norm": 1.1737237666373883,
      "learning_rate": 4.879234672335564e-06,
      "loss": 0.235,
      "step": 277
    },
    {
      "epoch": 0.14491431550140094,
      "grad_norm": 1.2092792760641806,
      "learning_rate": 4.8779075614079354e-06,
      "loss": 0.2395,
      "step": 278
    },
    {
      "epoch": 0.145435590017593,
      "grad_norm": 1.0839887894678346,
      "learning_rate": 4.87657338083117e-06,
      "loss": 0.2223,
      "step": 279
    },
    {
      "epoch": 0.1459568645337851,
      "grad_norm": 1.0999122775478736,
      "learning_rate": 4.875232134571863e-06,
      "loss": 0.2275,
      "step": 280
    },
    {
      "epoch": 0.1464781390499772,
      "grad_norm": 1.1636449280470644,
      "learning_rate": 4.8738838266176094e-06,
      "loss": 0.2211,
      "step": 281
    },
    {
      "epoch": 0.14699941356616927,
      "grad_norm": 1.1790289908665157,
      "learning_rate": 4.872528460977005e-06,
      "loss": 0.229,
      "step": 282
    },
    {
      "epoch": 0.14752068808236138,
      "grad_norm": 1.1989633537081639,
      "learning_rate": 4.871166041679626e-06,
      "loss": 0.2258,
      "step": 283
    },
    {
      "epoch": 0.14804196259855346,
      "grad_norm": 1.2423687257331835,
      "learning_rate": 4.869796572776018e-06,
      "loss": 0.2323,
      "step": 284
    },
    {
      "epoch": 0.14856323711474556,
      "grad_norm": 1.2228142589786297,
      "learning_rate": 4.868420058337687e-06,
      "loss": 0.2107,
      "step": 285
    },
    {
      "epoch": 0.14908451163093764,
      "grad_norm": 0.9926879454981049,
      "learning_rate": 4.867036502457087e-06,
      "loss": 0.2107,
      "step": 286
    },
    {
      "epoch": 0.14960578614712974,
      "grad_norm": 1.2008996239053435,
      "learning_rate": 4.865645909247604e-06,
      "loss": 0.2249,
      "step": 287
    },
    {
      "epoch": 0.15012706066332182,
      "grad_norm": 1.1542733944022654,
      "learning_rate": 4.864248282843548e-06,
      "loss": 0.2399,
      "step": 288
    },
    {
      "epoch": 0.1506483351795139,
      "grad_norm": 1.1691556753754107,
      "learning_rate": 4.862843627400139e-06,
      "loss": 0.2151,
      "step": 289
    },
    {
      "epoch": 0.151169609695706,
      "grad_norm": 1.3273619113907211,
      "learning_rate": 4.861431947093494e-06,
      "loss": 0.2175,
      "step": 290
    },
    {
      "epoch": 0.15169088421189808,
      "grad_norm": 1.2832476992194477,
      "learning_rate": 4.860013246120616e-06,
      "loss": 0.2298,
      "step": 291
    },
    {
      "epoch": 0.15221215872809019,
      "grad_norm": 1.084146667118211,
      "learning_rate": 4.85858752869938e-06,
      "loss": 0.2089,
      "step": 292
    },
    {
      "epoch": 0.15273343324428226,
      "grad_norm": 1.1508676857394082,
      "learning_rate": 4.857154799068522e-06,
      "loss": 0.2266,
      "step": 293
    },
    {
      "epoch": 0.15325470776047437,
      "grad_norm": 1.1639283738815098,
      "learning_rate": 4.855715061487626e-06,
      "loss": 0.2255,
      "step": 294
    },
    {
      "epoch": 0.15377598227666645,
      "grad_norm": 1.2554700518066422,
      "learning_rate": 4.8542683202371105e-06,
      "loss": 0.2223,
      "step": 295
    },
    {
      "epoch": 0.15429725679285855,
      "grad_norm": 1.3283787770601974,
      "learning_rate": 4.8528145796182155e-06,
      "loss": 0.2154,
      "step": 296
    },
    {
      "epoch": 0.15481853130905063,
      "grad_norm": 1.1601511824620128,
      "learning_rate": 4.851353843952992e-06,
      "loss": 0.2228,
      "step": 297
    },
    {
      "epoch": 0.1553398058252427,
      "grad_norm": 1.220795788128704,
      "learning_rate": 4.849886117584286e-06,
      "loss": 0.2335,
      "step": 298
    },
    {
      "epoch": 0.1558610803414348,
      "grad_norm": 1.2555523292108537,
      "learning_rate": 4.84841140487573e-06,
      "loss": 0.2097,
      "step": 299
    },
    {
      "epoch": 0.1563823548576269,
      "grad_norm": 1.164702238891539,
      "learning_rate": 4.846929710211724e-06,
      "loss": 0.2272,
      "step": 300
    },
    {
      "epoch": 0.156903629373819,
      "grad_norm": 1.183879685349843,
      "learning_rate": 4.845441037997428e-06,
      "loss": 0.2321,
      "step": 301
    },
    {
      "epoch": 0.15742490389001107,
      "grad_norm": 1.0808256708331534,
      "learning_rate": 4.843945392658744e-06,
      "loss": 0.2212,
      "step": 302
    },
    {
      "epoch": 0.15794617840620317,
      "grad_norm": 1.1207884103366483,
      "learning_rate": 4.84244277864231e-06,
      "loss": 0.2312,
      "step": 303
    },
    {
      "epoch": 0.15846745292239525,
      "grad_norm": 1.2420206921311776,
      "learning_rate": 4.840933200415479e-06,
      "loss": 0.2328,
      "step": 304
    },
    {
      "epoch": 0.15898872743858736,
      "grad_norm": 1.1302339000354495,
      "learning_rate": 4.839416662466307e-06,
      "loss": 0.2317,
      "step": 305
    },
    {
      "epoch": 0.15951000195477943,
      "grad_norm": 1.1663665978420636,
      "learning_rate": 4.837893169303548e-06,
      "loss": 0.2297,
      "step": 306
    },
    {
      "epoch": 0.1600312764709715,
      "grad_norm": 1.2551140786028032,
      "learning_rate": 4.836362725456628e-06,
      "loss": 0.2182,
      "step": 307
    },
    {
      "epoch": 0.16055255098716362,
      "grad_norm": 1.1278960330544754,
      "learning_rate": 4.834825335475641e-06,
      "loss": 0.2251,
      "step": 308
    },
    {
      "epoch": 0.1610738255033557,
      "grad_norm": 1.117111565112251,
      "learning_rate": 4.833281003931331e-06,
      "loss": 0.2165,
      "step": 309
    },
    {
      "epoch": 0.1615951000195478,
      "grad_norm": 1.0974889618517698,
      "learning_rate": 4.831729735415081e-06,
      "loss": 0.2333,
      "step": 310
    },
    {
      "epoch": 0.16211637453573988,
      "grad_norm": 1.1489672264426987,
      "learning_rate": 4.830171534538895e-06,
      "loss": 0.2408,
      "step": 311
    },
    {
      "epoch": 0.16263764905193198,
      "grad_norm": 1.0739457008165634,
      "learning_rate": 4.828606405935391e-06,
      "loss": 0.2118,
      "step": 312
    },
    {
      "epoch": 0.16315892356812406,
      "grad_norm": 1.1562271241695659,
      "learning_rate": 4.827034354257782e-06,
      "loss": 0.2259,
      "step": 313
    },
    {
      "epoch": 0.16368019808431616,
      "grad_norm": 1.146438735977002,
      "learning_rate": 4.825455384179864e-06,
      "loss": 0.2197,
      "step": 314
    },
    {
      "epoch": 0.16420147260050824,
      "grad_norm": 1.2160636638243234,
      "learning_rate": 4.823869500395999e-06,
      "loss": 0.2346,
      "step": 315
    },
    {
      "epoch": 0.16472274711670032,
      "grad_norm": 1.1626081501279093,
      "learning_rate": 4.822276707621109e-06,
      "loss": 0.2298,
      "step": 316
    },
    {
      "epoch": 0.16524402163289242,
      "grad_norm": 1.112204442031006,
      "learning_rate": 4.820677010590652e-06,
      "loss": 0.227,
      "step": 317
    },
    {
      "epoch": 0.1657652961490845,
      "grad_norm": 1.2103180677038683,
      "learning_rate": 4.819070414060616e-06,
      "loss": 0.2388,
      "step": 318
    },
    {
      "epoch": 0.1662865706652766,
      "grad_norm": 1.1743106791258793,
      "learning_rate": 4.817456922807499e-06,
      "loss": 0.2338,
      "step": 319
    },
    {
      "epoch": 0.16680784518146868,
      "grad_norm": 1.0896117602385076,
      "learning_rate": 4.815836541628299e-06,
      "loss": 0.2299,
      "step": 320
    },
    {
      "epoch": 0.1673291196976608,
      "grad_norm": 1.1892438430988612,
      "learning_rate": 4.814209275340498e-06,
      "loss": 0.2047,
      "step": 321
    },
    {
      "epoch": 0.16785039421385287,
      "grad_norm": 1.1452649056862731,
      "learning_rate": 4.8125751287820484e-06,
      "loss": 0.2189,
      "step": 322
    },
    {
      "epoch": 0.16837166873004497,
      "grad_norm": 1.1511519078648214,
      "learning_rate": 4.8109341068113566e-06,
      "loss": 0.2355,
      "step": 323
    },
    {
      "epoch": 0.16889294324623705,
      "grad_norm": 1.1351415607365043,
      "learning_rate": 4.8092862143072705e-06,
      "loss": 0.2311,
      "step": 324
    },
    {
      "epoch": 0.16941421776242913,
      "grad_norm": 1.1787027223084978,
      "learning_rate": 4.807631456169064e-06,
      "loss": 0.2316,
      "step": 325
    },
    {
      "epoch": 0.16993549227862123,
      "grad_norm": 1.1242332884299846,
      "learning_rate": 4.805969837316424e-06,
      "loss": 0.2108,
      "step": 326
    },
    {
      "epoch": 0.1704567667948133,
      "grad_norm": 1.1307767960947903,
      "learning_rate": 4.804301362689435e-06,
      "loss": 0.2292,
      "step": 327
    },
    {
      "epoch": 0.1709780413110054,
      "grad_norm": 1.1743519557918656,
      "learning_rate": 4.8026260372485625e-06,
      "loss": 0.2168,
      "step": 328
    },
    {
      "epoch": 0.1714993158271975,
      "grad_norm": 1.1654596044991834,
      "learning_rate": 4.8009438659746396e-06,
      "loss": 0.2119,
      "step": 329
    },
    {
      "epoch": 0.1720205903433896,
      "grad_norm": 1.0486212534056034,
      "learning_rate": 4.7992548538688554e-06,
      "loss": 0.2227,
      "step": 330
    },
    {
      "epoch": 0.17254186485958167,
      "grad_norm": 1.2639066123447578,
      "learning_rate": 4.797559005952733e-06,
      "loss": 0.2161,
      "step": 331
    },
    {
      "epoch": 0.17306313937577378,
      "grad_norm": 1.1820605824642645,
      "learning_rate": 4.795856327268124e-06,
      "loss": 0.2338,
      "step": 332
    },
    {
      "epoch": 0.17358441389196586,
      "grad_norm": 1.0878966501873548,
      "learning_rate": 4.794146822877182e-06,
      "loss": 0.2232,
      "step": 333
    },
    {
      "epoch": 0.17410568840815793,
      "grad_norm": 1.2325481338468462,
      "learning_rate": 4.792430497862358e-06,
      "loss": 0.2181,
      "step": 334
    },
    {
      "epoch": 0.17462696292435004,
      "grad_norm": 1.1599398884974863,
      "learning_rate": 4.790707357326381e-06,
      "loss": 0.2333,
      "step": 335
    },
    {
      "epoch": 0.17514823744054211,
      "grad_norm": 1.0837820673563148,
      "learning_rate": 4.788977406392242e-06,
      "loss": 0.2034,
      "step": 336
    },
    {
      "epoch": 0.17566951195673422,
      "grad_norm": 1.182057248623312,
      "learning_rate": 4.787240650203178e-06,
      "loss": 0.2272,
      "step": 337
    },
    {
      "epoch": 0.1761907864729263,
      "grad_norm": 1.2568951261399692,
      "learning_rate": 4.785497093922662e-06,
      "loss": 0.2376,
      "step": 338
    },
    {
      "epoch": 0.1767120609891184,
      "grad_norm": 1.2211840928254953,
      "learning_rate": 4.78374674273438e-06,
      "loss": 0.23,
      "step": 339
    },
    {
      "epoch": 0.17723333550531048,
      "grad_norm": 1.0705403957934083,
      "learning_rate": 4.781989601842224e-06,
      "loss": 0.2132,
      "step": 340
    },
    {
      "epoch": 0.17775461002150258,
      "grad_norm": 1.1202000299882193,
      "learning_rate": 4.780225676470268e-06,
      "loss": 0.1965,
      "step": 341
    },
    {
      "epoch": 0.17827588453769466,
      "grad_norm": 1.047916892629429,
      "learning_rate": 4.77845497186276e-06,
      "loss": 0.2076,
      "step": 342
    },
    {
      "epoch": 0.17879715905388674,
      "grad_norm": 1.1180752657115838,
      "learning_rate": 4.776677493284101e-06,
      "loss": 0.2195,
      "step": 343
    },
    {
      "epoch": 0.17931843357007884,
      "grad_norm": 1.1592347591044683,
      "learning_rate": 4.774893246018831e-06,
      "loss": 0.2293,
      "step": 344
    },
    {
      "epoch": 0.17983970808627092,
      "grad_norm": 1.088817972776965,
      "learning_rate": 4.773102235371617e-06,
      "loss": 0.218,
      "step": 345
    },
    {
      "epoch": 0.18036098260246303,
      "grad_norm": 1.2244227223086848,
      "learning_rate": 4.771304466667229e-06,
      "loss": 0.2174,
      "step": 346
    },
    {
      "epoch": 0.1808822571186551,
      "grad_norm": 1.1747749136702952,
      "learning_rate": 4.769499945250533e-06,
      "loss": 0.2237,
      "step": 347
    },
    {
      "epoch": 0.1814035316348472,
      "grad_norm": 1.0920641978765553,
      "learning_rate": 4.76768867648647e-06,
      "loss": 0.2323,
      "step": 348
    },
    {
      "epoch": 0.1819248061510393,
      "grad_norm": 1.0542260185395076,
      "learning_rate": 4.7658706657600395e-06,
      "loss": 0.2027,
      "step": 349
    },
    {
      "epoch": 0.1824460806672314,
      "grad_norm": 1.129211741927763,
      "learning_rate": 4.764045918476288e-06,
      "loss": 0.2298,
      "step": 350
    },
    {
      "epoch": 0.18296735518342347,
      "grad_norm": 1.1380740875574586,
      "learning_rate": 4.762214440060289e-06,
      "loss": 0.2217,
      "step": 351
    },
    {
      "epoch": 0.18348862969961555,
      "grad_norm": 1.136617520410274,
      "learning_rate": 4.760376235957127e-06,
      "loss": 0.2376,
      "step": 352
    },
    {
      "epoch": 0.18400990421580765,
      "grad_norm": 1.1274285674722149,
      "learning_rate": 4.758531311631884e-06,
      "loss": 0.2103,
      "step": 353
    },
    {
      "epoch": 0.18453117873199973,
      "grad_norm": 1.0510082657536814,
      "learning_rate": 4.756679672569621e-06,
      "loss": 0.2271,
      "step": 354
    },
    {
      "epoch": 0.18505245324819183,
      "grad_norm": 1.0689796820104887,
      "learning_rate": 4.7548213242753616e-06,
      "loss": 0.2176,
      "step": 355
    },
    {
      "epoch": 0.1855737277643839,
      "grad_norm": 1.0432439733778767,
      "learning_rate": 4.752956272274078e-06,
      "loss": 0.2211,
      "step": 356
    },
    {
      "epoch": 0.18609500228057602,
      "grad_norm": 1.1900272807021643,
      "learning_rate": 4.751084522110669e-06,
      "loss": 0.2225,
      "step": 357
    },
    {
      "epoch": 0.1866162767967681,
      "grad_norm": 1.074547857544035,
      "learning_rate": 4.749206079349952e-06,
      "loss": 0.2116,
      "step": 358
    },
    {
      "epoch": 0.1871375513129602,
      "grad_norm": 1.033594905483247,
      "learning_rate": 4.747320949576641e-06,
      "loss": 0.2142,
      "step": 359
    },
    {
      "epoch": 0.18765882582915228,
      "grad_norm": 1.1255484030942537,
      "learning_rate": 4.745429138395329e-06,
      "loss": 0.2107,
      "step": 360
    },
    {
      "epoch": 0.18818010034534435,
      "grad_norm": 1.1208575114062105,
      "learning_rate": 4.743530651430472e-06,
      "loss": 0.2199,
      "step": 361
    },
    {
      "epoch": 0.18870137486153646,
      "grad_norm": 1.1460239450962784,
      "learning_rate": 4.741625494326379e-06,
      "loss": 0.2232,
      "step": 362
    },
    {
      "epoch": 0.18922264937772854,
      "grad_norm": 1.1761939886658683,
      "learning_rate": 4.739713672747183e-06,
      "loss": 0.2193,
      "step": 363
    },
    {
      "epoch": 0.18974392389392064,
      "grad_norm": 1.19826819689747,
      "learning_rate": 4.737795192376836e-06,
      "loss": 0.226,
      "step": 364
    },
    {
      "epoch": 0.19026519841011272,
      "grad_norm": 1.103789018784415,
      "learning_rate": 4.735870058919084e-06,
      "loss": 0.2339,
      "step": 365
    },
    {
      "epoch": 0.19078647292630482,
      "grad_norm": 1.0832407294802735,
      "learning_rate": 4.733938278097456e-06,
      "loss": 0.2393,
      "step": 366
    },
    {
      "epoch": 0.1913077474424969,
      "grad_norm": 1.1848088227040534,
      "learning_rate": 4.731999855655239e-06,
      "loss": 0.224,
      "step": 367
    },
    {
      "epoch": 0.191829021958689,
      "grad_norm": 1.1838392460670388,
      "learning_rate": 4.730054797355471e-06,
      "loss": 0.2448,
      "step": 368
    },
    {
      "epoch": 0.19235029647488108,
      "grad_norm": 1.0617393105525352,
      "learning_rate": 4.728103108980915e-06,
      "loss": 0.2051,
      "step": 369
    },
    {
      "epoch": 0.19287157099107316,
      "grad_norm": 1.0844815033870783,
      "learning_rate": 4.726144796334049e-06,
      "loss": 0.2139,
      "step": 370
    },
    {
      "epoch": 0.19339284550726527,
      "grad_norm": 1.1834140709430048,
      "learning_rate": 4.724179865237042e-06,
      "loss": 0.2288,
      "step": 371
    },
    {
      "epoch": 0.19391412002345734,
      "grad_norm": 1.109820726429485,
      "learning_rate": 4.722208321531743e-06,
      "loss": 0.2107,
      "step": 372
    },
    {
      "epoch": 0.19443539453964945,
      "grad_norm": 1.1351904940122715,
      "learning_rate": 4.720230171079657e-06,
      "loss": 0.2192,
      "step": 373
    },
    {
      "epoch": 0.19495666905584152,
      "grad_norm": 1.2167214965769027,
      "learning_rate": 4.7182454197619355e-06,
      "loss": 0.2256,
      "step": 374
    },
    {
      "epoch": 0.19547794357203363,
      "grad_norm": 1.1604275914972113,
      "learning_rate": 4.716254073479352e-06,
      "loss": 0.2274,
      "step": 375
    },
    {
      "epoch": 0.1959992180882257,
      "grad_norm": 1.1426123348996993,
      "learning_rate": 4.714256138152287e-06,
      "loss": 0.2145,
      "step": 376
    },
    {
      "epoch": 0.1965204926044178,
      "grad_norm": 1.1541949632639865,
      "learning_rate": 4.712251619720712e-06,
      "loss": 0.2236,
      "step": 377
    },
    {
      "epoch": 0.1970417671206099,
      "grad_norm": 1.1680074383865588,
      "learning_rate": 4.71024052414417e-06,
      "loss": 0.224,
      "step": 378
    },
    {
      "epoch": 0.19756304163680197,
      "grad_norm": 1.093211547348792,
      "learning_rate": 4.70822285740176e-06,
      "loss": 0.2167,
      "step": 379
    },
    {
      "epoch": 0.19808431615299407,
      "grad_norm": 1.0429291132552085,
      "learning_rate": 4.706198625492111e-06,
      "loss": 0.225,
      "step": 380
    },
    {
      "epoch": 0.19860559066918615,
      "grad_norm": 1.1437201685657041,
      "learning_rate": 4.704167834433378e-06,
      "loss": 0.2154,
      "step": 381
    },
    {
      "epoch": 0.19912686518537825,
      "grad_norm": 1.0255764432733163,
      "learning_rate": 4.702130490263215e-06,
      "loss": 0.2211,
      "step": 382
    },
    {
      "epoch": 0.19964813970157033,
      "grad_norm": 1.079528796981029,
      "learning_rate": 4.700086599038755e-06,
      "loss": 0.2067,
      "step": 383
    },
    {
      "epoch": 0.20016941421776244,
      "grad_norm": 1.1198477995083633,
      "learning_rate": 4.698036166836598e-06,
      "loss": 0.2325,
      "step": 384
    },
    {
      "epoch": 0.20069068873395451,
      "grad_norm": 0.9961051646512427,
      "learning_rate": 4.695979199752794e-06,
      "loss": 0.2095,
      "step": 385
    },
    {
      "epoch": 0.20121196325014662,
      "grad_norm": 1.115542929205913,
      "learning_rate": 4.693915703902816e-06,
      "loss": 0.2222,
      "step": 386
    },
    {
      "epoch": 0.2017332377663387,
      "grad_norm": 1.0607550045367906,
      "learning_rate": 4.691845685421551e-06,
      "loss": 0.2103,
      "step": 387
    },
    {
      "epoch": 0.20225451228253077,
      "grad_norm": 1.0383685809934824,
      "learning_rate": 4.689769150463277e-06,
      "loss": 0.2196,
      "step": 388
    },
    {
      "epoch": 0.20277578679872288,
      "grad_norm": 1.1801982817474617,
      "learning_rate": 4.687686105201645e-06,
      "loss": 0.226,
      "step": 389
    },
    {
      "epoch": 0.20329706131491496,
      "grad_norm": 1.1154931717291245,
      "learning_rate": 4.685596555829664e-06,
      "loss": 0.2047,
      "step": 390
    },
    {
      "epoch": 0.20381833583110706,
      "grad_norm": 1.032527412329601,
      "learning_rate": 4.683500508559676e-06,
      "loss": 0.2083,
      "step": 391
    },
    {
      "epoch": 0.20433961034729914,
      "grad_norm": 1.2019006510748693,
      "learning_rate": 4.681397969623347e-06,
      "loss": 0.2065,
      "step": 392
    },
    {
      "epoch": 0.20486088486349124,
      "grad_norm": 1.1978210836606196,
      "learning_rate": 4.679288945271639e-06,
      "loss": 0.2444,
      "step": 393
    },
    {
      "epoch": 0.20538215937968332,
      "grad_norm": 1.0218474068950976,
      "learning_rate": 4.677173441774796e-06,
      "loss": 0.2236,
      "step": 394
    },
    {
      "epoch": 0.20590343389587543,
      "grad_norm": 1.1665523253520813,
      "learning_rate": 4.675051465422326e-06,
      "loss": 0.2337,
      "step": 395
    },
    {
      "epoch": 0.2064247084120675,
      "grad_norm": 1.132284782223571,
      "learning_rate": 4.6729230225229815e-06,
      "loss": 0.2195,
      "step": 396
    },
    {
      "epoch": 0.20694598292825958,
      "grad_norm": 1.193184872531427,
      "learning_rate": 4.670788119404739e-06,
      "loss": 0.2287,
      "step": 397
    },
    {
      "epoch": 0.20746725744445169,
      "grad_norm": 1.049891267841484,
      "learning_rate": 4.66864676241478e-06,
      "loss": 0.2071,
      "step": 398
    },
    {
      "epoch": 0.20798853196064376,
      "grad_norm": 1.0641364325724467,
      "learning_rate": 4.666498957919479e-06,
      "loss": 0.2168,
      "step": 399
    },
    {
      "epoch": 0.20850980647683587,
      "grad_norm": 1.0981712069511067,
      "learning_rate": 4.664344712304375e-06,
      "loss": 0.2176,
      "step": 400
    },
    {
      "epoch": 0.20903108099302795,
      "grad_norm": 1.1645379745244755,
      "learning_rate": 4.6621840319741576e-06,
      "loss": 0.2306,
      "step": 401
    },
    {
      "epoch": 0.20955235550922005,
      "grad_norm": 1.0935765067331227,
      "learning_rate": 4.660016923352648e-06,
      "loss": 0.219,
      "step": 402
    },
    {
      "epoch": 0.21007363002541213,
      "grad_norm": 1.0162370357327695,
      "learning_rate": 4.657843392882778e-06,
      "loss": 0.2164,
      "step": 403
    },
    {
      "epoch": 0.21059490454160423,
      "grad_norm": 1.0685832938670983,
      "learning_rate": 4.6556634470265725e-06,
      "loss": 0.2082,
      "step": 404
    },
    {
      "epoch": 0.2111161790577963,
      "grad_norm": 1.1207863558203524,
      "learning_rate": 4.6534770922651305e-06,
      "loss": 0.2209,
      "step": 405
    },
    {
      "epoch": 0.21163745357398842,
      "grad_norm": 1.2696785852584918,
      "learning_rate": 4.651284335098603e-06,
      "loss": 0.2335,
      "step": 406
    },
    {
      "epoch": 0.2121587280901805,
      "grad_norm": 1.1354128670205956,
      "learning_rate": 4.6490851820461785e-06,
      "loss": 0.2224,
      "step": 407
    },
    {
      "epoch": 0.21268000260637257,
      "grad_norm": 1.0975319534286776,
      "learning_rate": 4.646879639646058e-06,
      "loss": 0.212,
      "step": 408
    },
    {
      "epoch": 0.21320127712256468,
      "grad_norm": 1.095184660583565,
      "learning_rate": 4.64466771445544e-06,
      "loss": 0.2136,
      "step": 409
    },
    {
      "epoch": 0.21372255163875675,
      "grad_norm": 1.064612547677763,
      "learning_rate": 4.642449413050499e-06,
      "loss": 0.1937,
      "step": 410
    },
    {
      "epoch": 0.21424382615494886,
      "grad_norm": 1.1115362759791394,
      "learning_rate": 4.640224742026365e-06,
      "loss": 0.2254,
      "step": 411
    },
    {
      "epoch": 0.21476510067114093,
      "grad_norm": 1.0348862046404177,
      "learning_rate": 4.637993707997107e-06,
      "loss": 0.2112,
      "step": 412
    },
    {
      "epoch": 0.21528637518733304,
      "grad_norm": 1.0818172703394755,
      "learning_rate": 4.635756317595714e-06,
      "loss": 0.206,
      "step": 413
    },
    {
      "epoch": 0.21580764970352512,
      "grad_norm": 1.0260926284540925,
      "learning_rate": 4.6335125774740665e-06,
      "loss": 0.2076,
      "step": 414
    },
    {
      "epoch": 0.21632892421971722,
      "grad_norm": 1.0429305974716216,
      "learning_rate": 4.6312624943029275e-06,
      "loss": 0.2177,
      "step": 415
    },
    {
      "epoch": 0.2168501987359093,
      "grad_norm": 1.1280018957375995,
      "learning_rate": 4.629006074771918e-06,
      "loss": 0.2373,
      "step": 416
    },
    {
      "epoch": 0.21737147325210138,
      "grad_norm": 1.1275683844242368,
      "learning_rate": 4.626743325589496e-06,
      "loss": 0.211,
      "step": 417
    },
    {
      "epoch": 0.21789274776829348,
      "grad_norm": 1.07286210999913,
      "learning_rate": 4.624474253482938e-06,
      "loss": 0.2225,
      "step": 418
    },
    {
      "epoch": 0.21841402228448556,
      "grad_norm": 1.097889164710293,
      "learning_rate": 4.622198865198321e-06,
      "loss": 0.2151,
      "step": 419
    },
    {
      "epoch": 0.21893529680067766,
      "grad_norm": 1.1004095694814584,
      "learning_rate": 4.619917167500496e-06,
      "loss": 0.2089,
      "step": 420
    },
    {
      "epoch": 0.21945657131686974,
      "grad_norm": 1.1048350999571594,
      "learning_rate": 4.617629167173078e-06,
      "loss": 0.2266,
      "step": 421
    },
    {
      "epoch": 0.21997784583306185,
      "grad_norm": 1.0912873499468907,
      "learning_rate": 4.615334871018415e-06,
      "loss": 0.231,
      "step": 422
    },
    {
      "epoch": 0.22049912034925392,
      "grad_norm": 1.0408222898191029,
      "learning_rate": 4.613034285857575e-06,
      "loss": 0.1894,
      "step": 423
    },
    {
      "epoch": 0.22102039486544603,
      "grad_norm": 1.1377530366999036,
      "learning_rate": 4.610727418530324e-06,
      "loss": 0.2273,
      "step": 424
    },
    {
      "epoch": 0.2215416693816381,
      "grad_norm": 1.070826056404283,
      "learning_rate": 4.6084142758951055e-06,
      "loss": 0.2196,
      "step": 425
    },
    {
      "epoch": 0.22206294389783018,
      "grad_norm": 1.1473399605841177,
      "learning_rate": 4.606094864829016e-06,
      "loss": 0.2109,
      "step": 426
    },
    {
      "epoch": 0.2225842184140223,
      "grad_norm": 1.1432744764899827,
      "learning_rate": 4.603769192227795e-06,
      "loss": 0.2213,
      "step": 427
    },
    {
      "epoch": 0.22310549293021437,
      "grad_norm": 1.072285539296878,
      "learning_rate": 4.601437265005792e-06,
      "loss": 0.2166,
      "step": 428
    },
    {
      "epoch": 0.22362676744640647,
      "grad_norm": 1.1618829602109915,
      "learning_rate": 4.599099090095955e-06,
      "loss": 0.207,
      "step": 429
    },
    {
      "epoch": 0.22414804196259855,
      "grad_norm": 1.2500972677894722,
      "learning_rate": 4.5967546744498044e-06,
      "loss": 0.2193,
      "step": 430
    },
    {
      "epoch": 0.22466931647879065,
      "grad_norm": 1.1657729036776328,
      "learning_rate": 4.594404025037418e-06,
      "loss": 0.2113,
      "step": 431
    },
    {
      "epoch": 0.22519059099498273,
      "grad_norm": 1.249964132485446,
      "learning_rate": 4.592047148847404e-06,
      "loss": 0.2256,
      "step": 432
    },
    {
      "epoch": 0.22571186551117484,
      "grad_norm": 1.1644116306373535,
      "learning_rate": 4.589684052886884e-06,
      "loss": 0.2198,
      "step": 433
    },
    {
      "epoch": 0.2262331400273669,
      "grad_norm": 1.2065528009787465,
      "learning_rate": 4.587314744181471e-06,
      "loss": 0.231,
      "step": 434
    },
    {
      "epoch": 0.226754414543559,
      "grad_norm": 1.143504671323367,
      "learning_rate": 4.58493922977525e-06,
      "loss": 0.209,
      "step": 435
    },
    {
      "epoch": 0.2272756890597511,
      "grad_norm": 1.0472266892118371,
      "learning_rate": 4.582557516730755e-06,
      "loss": 0.2072,
      "step": 436
    },
    {
      "epoch": 0.22779696357594317,
      "grad_norm": 1.0361215331977902,
      "learning_rate": 4.58016961212895e-06,
      "loss": 0.196,
      "step": 437
    },
    {
      "epoch": 0.22831823809213528,
      "grad_norm": 1.1481051690496913,
      "learning_rate": 4.577775523069204e-06,
      "loss": 0.2277,
      "step": 438
    },
    {
      "epoch": 0.22883951260832736,
      "grad_norm": 1.2547761237172292,
      "learning_rate": 4.575375256669276e-06,
      "loss": 0.2407,
      "step": 439
    },
    {
      "epoch": 0.22936078712451946,
      "grad_norm": 0.9650748955192284,
      "learning_rate": 4.572968820065288e-06,
      "loss": 0.1848,
      "step": 440
    },
    {
      "epoch": 0.22988206164071154,
      "grad_norm": 1.1719657520615745,
      "learning_rate": 4.570556220411708e-06,
      "loss": 0.2163,
      "step": 441
    },
    {
      "epoch": 0.23040333615690364,
      "grad_norm": 1.2562943824109747,
      "learning_rate": 4.568137464881328e-06,
      "loss": 0.2093,
      "step": 442
    },
    {
      "epoch": 0.23092461067309572,
      "grad_norm": 1.0944459906859987,
      "learning_rate": 4.5657125606652385e-06,
      "loss": 0.2132,
      "step": 443
    },
    {
      "epoch": 0.2314458851892878,
      "grad_norm": 1.2155693533167746,
      "learning_rate": 4.563281514972814e-06,
      "loss": 0.2262,
      "step": 444
    },
    {
      "epoch": 0.2319671597054799,
      "grad_norm": 1.2337605671295995,
      "learning_rate": 4.560844335031684e-06,
      "loss": 0.2404,
      "step": 445
    },
    {
      "epoch": 0.23248843422167198,
      "grad_norm": 1.1757287729273904,
      "learning_rate": 4.55840102808772e-06,
      "loss": 0.2343,
      "step": 446
    },
    {
      "epoch": 0.23300970873786409,
      "grad_norm": 1.1150156127039372,
      "learning_rate": 4.555951601405005e-06,
      "loss": 0.2063,
      "step": 447
    },
    {
      "epoch": 0.23353098325405616,
      "grad_norm": 1.1301475531038054,
      "learning_rate": 4.55349606226582e-06,
      "loss": 0.2202,
      "step": 448
    },
    {
      "epoch": 0.23405225777024827,
      "grad_norm": 1.1388989584706561,
      "learning_rate": 4.551034417970616e-06,
      "loss": 0.2324,
      "step": 449
    },
    {
      "epoch": 0.23457353228644034,
      "grad_norm": 1.0575247489861925,
      "learning_rate": 4.548566675837996e-06,
      "loss": 0.2186,
      "step": 450
    },
    {
      "epoch": 0.23509480680263245,
      "grad_norm": 1.0204052134233565,
      "learning_rate": 4.546092843204694e-06,
      "loss": 0.2101,
      "step": 451
    },
    {
      "epoch": 0.23561608131882453,
      "grad_norm": 1.2251080937053365,
      "learning_rate": 4.543612927425547e-06,
      "loss": 0.2248,
      "step": 452
    },
    {
      "epoch": 0.2361373558350166,
      "grad_norm": 1.1642657736807371,
      "learning_rate": 4.541126935873481e-06,
      "loss": 0.2082,
      "step": 453
    },
    {
      "epoch": 0.2366586303512087,
      "grad_norm": 1.0770735004827978,
      "learning_rate": 4.538634875939486e-06,
      "loss": 0.2114,
      "step": 454
    },
    {
      "epoch": 0.2371799048674008,
      "grad_norm": 1.1584724096430439,
      "learning_rate": 4.536136755032592e-06,
      "loss": 0.2133,
      "step": 455
    },
    {
      "epoch": 0.2377011793835929,
      "grad_norm": 1.1493910338394913,
      "learning_rate": 4.5336325805798475e-06,
      "loss": 0.214,
      "step": 456
    },
    {
      "epoch": 0.23822245389978497,
      "grad_norm": 0.9889705744340543,
      "learning_rate": 4.5311223600263016e-06,
      "loss": 0.2074,
      "step": 457
    },
    {
      "epoch": 0.23874372841597707,
      "grad_norm": 1.0191092901261611,
      "learning_rate": 4.528606100834976e-06,
      "loss": 0.208,
      "step": 458
    },
    {
      "epoch": 0.23926500293216915,
      "grad_norm": 1.0516379799007636,
      "learning_rate": 4.526083810486848e-06,
      "loss": 0.2108,
      "step": 459
    },
    {
      "epoch": 0.23978627744836126,
      "grad_norm": 1.1330898073365478,
      "learning_rate": 4.523555496480824e-06,
      "loss": 0.2126,
      "step": 460
    },
    {
      "epoch": 0.24030755196455333,
      "grad_norm": 1.0796213356859856,
      "learning_rate": 4.5210211663337195e-06,
      "loss": 0.2218,
      "step": 461
    },
    {
      "epoch": 0.2408288264807454,
      "grad_norm": 1.0355251655845654,
      "learning_rate": 4.518480827580237e-06,
      "loss": 0.2115,
      "step": 462
    },
    {
      "epoch": 0.24135010099693752,
      "grad_norm": 1.153573794271069,
      "learning_rate": 4.515934487772942e-06,
      "loss": 0.2142,
      "step": 463
    },
    {
      "epoch": 0.2418713755131296,
      "grad_norm": 0.9815120251789137,
      "learning_rate": 4.513382154482242e-06,
      "loss": 0.198,
      "step": 464
    },
    {
      "epoch": 0.2423926500293217,
      "grad_norm": 1.1495741092058984,
      "learning_rate": 4.510823835296364e-06,
      "loss": 0.2272,
      "step": 465
    },
    {
      "epoch": 0.24291392454551378,
      "grad_norm": 1.102910606321348,
      "learning_rate": 4.50825953782133e-06,
      "loss": 0.2001,
      "step": 466
    },
    {
      "epoch": 0.24343519906170588,
      "grad_norm": 1.0645645708742124,
      "learning_rate": 4.505689269680937e-06,
      "loss": 0.2124,
      "step": 467
    },
    {
      "epoch": 0.24395647357789796,
      "grad_norm": 1.0189195608220585,
      "learning_rate": 4.503113038516732e-06,
      "loss": 0.2126,
      "step": 468
    },
    {
      "epoch": 0.24447774809409006,
      "grad_norm": 1.067814056299296,
      "learning_rate": 4.500530851987992e-06,
      "loss": 0.2225,
      "step": 469
    },
    {
      "epoch": 0.24499902261028214,
      "grad_norm": 1.061499241529969,
      "learning_rate": 4.4979427177716974e-06,
      "loss": 0.1993,
      "step": 470
    },
    {
      "epoch": 0.24552029712647422,
      "grad_norm": 1.0402414395794188,
      "learning_rate": 4.495348643562514e-06,
      "loss": 0.209,
      "step": 471
    },
    {
      "epoch": 0.24604157164266632,
      "grad_norm": 1.0418834782414723,
      "learning_rate": 4.4927486370727656e-06,
      "loss": 0.2205,
      "step": 472
    },
    {
      "epoch": 0.2465628461588584,
      "grad_norm": 1.0972428443151883,
      "learning_rate": 4.4901427060324135e-06,
      "loss": 0.2306,
      "step": 473
    },
    {
      "epoch": 0.2470841206750505,
      "grad_norm": 1.1155839876556997,
      "learning_rate": 4.487530858189033e-06,
      "loss": 0.2187,
      "step": 474
    },
    {
      "epoch": 0.24760539519124258,
      "grad_norm": 1.1300995264380476,
      "learning_rate": 4.4849131013077915e-06,
      "loss": 0.2314,
      "step": 475
    },
    {
      "epoch": 0.2481266697074347,
      "grad_norm": 1.028559227898691,
      "learning_rate": 4.482289443171421e-06,
      "loss": 0.2175,
      "step": 476
    },
    {
      "epoch": 0.24864794422362677,
      "grad_norm": 1.133823139878735,
      "learning_rate": 4.479659891580203e-06,
      "loss": 0.2151,
      "step": 477
    },
    {
      "epoch": 0.24916921873981887,
      "grad_norm": 1.1948210325504214,
      "learning_rate": 4.477024454351937e-06,
      "loss": 0.2243,
      "step": 478
    },
    {
      "epoch": 0.24969049325601095,
      "grad_norm": 1.1155669981436795,
      "learning_rate": 4.4743831393219215e-06,
      "loss": 0.2246,
      "step": 479
    },
    {
      "epoch": 0.25021176777220305,
      "grad_norm": 1.0870863630319363,
      "learning_rate": 4.471735954342932e-06,
      "loss": 0.2194,
      "step": 480
    },
    {
      "epoch": 0.2507330422883951,
      "grad_norm": 1.021947377733924,
      "learning_rate": 4.469082907285192e-06,
      "loss": 0.214,
      "step": 481
    },
    {
      "epoch": 0.2512543168045872,
      "grad_norm": 1.0696327117947075,
      "learning_rate": 4.4664240060363565e-06,
      "loss": 0.2195,
      "step": 482
    },
    {
      "epoch": 0.2517755913207793,
      "grad_norm": 1.080457437808401,
      "learning_rate": 4.463759258501485e-06,
      "loss": 0.2058,
      "step": 483
    },
    {
      "epoch": 0.2522968658369714,
      "grad_norm": 1.165235624176743,
      "learning_rate": 4.461088672603015e-06,
      "loss": 0.2198,
      "step": 484
    },
    {
      "epoch": 0.25281814035316347,
      "grad_norm": 1.1277534771234747,
      "learning_rate": 4.458412256280747e-06,
      "loss": 0.225,
      "step": 485
    },
    {
      "epoch": 0.2533394148693556,
      "grad_norm": 1.0599584663873574,
      "learning_rate": 4.455730017491812e-06,
      "loss": 0.2006,
      "step": 486
    },
    {
      "epoch": 0.2538606893855477,
      "grad_norm": 1.0249989199079323,
      "learning_rate": 4.453041964210653e-06,
      "loss": 0.2041,
      "step": 487
    },
    {
      "epoch": 0.2543819639017397,
      "grad_norm": 1.0814394140585897,
      "learning_rate": 4.450348104428998e-06,
      "loss": 0.2215,
      "step": 488
    },
    {
      "epoch": 0.25490323841793183,
      "grad_norm": 1.0858602425123827,
      "learning_rate": 4.447648446155841e-06,
      "loss": 0.222,
      "step": 489
    },
    {
      "epoch": 0.25542451293412394,
      "grad_norm": 1.0278147977821745,
      "learning_rate": 4.4449429974174115e-06,
      "loss": 0.2114,
      "step": 490
    },
    {
      "epoch": 0.25594578745031604,
      "grad_norm": 0.9653873167735169,
      "learning_rate": 4.442231766257159e-06,
      "loss": 0.1928,
      "step": 491
    },
    {
      "epoch": 0.2564670619665081,
      "grad_norm": 1.0141617541982602,
      "learning_rate": 4.43951476073572e-06,
      "loss": 0.2,
      "step": 492
    },
    {
      "epoch": 0.2569883364827002,
      "grad_norm": 1.0782517271477747,
      "learning_rate": 4.436791988930901e-06,
      "loss": 0.2182,
      "step": 493
    },
    {
      "epoch": 0.2575096109988923,
      "grad_norm": 1.0883211480385289,
      "learning_rate": 4.434063458937652e-06,
      "loss": 0.2096,
      "step": 494
    },
    {
      "epoch": 0.2580308855150844,
      "grad_norm": 1.0827340273369845,
      "learning_rate": 4.43132917886804e-06,
      "loss": 0.2257,
      "step": 495
    },
    {
      "epoch": 0.25855216003127646,
      "grad_norm": 1.0410081603932166,
      "learning_rate": 4.428589156851231e-06,
      "loss": 0.2181,
      "step": 496
    },
    {
      "epoch": 0.25907343454746856,
      "grad_norm": 1.0541847604693613,
      "learning_rate": 4.42584340103346e-06,
      "loss": 0.2162,
      "step": 497
    },
    {
      "epoch": 0.25959470906366067,
      "grad_norm": 1.143645122965762,
      "learning_rate": 4.423091919578008e-06,
      "loss": 0.2301,
      "step": 498
    },
    {
      "epoch": 0.2601159835798527,
      "grad_norm": 1.0639180365437577,
      "learning_rate": 4.4203347206651805e-06,
      "loss": 0.2355,
      "step": 499
    },
    {
      "epoch": 0.2606372580960448,
      "grad_norm": 0.9614473295811211,
      "learning_rate": 4.417571812492279e-06,
      "loss": 0.1954,
      "step": 500
    },
    {
      "epoch": 0.2606372580960448,
      "eval_loss": 0.21339768171310425,
      "eval_runtime": 52.4959,
      "eval_samples_per_second": 23.64,
      "eval_steps_per_second": 2.972,
      "step": 500
    },
    {
      "epoch": 0.2611585326122369,
      "grad_norm": 1.0841579950020523,
      "learning_rate": 4.4148032032735835e-06,
      "loss": 0.2233,
      "step": 501
    },
    {
      "epoch": 0.26167980712842903,
      "grad_norm": 1.0055568844787404,
      "learning_rate": 4.4120289012403185e-06,
      "loss": 0.205,
      "step": 502
    },
    {
      "epoch": 0.2622010816446211,
      "grad_norm": 1.0687056061935238,
      "learning_rate": 4.409248914640636e-06,
      "loss": 0.2156,
      "step": 503
    },
    {
      "epoch": 0.2627223561608132,
      "grad_norm": 1.0771229976941505,
      "learning_rate": 4.4064632517395875e-06,
      "loss": 0.2247,
      "step": 504
    },
    {
      "epoch": 0.2632436306770053,
      "grad_norm": 1.0481049423895688,
      "learning_rate": 4.4036719208191025e-06,
      "loss": 0.2234,
      "step": 505
    },
    {
      "epoch": 0.26376490519319734,
      "grad_norm": 1.162506231563762,
      "learning_rate": 4.400874930177959e-06,
      "loss": 0.2219,
      "step": 506
    },
    {
      "epoch": 0.26428617970938945,
      "grad_norm": 1.0912372610620908,
      "learning_rate": 4.398072288131763e-06,
      "loss": 0.1922,
      "step": 507
    },
    {
      "epoch": 0.26480745422558155,
      "grad_norm": 1.0867839888631574,
      "learning_rate": 4.395264003012924e-06,
      "loss": 0.2175,
      "step": 508
    },
    {
      "epoch": 0.26532872874177366,
      "grad_norm": 1.1805618069379247,
      "learning_rate": 4.392450083170625e-06,
      "loss": 0.218,
      "step": 509
    },
    {
      "epoch": 0.2658500032579657,
      "grad_norm": 1.0889255347942146,
      "learning_rate": 4.389630536970806e-06,
      "loss": 0.2064,
      "step": 510
    },
    {
      "epoch": 0.2663712777741578,
      "grad_norm": 1.1116894634969319,
      "learning_rate": 4.386805372796129e-06,
      "loss": 0.2181,
      "step": 511
    },
    {
      "epoch": 0.2668925522903499,
      "grad_norm": 1.0586197466332472,
      "learning_rate": 4.383974599045963e-06,
      "loss": 0.2093,
      "step": 512
    },
    {
      "epoch": 0.267413826806542,
      "grad_norm": 1.0145935887387179,
      "learning_rate": 4.3811382241363545e-06,
      "loss": 0.2113,
      "step": 513
    },
    {
      "epoch": 0.26793510132273407,
      "grad_norm": 1.0435500323197133,
      "learning_rate": 4.378296256499998e-06,
      "loss": 0.2103,
      "step": 514
    },
    {
      "epoch": 0.2684563758389262,
      "grad_norm": 1.0479347125521619,
      "learning_rate": 4.375448704586221e-06,
      "loss": 0.208,
      "step": 515
    },
    {
      "epoch": 0.2689776503551183,
      "grad_norm": 1.0710861477938967,
      "learning_rate": 4.37259557686095e-06,
      "loss": 0.2001,
      "step": 516
    },
    {
      "epoch": 0.26949892487131033,
      "grad_norm": 1.0438111670635168,
      "learning_rate": 4.369736881806691e-06,
      "loss": 0.2155,
      "step": 517
    },
    {
      "epoch": 0.27002019938750244,
      "grad_norm": 1.071314827976162,
      "learning_rate": 4.366872627922498e-06,
      "loss": 0.2049,
      "step": 518
    },
    {
      "epoch": 0.27054147390369454,
      "grad_norm": 1.1105126190630343,
      "learning_rate": 4.364002823723956e-06,
      "loss": 0.2161,
      "step": 519
    },
    {
      "epoch": 0.27106274841988665,
      "grad_norm": 1.1128881857208017,
      "learning_rate": 4.36112747774315e-06,
      "loss": 0.2178,
      "step": 520
    },
    {
      "epoch": 0.2715840229360787,
      "grad_norm": 1.127620184184549,
      "learning_rate": 4.358246598528641e-06,
      "loss": 0.2008,
      "step": 521
    },
    {
      "epoch": 0.2721052974522708,
      "grad_norm": 1.10004250359709,
      "learning_rate": 4.355360194645439e-06,
      "loss": 0.1985,
      "step": 522
    },
    {
      "epoch": 0.2726265719684629,
      "grad_norm": 1.0272279568989093,
      "learning_rate": 4.35246827467498e-06,
      "loss": 0.2138,
      "step": 523
    },
    {
      "epoch": 0.27314784648465495,
      "grad_norm": 1.106632586603704,
      "learning_rate": 4.349570847215104e-06,
      "loss": 0.2155,
      "step": 524
    },
    {
      "epoch": 0.27366912100084706,
      "grad_norm": 1.194736618683367,
      "learning_rate": 4.346667920880016e-06,
      "loss": 0.2197,
      "step": 525
    },
    {
      "epoch": 0.27419039551703916,
      "grad_norm": 1.0801091301725214,
      "learning_rate": 4.343759504300278e-06,
      "loss": 0.2074,
      "step": 526
    },
    {
      "epoch": 0.27471167003323127,
      "grad_norm": 1.0614194685397207,
      "learning_rate": 4.34084560612277e-06,
      "loss": 0.2136,
      "step": 527
    },
    {
      "epoch": 0.2752329445494233,
      "grad_norm": 1.0531834029204565,
      "learning_rate": 4.337926235010672e-06,
      "loss": 0.2053,
      "step": 528
    },
    {
      "epoch": 0.2757542190656154,
      "grad_norm": 1.1801906881402193,
      "learning_rate": 4.335001399643433e-06,
      "loss": 0.2194,
      "step": 529
    },
    {
      "epoch": 0.27627549358180753,
      "grad_norm": 1.0644262272232463,
      "learning_rate": 4.332071108716747e-06,
      "loss": 0.211,
      "step": 530
    },
    {
      "epoch": 0.27679676809799963,
      "grad_norm": 1.0440684724636122,
      "learning_rate": 4.329135370942531e-06,
      "loss": 0.2066,
      "step": 531
    },
    {
      "epoch": 0.2773180426141917,
      "grad_norm": 1.0480131480389971,
      "learning_rate": 4.326194195048894e-06,
      "loss": 0.2125,
      "step": 532
    },
    {
      "epoch": 0.2778393171303838,
      "grad_norm": 1.0574201215221624,
      "learning_rate": 4.323247589780111e-06,
      "loss": 0.2073,
      "step": 533
    },
    {
      "epoch": 0.2783605916465759,
      "grad_norm": 1.2097127844915045,
      "learning_rate": 4.320295563896601e-06,
      "loss": 0.2035,
      "step": 534
    },
    {
      "epoch": 0.27888186616276794,
      "grad_norm": 1.040708567287013,
      "learning_rate": 4.317338126174899e-06,
      "loss": 0.2262,
      "step": 535
    },
    {
      "epoch": 0.27940314067896005,
      "grad_norm": 1.0670609705734364,
      "learning_rate": 4.314375285407629e-06,
      "loss": 0.2178,
      "step": 536
    },
    {
      "epoch": 0.27992441519515215,
      "grad_norm": 1.0480234772110812,
      "learning_rate": 4.311407050403479e-06,
      "loss": 0.1976,
      "step": 537
    },
    {
      "epoch": 0.28044568971134426,
      "grad_norm": 1.0042327964128128,
      "learning_rate": 4.308433429987172e-06,
      "loss": 0.2111,
      "step": 538
    },
    {
      "epoch": 0.2809669642275363,
      "grad_norm": 1.138894159548414,
      "learning_rate": 4.305454432999445e-06,
      "loss": 0.2085,
      "step": 539
    },
    {
      "epoch": 0.2814882387437284,
      "grad_norm": 1.0516652868157126,
      "learning_rate": 4.302470068297019e-06,
      "loss": 0.2132,
      "step": 540
    },
    {
      "epoch": 0.2820095132599205,
      "grad_norm": 1.1303509110253085,
      "learning_rate": 4.2994803447525735e-06,
      "loss": 0.2119,
      "step": 541
    },
    {
      "epoch": 0.28253078777611257,
      "grad_norm": 0.990081843922128,
      "learning_rate": 4.29648527125472e-06,
      "loss": 0.1951,
      "step": 542
    },
    {
      "epoch": 0.2830520622923047,
      "grad_norm": 0.9709694070302939,
      "learning_rate": 4.2934848567079745e-06,
      "loss": 0.2034,
      "step": 543
    },
    {
      "epoch": 0.2835733368084968,
      "grad_norm": 1.1024012313744909,
      "learning_rate": 4.290479110032735e-06,
      "loss": 0.2195,
      "step": 544
    },
    {
      "epoch": 0.2840946113246889,
      "grad_norm": 0.9752023861970188,
      "learning_rate": 4.28746804016525e-06,
      "loss": 0.2013,
      "step": 545
    },
    {
      "epoch": 0.28461588584088093,
      "grad_norm": 1.051607491477118,
      "learning_rate": 4.284451656057595e-06,
      "loss": 0.2244,
      "step": 546
    },
    {
      "epoch": 0.28513716035707304,
      "grad_norm": 1.0474182907791254,
      "learning_rate": 4.281429966677644e-06,
      "loss": 0.2199,
      "step": 547
    },
    {
      "epoch": 0.28565843487326514,
      "grad_norm": 1.0251588658877104,
      "learning_rate": 4.2784029810090456e-06,
      "loss": 0.2042,
      "step": 548
    },
    {
      "epoch": 0.28617970938945725,
      "grad_norm": 1.046698681370147,
      "learning_rate": 4.275370708051194e-06,
      "loss": 0.2119,
      "step": 549
    },
    {
      "epoch": 0.2867009839056493,
      "grad_norm": 1.0824173335506142,
      "learning_rate": 4.2723331568192004e-06,
      "loss": 0.2214,
      "step": 550
    },
    {
      "epoch": 0.2872222584218414,
      "grad_norm": 1.0746888933028043,
      "learning_rate": 4.269290336343873e-06,
      "loss": 0.2169,
      "step": 551
    },
    {
      "epoch": 0.2877435329380335,
      "grad_norm": 1.0756144814428263,
      "learning_rate": 4.266242255671681e-06,
      "loss": 0.2117,
      "step": 552
    },
    {
      "epoch": 0.28826480745422556,
      "grad_norm": 1.0690384595853772,
      "learning_rate": 4.2631889238647375e-06,
      "loss": 0.2108,
      "step": 553
    },
    {
      "epoch": 0.28878608197041766,
      "grad_norm": 1.0994939303823907,
      "learning_rate": 4.260130350000763e-06,
      "loss": 0.2121,
      "step": 554
    },
    {
      "epoch": 0.28930735648660977,
      "grad_norm": 1.1317609730907072,
      "learning_rate": 4.257066543173064e-06,
      "loss": 0.2157,
      "step": 555
    },
    {
      "epoch": 0.2898286310028019,
      "grad_norm": 1.1807722421341038,
      "learning_rate": 4.253997512490507e-06,
      "loss": 0.2138,
      "step": 556
    },
    {
      "epoch": 0.2903499055189939,
      "grad_norm": 1.0766939997287335,
      "learning_rate": 4.250923267077489e-06,
      "loss": 0.2251,
      "step": 557
    },
    {
      "epoch": 0.290871180035186,
      "grad_norm": 1.015097333754172,
      "learning_rate": 4.247843816073909e-06,
      "loss": 0.2168,
      "step": 558
    },
    {
      "epoch": 0.29139245455137813,
      "grad_norm": 1.0611808878431113,
      "learning_rate": 4.2447591686351406e-06,
      "loss": 0.2151,
      "step": 559
    },
    {
      "epoch": 0.2919137290675702,
      "grad_norm": 1.081321925415324,
      "learning_rate": 4.2416693339320115e-06,
      "loss": 0.2137,
      "step": 560
    },
    {
      "epoch": 0.2924350035837623,
      "grad_norm": 1.0361316285970896,
      "learning_rate": 4.238574321150769e-06,
      "loss": 0.2001,
      "step": 561
    },
    {
      "epoch": 0.2929562780999544,
      "grad_norm": 1.0732683873731856,
      "learning_rate": 4.235474139493055e-06,
      "loss": 0.2126,
      "step": 562
    },
    {
      "epoch": 0.2934775526161465,
      "grad_norm": 1.042382014216991,
      "learning_rate": 4.23236879817588e-06,
      "loss": 0.2198,
      "step": 563
    },
    {
      "epoch": 0.29399882713233855,
      "grad_norm": 1.05888474945225,
      "learning_rate": 4.229258306431592e-06,
      "loss": 0.2026,
      "step": 564
    },
    {
      "epoch": 0.29452010164853065,
      "grad_norm": 1.096124873852566,
      "learning_rate": 4.226142673507852e-06,
      "loss": 0.2081,
      "step": 565
    },
    {
      "epoch": 0.29504137616472276,
      "grad_norm": 1.0846747928553437,
      "learning_rate": 4.22302190866761e-06,
      "loss": 0.2082,
      "step": 566
    },
    {
      "epoch": 0.29556265068091486,
      "grad_norm": 1.034256519119133,
      "learning_rate": 4.219896021189067e-06,
      "loss": 0.2058,
      "step": 567
    },
    {
      "epoch": 0.2960839251971069,
      "grad_norm": 1.0958509973103114,
      "learning_rate": 4.2167650203656605e-06,
      "loss": 0.2076,
      "step": 568
    },
    {
      "epoch": 0.296605199713299,
      "grad_norm": 1.0219802672482143,
      "learning_rate": 4.213628915506025e-06,
      "loss": 0.2098,
      "step": 569
    },
    {
      "epoch": 0.2971264742294911,
      "grad_norm": 1.0725760126979913,
      "learning_rate": 4.210487715933973e-06,
      "loss": 0.2004,
      "step": 570
    },
    {
      "epoch": 0.29764774874568317,
      "grad_norm": 1.1092050995305374,
      "learning_rate": 4.207341430988461e-06,
      "loss": 0.2179,
      "step": 571
    },
    {
      "epoch": 0.2981690232618753,
      "grad_norm": 1.0116440915769094,
      "learning_rate": 4.204190070023567e-06,
      "loss": 0.2103,
      "step": 572
    },
    {
      "epoch": 0.2986902977780674,
      "grad_norm": 1.0820906649909003,
      "learning_rate": 4.2010336424084596e-06,
      "loss": 0.2108,
      "step": 573
    },
    {
      "epoch": 0.2992115722942595,
      "grad_norm": 1.150843702137832,
      "learning_rate": 4.19787215752737e-06,
      "loss": 0.2114,
      "step": 574
    },
    {
      "epoch": 0.29973284681045154,
      "grad_norm": 1.0751806915419084,
      "learning_rate": 4.194705624779566e-06,
      "loss": 0.2185,
      "step": 575
    },
    {
      "epoch": 0.30025412132664364,
      "grad_norm": 1.0417628113570552,
      "learning_rate": 4.191534053579322e-06,
      "loss": 0.2115,
      "step": 576
    },
    {
      "epoch": 0.30077539584283575,
      "grad_norm": 1.2325769173664691,
      "learning_rate": 4.188357453355893e-06,
      "loss": 0.205,
      "step": 577
    },
    {
      "epoch": 0.3012966703590278,
      "grad_norm": 1.1461454843205805,
      "learning_rate": 4.1851758335534844e-06,
      "loss": 0.2223,
      "step": 578
    },
    {
      "epoch": 0.3018179448752199,
      "grad_norm": 1.0194539603948622,
      "learning_rate": 4.181989203631227e-06,
      "loss": 0.1904,
      "step": 579
    },
    {
      "epoch": 0.302339219391412,
      "grad_norm": 1.082325223841357,
      "learning_rate": 4.178797573063144e-06,
      "loss": 0.2093,
      "step": 580
    },
    {
      "epoch": 0.3028604939076041,
      "grad_norm": 1.0988623753150655,
      "learning_rate": 4.175600951338129e-06,
      "loss": 0.2124,
      "step": 581
    },
    {
      "epoch": 0.30338176842379616,
      "grad_norm": 1.1122185167686438,
      "learning_rate": 4.172399347959912e-06,
      "loss": 0.2293,
      "step": 582
    },
    {
      "epoch": 0.30390304293998827,
      "grad_norm": 1.1104488075802776,
      "learning_rate": 4.169192772447036e-06,
      "loss": 0.2094,
      "step": 583
    },
    {
      "epoch": 0.30442431745618037,
      "grad_norm": 1.1498498606575311,
      "learning_rate": 4.1659812343328246e-06,
      "loss": 0.2083,
      "step": 584
    },
    {
      "epoch": 0.3049455919723725,
      "grad_norm": 1.073342259019356,
      "learning_rate": 4.162764743165355e-06,
      "loss": 0.2062,
      "step": 585
    },
    {
      "epoch": 0.3054668664885645,
      "grad_norm": 1.0622125095205464,
      "learning_rate": 4.1595433085074334e-06,
      "loss": 0.2024,
      "step": 586
    },
    {
      "epoch": 0.30598814100475663,
      "grad_norm": 1.0668772967693896,
      "learning_rate": 4.156316939936559e-06,
      "loss": 0.2212,
      "step": 587
    },
    {
      "epoch": 0.30650941552094874,
      "grad_norm": 1.0675400676528097,
      "learning_rate": 4.153085647044904e-06,
      "loss": 0.2141,
      "step": 588
    },
    {
      "epoch": 0.3070306900371408,
      "grad_norm": 1.0197324992856283,
      "learning_rate": 4.149849439439277e-06,
      "loss": 0.1918,
      "step": 589
    },
    {
      "epoch": 0.3075519645533329,
      "grad_norm": 1.0208947839202611,
      "learning_rate": 4.146608326741101e-06,
      "loss": 0.1979,
      "step": 590
    },
    {
      "epoch": 0.308073239069525,
      "grad_norm": 1.0241897621000515,
      "learning_rate": 4.1433623185863805e-06,
      "loss": 0.2222,
      "step": 591
    },
    {
      "epoch": 0.3085945135857171,
      "grad_norm": 0.9958526236943049,
      "learning_rate": 4.140111424625676e-06,
      "loss": 0.2026,
      "step": 592
    },
    {
      "epoch": 0.30911578810190915,
      "grad_norm": 1.1010645029579753,
      "learning_rate": 4.1368556545240724e-06,
      "loss": 0.2111,
      "step": 593
    },
    {
      "epoch": 0.30963706261810126,
      "grad_norm": 1.020348361008944,
      "learning_rate": 4.133595017961152e-06,
      "loss": 0.2121,
      "step": 594
    },
    {
      "epoch": 0.31015833713429336,
      "grad_norm": 1.0634652896832968,
      "learning_rate": 4.130329524630966e-06,
      "loss": 0.2092,
      "step": 595
    },
    {
      "epoch": 0.3106796116504854,
      "grad_norm": 1.22594320649436,
      "learning_rate": 4.127059184242004e-06,
      "loss": 0.1888,
      "step": 596
    },
    {
      "epoch": 0.3112008861666775,
      "grad_norm": 1.084220606584145,
      "learning_rate": 4.123784006517166e-06,
      "loss": 0.2041,
      "step": 597
    },
    {
      "epoch": 0.3117221606828696,
      "grad_norm": 1.2076936308911805,
      "learning_rate": 4.120504001193737e-06,
      "loss": 0.2247,
      "step": 598
    },
    {
      "epoch": 0.3122434351990617,
      "grad_norm": 1.110425661526799,
      "learning_rate": 4.117219178023349e-06,
      "loss": 0.2187,
      "step": 599
    },
    {
      "epoch": 0.3127647097152538,
      "grad_norm": 1.0165222874522115,
      "learning_rate": 4.113929546771963e-06,
      "loss": 0.2212,
      "step": 600
    },
    {
      "epoch": 0.3132859842314459,
      "grad_norm": 1.01907832542922,
      "learning_rate": 4.1106351172198325e-06,
      "loss": 0.2059,
      "step": 601
    },
    {
      "epoch": 0.313807258747638,
      "grad_norm": 1.1113618660877311,
      "learning_rate": 4.1073358991614745e-06,
      "loss": 0.2056,
      "step": 602
    },
    {
      "epoch": 0.3143285332638301,
      "grad_norm": 1.061394484619993,
      "learning_rate": 4.1040319024056465e-06,
      "loss": 0.2157,
      "step": 603
    },
    {
      "epoch": 0.31484980778002214,
      "grad_norm": 1.0770711874914694,
      "learning_rate": 4.10072313677531e-06,
      "loss": 0.2267,
      "step": 604
    },
    {
      "epoch": 0.31537108229621424,
      "grad_norm": 1.1000479766093059,
      "learning_rate": 4.0974096121076076e-06,
      "loss": 0.2274,
      "step": 605
    },
    {
      "epoch": 0.31589235681240635,
      "grad_norm": 1.030867505172163,
      "learning_rate": 4.094091338253829e-06,
      "loss": 0.2042,
      "step": 606
    },
    {
      "epoch": 0.3164136313285984,
      "grad_norm": 1.0592868986180395,
      "learning_rate": 4.0907683250793814e-06,
      "loss": 0.1994,
      "step": 607
    },
    {
      "epoch": 0.3169349058447905,
      "grad_norm": 1.0149014356934085,
      "learning_rate": 4.0874405824637676e-06,
      "loss": 0.2021,
      "step": 608
    },
    {
      "epoch": 0.3174561803609826,
      "grad_norm": 1.0683812405742192,
      "learning_rate": 4.084108120300546e-06,
      "loss": 0.2047,
      "step": 609
    },
    {
      "epoch": 0.3179774548771747,
      "grad_norm": 1.1387780132206826,
      "learning_rate": 4.080770948497311e-06,
      "loss": 0.1883,
      "step": 610
    },
    {
      "epoch": 0.31849872939336676,
      "grad_norm": 1.1221245908523543,
      "learning_rate": 4.077429076975655e-06,
      "loss": 0.2095,
      "step": 611
    },
    {
      "epoch": 0.31902000390955887,
      "grad_norm": 1.0282758937773733,
      "learning_rate": 4.074082515671145e-06,
      "loss": 0.2145,
      "step": 612
    },
    {
      "epoch": 0.319541278425751,
      "grad_norm": 1.110630040223085,
      "learning_rate": 4.070731274533291e-06,
      "loss": 0.2049,
      "step": 613
    },
    {
      "epoch": 0.320062552941943,
      "grad_norm": 1.056095149339451,
      "learning_rate": 4.067375363525516e-06,
      "loss": 0.2137,
      "step": 614
    },
    {
      "epoch": 0.32058382745813513,
      "grad_norm": 1.0650979149433963,
      "learning_rate": 4.064014792625126e-06,
      "loss": 0.2003,
      "step": 615
    },
    {
      "epoch": 0.32110510197432723,
      "grad_norm": 1.099715369531199,
      "learning_rate": 4.060649571823284e-06,
      "loss": 0.2091,
      "step": 616
    },
    {
      "epoch": 0.32162637649051934,
      "grad_norm": 0.9805781333526467,
      "learning_rate": 4.057279711124973e-06,
      "loss": 0.2096,
      "step": 617
    },
    {
      "epoch": 0.3221476510067114,
      "grad_norm": 1.155776870795669,
      "learning_rate": 4.053905220548974e-06,
      "loss": 0.2212,
      "step": 618
    },
    {
      "epoch": 0.3226689255229035,
      "grad_norm": 1.0538004320056853,
      "learning_rate": 4.050526110127832e-06,
      "loss": 0.1766,
      "step": 619
    },
    {
      "epoch": 0.3231902000390956,
      "grad_norm": 1.0618488101877996,
      "learning_rate": 4.047142389907827e-06,
      "loss": 0.2056,
      "step": 620
    },
    {
      "epoch": 0.3237114745552877,
      "grad_norm": 1.0565037231122896,
      "learning_rate": 4.043754069948944e-06,
      "loss": 0.2175,
      "step": 621
    },
    {
      "epoch": 0.32423274907147975,
      "grad_norm": 1.0734547383525697,
      "learning_rate": 4.040361160324844e-06,
      "loss": 0.2108,
      "step": 622
    },
    {
      "epoch": 0.32475402358767186,
      "grad_norm": 1.1404843494285408,
      "learning_rate": 4.036963671122831e-06,
      "loss": 0.1984,
      "step": 623
    },
    {
      "epoch": 0.32527529810386396,
      "grad_norm": 1.0409356970425339,
      "learning_rate": 4.033561612443829e-06,
      "loss": 0.2186,
      "step": 624
    },
    {
      "epoch": 0.325796572620056,
      "grad_norm": 1.0275010010679515,
      "learning_rate": 4.030154994402341e-06,
      "loss": 0.202,
      "step": 625
    },
    {
      "epoch": 0.3263178471362481,
      "grad_norm": 1.0483642396947566,
      "learning_rate": 4.0267438271264304e-06,
      "loss": 0.1998,
      "step": 626
    },
    {
      "epoch": 0.3268391216524402,
      "grad_norm": 1.2195252537996888,
      "learning_rate": 4.023328120757685e-06,
      "loss": 0.2196,
      "step": 627
    },
    {
      "epoch": 0.32736039616863233,
      "grad_norm": 1.0662345771718675,
      "learning_rate": 4.019907885451184e-06,
      "loss": 0.2065,
      "step": 628
    },
    {
      "epoch": 0.3278816706848244,
      "grad_norm": 1.0183412109857055,
      "learning_rate": 4.016483131375476e-06,
      "loss": 0.1978,
      "step": 629
    },
    {
      "epoch": 0.3284029452010165,
      "grad_norm": 1.1051721887857624,
      "learning_rate": 4.01305386871254e-06,
      "loss": 0.2181,
      "step": 630
    },
    {
      "epoch": 0.3289242197172086,
      "grad_norm": 1.0671144442819984,
      "learning_rate": 4.009620107657763e-06,
      "loss": 0.2189,
      "step": 631
    },
    {
      "epoch": 0.32944549423340064,
      "grad_norm": 1.0747517554303343,
      "learning_rate": 4.006181858419905e-06,
      "loss": 0.203,
      "step": 632
    },
    {
      "epoch": 0.32996676874959274,
      "grad_norm": 0.9833069358480406,
      "learning_rate": 4.002739131221066e-06,
      "loss": 0.1977,
      "step": 633
    },
    {
      "epoch": 0.33048804326578485,
      "grad_norm": 1.075564747056274,
      "learning_rate": 3.999291936296664e-06,
      "loss": 0.2003,
      "step": 634
    },
    {
      "epoch": 0.33100931778197695,
      "grad_norm": 1.0211300435597885,
      "learning_rate": 3.995840283895399e-06,
      "loss": 0.1925,
      "step": 635
    },
    {
      "epoch": 0.331530592298169,
      "grad_norm": 1.055480781342074,
      "learning_rate": 3.99238418427922e-06,
      "loss": 0.2083,
      "step": 636
    },
    {
      "epoch": 0.3320518668143611,
      "grad_norm": 1.1455013890415642,
      "learning_rate": 3.988923647723301e-06,
      "loss": 0.2245,
      "step": 637
    },
    {
      "epoch": 0.3325731413305532,
      "grad_norm": 1.138804936681144,
      "learning_rate": 3.9854586845160055e-06,
      "loss": 0.2097,
      "step": 638
    },
    {
      "epoch": 0.3330944158467453,
      "grad_norm": 1.0391589155707512,
      "learning_rate": 3.981989304958861e-06,
      "loss": 0.1979,
      "step": 639
    },
    {
      "epoch": 0.33361569036293737,
      "grad_norm": 0.9846771488041198,
      "learning_rate": 3.978515519366519e-06,
      "loss": 0.191,
      "step": 640
    },
    {
      "epoch": 0.33413696487912947,
      "grad_norm": 1.160705167196543,
      "learning_rate": 3.975037338066736e-06,
      "loss": 0.2269,
      "step": 641
    },
    {
      "epoch": 0.3346582393953216,
      "grad_norm": 1.0341541389157343,
      "learning_rate": 3.9715547714003355e-06,
      "loss": 0.212,
      "step": 642
    },
    {
      "epoch": 0.3351795139115136,
      "grad_norm": 0.9669202329689861,
      "learning_rate": 3.968067829721178e-06,
      "loss": 0.1926,
      "step": 643
    },
    {
      "epoch": 0.33570078842770573,
      "grad_norm": 1.0646552233485294,
      "learning_rate": 3.96457652339613e-06,
      "loss": 0.2213,
      "step": 644
    },
    {
      "epoch": 0.33622206294389784,
      "grad_norm": 1.0267029973506987,
      "learning_rate": 3.961080862805039e-06,
      "loss": 0.2033,
      "step": 645
    },
    {
      "epoch": 0.33674333746008994,
      "grad_norm": 1.0386343031224388,
      "learning_rate": 3.9575808583406926e-06,
      "loss": 0.1999,
      "step": 646
    },
    {
      "epoch": 0.337264611976282,
      "grad_norm": 1.0609243483484754,
      "learning_rate": 3.954076520408796e-06,
      "loss": 0.2067,
      "step": 647
    },
    {
      "epoch": 0.3377858864924741,
      "grad_norm": 1.0156303481124798,
      "learning_rate": 3.950567859427938e-06,
      "loss": 0.1966,
      "step": 648
    },
    {
      "epoch": 0.3383071610086662,
      "grad_norm": 1.0521948883806702,
      "learning_rate": 3.947054885829559e-06,
      "loss": 0.1943,
      "step": 649
    },
    {
      "epoch": 0.33882843552485825,
      "grad_norm": 1.1434498924585015,
      "learning_rate": 3.943537610057921e-06,
      "loss": 0.2166,
      "step": 650
    },
    {
      "epoch": 0.33934971004105036,
      "grad_norm": 1.1505038673012769,
      "learning_rate": 3.940016042570079e-06,
      "loss": 0.2182,
      "step": 651
    },
    {
      "epoch": 0.33987098455724246,
      "grad_norm": 1.0440410659240964,
      "learning_rate": 3.936490193835843e-06,
      "loss": 0.2093,
      "step": 652
    },
    {
      "epoch": 0.34039225907343457,
      "grad_norm": 1.1208632283728355,
      "learning_rate": 3.932960074337755e-06,
      "loss": 0.2269,
      "step": 653
    },
    {
      "epoch": 0.3409135335896266,
      "grad_norm": 1.0861275993743262,
      "learning_rate": 3.929425694571055e-06,
      "loss": 0.2035,
      "step": 654
    },
    {
      "epoch": 0.3414348081058187,
      "grad_norm": 1.0290460624684266,
      "learning_rate": 3.925887065043643e-06,
      "loss": 0.2122,
      "step": 655
    },
    {
      "epoch": 0.3419560826220108,
      "grad_norm": 1.0884709264769998,
      "learning_rate": 3.922344196276063e-06,
      "loss": 0.2139,
      "step": 656
    },
    {
      "epoch": 0.34247735713820293,
      "grad_norm": 1.0532577216421508,
      "learning_rate": 3.918797098801453e-06,
      "loss": 0.2027,
      "step": 657
    },
    {
      "epoch": 0.342998631654395,
      "grad_norm": 1.0704967165140549,
      "learning_rate": 3.915245783165531e-06,
      "loss": 0.1975,
      "step": 658
    },
    {
      "epoch": 0.3435199061705871,
      "grad_norm": 1.1395825307815826,
      "learning_rate": 3.911690259926551e-06,
      "loss": 0.21,
      "step": 659
    },
    {
      "epoch": 0.3440411806867792,
      "grad_norm": 1.0578163243072147,
      "learning_rate": 3.908130539655278e-06,
      "loss": 0.2029,
      "step": 660
    },
    {
      "epoch": 0.34456245520297124,
      "grad_norm": 1.1280961630838429,
      "learning_rate": 3.904566632934955e-06,
      "loss": 0.2265,
      "step": 661
    },
    {
      "epoch": 0.34508372971916335,
      "grad_norm": 1.0471018683165267,
      "learning_rate": 3.900998550361271e-06,
      "loss": 0.2064,
      "step": 662
    },
    {
      "epoch": 0.34560500423535545,
      "grad_norm": 1.051587787475786,
      "learning_rate": 3.897426302542331e-06,
      "loss": 0.2009,
      "step": 663
    },
    {
      "epoch": 0.34612627875154756,
      "grad_norm": 1.07096724118746,
      "learning_rate": 3.893849900098623e-06,
      "loss": 0.1984,
      "step": 664
    },
    {
      "epoch": 0.3466475532677396,
      "grad_norm": 1.1311625175456812,
      "learning_rate": 3.890269353662987e-06,
      "loss": 0.2042,
      "step": 665
    },
    {
      "epoch": 0.3471688277839317,
      "grad_norm": 1.0603390750522113,
      "learning_rate": 3.886684673880583e-06,
      "loss": 0.2077,
      "step": 666
    },
    {
      "epoch": 0.3476901023001238,
      "grad_norm": 1.0002435941995667,
      "learning_rate": 3.8830958714088595e-06,
      "loss": 0.2036,
      "step": 667
    },
    {
      "epoch": 0.34821137681631587,
      "grad_norm": 1.09095658907915,
      "learning_rate": 3.879502956917524e-06,
      "loss": 0.2034,
      "step": 668
    },
    {
      "epoch": 0.34873265133250797,
      "grad_norm": 1.1310309501196332,
      "learning_rate": 3.875905941088505e-06,
      "loss": 0.2075,
      "step": 669
    },
    {
      "epoch": 0.3492539258487001,
      "grad_norm": 1.0418279096661451,
      "learning_rate": 3.872304834615929e-06,
      "loss": 0.1985,
      "step": 670
    },
    {
      "epoch": 0.3497752003648922,
      "grad_norm": 0.9684063606430465,
      "learning_rate": 3.868699648206081e-06,
      "loss": 0.1895,
      "step": 671
    },
    {
      "epoch": 0.35029647488108423,
      "grad_norm": 1.1204415636477791,
      "learning_rate": 3.8650903925773795e-06,
      "loss": 0.207,
      "step": 672
    },
    {
      "epoch": 0.35081774939727633,
      "grad_norm": 1.1243132412029433,
      "learning_rate": 3.861477078460337e-06,
      "loss": 0.2111,
      "step": 673
    },
    {
      "epoch": 0.35133902391346844,
      "grad_norm": 1.0979496907292703,
      "learning_rate": 3.857859716597534e-06,
      "loss": 0.2148,
      "step": 674
    },
    {
      "epoch": 0.35186029842966055,
      "grad_norm": 1.0443405543071218,
      "learning_rate": 3.854238317743586e-06,
      "loss": 0.2196,
      "step": 675
    },
    {
      "epoch": 0.3523815729458526,
      "grad_norm": 1.0762635593895937,
      "learning_rate": 3.8506128926651095e-06,
      "loss": 0.2105,
      "step": 676
    },
    {
      "epoch": 0.3529028474620447,
      "grad_norm": 1.0338762149355627,
      "learning_rate": 3.846983452140689e-06,
      "loss": 0.2132,
      "step": 677
    },
    {
      "epoch": 0.3534241219782368,
      "grad_norm": 1.0074710371606315,
      "learning_rate": 3.843350006960852e-06,
      "loss": 0.2053,
      "step": 678
    },
    {
      "epoch": 0.35394539649442885,
      "grad_norm": 1.0126329998732087,
      "learning_rate": 3.83971256792803e-06,
      "loss": 0.2102,
      "step": 679
    },
    {
      "epoch": 0.35446667101062096,
      "grad_norm": 1.0068494471356129,
      "learning_rate": 3.836071145856526e-06,
      "loss": 0.2021,
      "step": 680
    },
    {
      "epoch": 0.35498794552681306,
      "grad_norm": 1.0269809005666712,
      "learning_rate": 3.832425751572488e-06,
      "loss": 0.2034,
      "step": 681
    },
    {
      "epoch": 0.35550922004300517,
      "grad_norm": 0.9716053014003645,
      "learning_rate": 3.828776395913872e-06,
      "loss": 0.1993,
      "step": 682
    },
    {
      "epoch": 0.3560304945591972,
      "grad_norm": 1.0117095527596758,
      "learning_rate": 3.825123089730413e-06,
      "loss": 0.2096,
      "step": 683
    },
    {
      "epoch": 0.3565517690753893,
      "grad_norm": 1.0729107646685034,
      "learning_rate": 3.821465843883588e-06,
      "loss": 0.1915,
      "step": 684
    },
    {
      "epoch": 0.35707304359158143,
      "grad_norm": 1.0320361681933137,
      "learning_rate": 3.81780466924659e-06,
      "loss": 0.1905,
      "step": 685
    },
    {
      "epoch": 0.3575943181077735,
      "grad_norm": 0.9715658008355695,
      "learning_rate": 3.814139576704291e-06,
      "loss": 0.1969,
      "step": 686
    },
    {
      "epoch": 0.3581155926239656,
      "grad_norm": 1.0932036907834042,
      "learning_rate": 3.810470577153212e-06,
      "loss": 0.2115,
      "step": 687
    },
    {
      "epoch": 0.3586368671401577,
      "grad_norm": 1.029912994292444,
      "learning_rate": 3.8067976815014885e-06,
      "loss": 0.1997,
      "step": 688
    },
    {
      "epoch": 0.3591581416563498,
      "grad_norm": 1.0811060169767797,
      "learning_rate": 3.8031209006688397e-06,
      "loss": 0.2077,
      "step": 689
    },
    {
      "epoch": 0.35967941617254184,
      "grad_norm": 1.0421585599821286,
      "learning_rate": 3.7994402455865375e-06,
      "loss": 0.2079,
      "step": 690
    },
    {
      "epoch": 0.36020069068873395,
      "grad_norm": 1.0377992818502029,
      "learning_rate": 3.795755727197368e-06,
      "loss": 0.2017,
      "step": 691
    },
    {
      "epoch": 0.36072196520492605,
      "grad_norm": 1.0065660374049659,
      "learning_rate": 3.7920673564556083e-06,
      "loss": 0.2221,
      "step": 692
    },
    {
      "epoch": 0.36124323972111816,
      "grad_norm": 0.979282912950048,
      "learning_rate": 3.788375144326985e-06,
      "loss": 0.1955,
      "step": 693
    },
    {
      "epoch": 0.3617645142373102,
      "grad_norm": 1.0026392413250227,
      "learning_rate": 3.784679101788647e-06,
      "loss": 0.2116,
      "step": 694
    },
    {
      "epoch": 0.3622857887535023,
      "grad_norm": 0.9986548278232529,
      "learning_rate": 3.78097923982913e-06,
      "loss": 0.214,
      "step": 695
    },
    {
      "epoch": 0.3628070632696944,
      "grad_norm": 0.9585609430319894,
      "learning_rate": 3.7772755694483265e-06,
      "loss": 0.1905,
      "step": 696
    },
    {
      "epoch": 0.36332833778588647,
      "grad_norm": 1.0401759307528902,
      "learning_rate": 3.7735681016574504e-06,
      "loss": 0.1971,
      "step": 697
    },
    {
      "epoch": 0.3638496123020786,
      "grad_norm": 1.0360631930811912,
      "learning_rate": 3.7698568474790064e-06,
      "loss": 0.1996,
      "step": 698
    },
    {
      "epoch": 0.3643708868182707,
      "grad_norm": 1.0178019461687884,
      "learning_rate": 3.766141817946757e-06,
      "loss": 0.2161,
      "step": 699
    },
    {
      "epoch": 0.3648921613344628,
      "grad_norm": 0.9971708011450119,
      "learning_rate": 3.7624230241056854e-06,
      "loss": 0.2008,
      "step": 700
    },
    {
      "epoch": 0.36541343585065483,
      "grad_norm": 1.1371013279219415,
      "learning_rate": 3.7587004770119716e-06,
      "loss": 0.2293,
      "step": 701
    },
    {
      "epoch": 0.36593471036684694,
      "grad_norm": 1.0637851468087032,
      "learning_rate": 3.7549741877329504e-06,
      "loss": 0.2015,
      "step": 702
    },
    {
      "epoch": 0.36645598488303904,
      "grad_norm": 1.0182788467476143,
      "learning_rate": 3.7512441673470836e-06,
      "loss": 0.2097,
      "step": 703
    },
    {
      "epoch": 0.3669772593992311,
      "grad_norm": 0.9873903086378643,
      "learning_rate": 3.747510426943925e-06,
      "loss": 0.1945,
      "step": 704
    },
    {
      "epoch": 0.3674985339154232,
      "grad_norm": 1.066187464643768,
      "learning_rate": 3.7437729776240894e-06,
      "loss": 0.2033,
      "step": 705
    },
    {
      "epoch": 0.3680198084316153,
      "grad_norm": 1.0109132335088629,
      "learning_rate": 3.740031830499219e-06,
      "loss": 0.187,
      "step": 706
    },
    {
      "epoch": 0.3685410829478074,
      "grad_norm": 1.087352920662628,
      "learning_rate": 3.7362869966919467e-06,
      "loss": 0.2017,
      "step": 707
    },
    {
      "epoch": 0.36906235746399946,
      "grad_norm": 1.0923626972489895,
      "learning_rate": 3.7325384873358695e-06,
      "loss": 0.2112,
      "step": 708
    },
    {
      "epoch": 0.36958363198019156,
      "grad_norm": 1.078860234772305,
      "learning_rate": 3.7287863135755098e-06,
      "loss": 0.2117,
      "step": 709
    },
    {
      "epoch": 0.37010490649638367,
      "grad_norm": 1.1449021033180053,
      "learning_rate": 3.7250304865662857e-06,
      "loss": 0.1984,
      "step": 710
    },
    {
      "epoch": 0.3706261810125758,
      "grad_norm": 1.0614686156604147,
      "learning_rate": 3.7212710174744753e-06,
      "loss": 0.2079,
      "step": 711
    },
    {
      "epoch": 0.3711474555287678,
      "grad_norm": 1.037377272701006,
      "learning_rate": 3.7175079174771872e-06,
      "loss": 0.2022,
      "step": 712
    },
    {
      "epoch": 0.3716687300449599,
      "grad_norm": 1.0516583337554275,
      "learning_rate": 3.713741197762323e-06,
      "loss": 0.2009,
      "step": 713
    },
    {
      "epoch": 0.37219000456115203,
      "grad_norm": 1.0268348370799318,
      "learning_rate": 3.7099708695285436e-06,
      "loss": 0.1955,
      "step": 714
    },
    {
      "epoch": 0.3727112790773441,
      "grad_norm": 0.9891691053873646,
      "learning_rate": 3.706196943985245e-06,
      "loss": 0.198,
      "step": 715
    },
    {
      "epoch": 0.3732325535935362,
      "grad_norm": 1.0507470048468373,
      "learning_rate": 3.7024194323525115e-06,
      "loss": 0.2183,
      "step": 716
    },
    {
      "epoch": 0.3737538281097283,
      "grad_norm": 1.0441076323714367,
      "learning_rate": 3.6986383458610925e-06,
      "loss": 0.1982,
      "step": 717
    },
    {
      "epoch": 0.3742751026259204,
      "grad_norm": 0.9343537658012983,
      "learning_rate": 3.694853695752364e-06,
      "loss": 0.1944,
      "step": 718
    },
    {
      "epoch": 0.37479637714211245,
      "grad_norm": 0.9997766034036779,
      "learning_rate": 3.6910654932782984e-06,
      "loss": 0.2017,
      "step": 719
    },
    {
      "epoch": 0.37531765165830455,
      "grad_norm": 1.028401152906009,
      "learning_rate": 3.6872737497014286e-06,
      "loss": 0.2002,
      "step": 720
    },
    {
      "epoch": 0.37583892617449666,
      "grad_norm": 1.0087687788014903,
      "learning_rate": 3.6834784762948156e-06,
      "loss": 0.1982,
      "step": 721
    },
    {
      "epoch": 0.3763602006906887,
      "grad_norm": 0.9997736649547044,
      "learning_rate": 3.6796796843420134e-06,
      "loss": 0.2071,
      "step": 722
    },
    {
      "epoch": 0.3768814752068808,
      "grad_norm": 0.9721162464369297,
      "learning_rate": 3.67587738513704e-06,
      "loss": 0.1884,
      "step": 723
    },
    {
      "epoch": 0.3774027497230729,
      "grad_norm": 1.0304064094853025,
      "learning_rate": 3.672071589984337e-06,
      "loss": 0.1995,
      "step": 724
    },
    {
      "epoch": 0.377924024239265,
      "grad_norm": 1.0589375873312883,
      "learning_rate": 3.6682623101987423e-06,
      "loss": 0.1981,
      "step": 725
    },
    {
      "epoch": 0.37844529875545707,
      "grad_norm": 0.9710396112430536,
      "learning_rate": 3.664449557105454e-06,
      "loss": 0.2,
      "step": 726
    },
    {
      "epoch": 0.3789665732716492,
      "grad_norm": 1.0522926935967352,
      "learning_rate": 3.6606333420399933e-06,
      "loss": 0.2091,
      "step": 727
    },
    {
      "epoch": 0.3794878477878413,
      "grad_norm": 1.0852598572018897,
      "learning_rate": 3.6568136763481766e-06,
      "loss": 0.2079,
      "step": 728
    },
    {
      "epoch": 0.3800091223040334,
      "grad_norm": 1.0492806894049111,
      "learning_rate": 3.652990571386079e-06,
      "loss": 0.2074,
      "step": 729
    },
    {
      "epoch": 0.38053039682022544,
      "grad_norm": 1.0085185599351503,
      "learning_rate": 3.649164038520001e-06,
      "loss": 0.1935,
      "step": 730
    },
    {
      "epoch": 0.38105167133641754,
      "grad_norm": 1.0857364772225881,
      "learning_rate": 3.6453340891264344e-06,
      "loss": 0.2052,
      "step": 731
    },
    {
      "epoch": 0.38157294585260965,
      "grad_norm": 1.1224287433484814,
      "learning_rate": 3.641500734592026e-06,
      "loss": 0.205,
      "step": 732
    },
    {
      "epoch": 0.3820942203688017,
      "grad_norm": 1.155690013988157,
      "learning_rate": 3.63766398631355e-06,
      "loss": 0.2281,
      "step": 733
    },
    {
      "epoch": 0.3826154948849938,
      "grad_norm": 0.9944633480040708,
      "learning_rate": 3.633823855697869e-06,
      "loss": 0.1834,
      "step": 734
    },
    {
      "epoch": 0.3831367694011859,
      "grad_norm": 1.016517755609509,
      "learning_rate": 3.6299803541619e-06,
      "loss": 0.2101,
      "step": 735
    },
    {
      "epoch": 0.383658043917378,
      "grad_norm": 0.965254600627704,
      "learning_rate": 3.6261334931325833e-06,
      "loss": 0.1819,
      "step": 736
    },
    {
      "epoch": 0.38417931843357006,
      "grad_norm": 1.0509533639633712,
      "learning_rate": 3.622283284046847e-06,
      "loss": 0.1969,
      "step": 737
    },
    {
      "epoch": 0.38470059294976217,
      "grad_norm": 1.0273270859307988,
      "learning_rate": 3.618429738351574e-06,
      "loss": 0.1955,
      "step": 738
    },
    {
      "epoch": 0.38522186746595427,
      "grad_norm": 1.0828612063888188,
      "learning_rate": 3.6145728675035658e-06,
      "loss": 0.2083,
      "step": 739
    },
    {
      "epoch": 0.3857431419821463,
      "grad_norm": 1.037973214218959,
      "learning_rate": 3.6107126829695093e-06,
      "loss": 0.1959,
      "step": 740
    },
    {
      "epoch": 0.3862644164983384,
      "grad_norm": 0.9740301727879055,
      "learning_rate": 3.6068491962259457e-06,
      "loss": 0.1981,
      "step": 741
    },
    {
      "epoch": 0.38678569101453053,
      "grad_norm": 0.9675934276994784,
      "learning_rate": 3.6029824187592304e-06,
      "loss": 0.1905,
      "step": 742
    },
    {
      "epoch": 0.38730696553072264,
      "grad_norm": 0.9985279581622839,
      "learning_rate": 3.599112362065506e-06,
      "loss": 0.1946,
      "step": 743
    },
    {
      "epoch": 0.3878282400469147,
      "grad_norm": 1.0065347198646675,
      "learning_rate": 3.5952390376506614e-06,
      "loss": 0.1982,
      "step": 744
    },
    {
      "epoch": 0.3883495145631068,
      "grad_norm": 1.0684404764047237,
      "learning_rate": 3.591362457030302e-06,
      "loss": 0.2241,
      "step": 745
    },
    {
      "epoch": 0.3888707890792989,
      "grad_norm": 1.0386178441721692,
      "learning_rate": 3.5874826317297135e-06,
      "loss": 0.2158,
      "step": 746
    },
    {
      "epoch": 0.389392063595491,
      "grad_norm": 1.1467068947450896,
      "learning_rate": 3.5835995732838292e-06,
      "loss": 0.2187,
      "step": 747
    },
    {
      "epoch": 0.38991333811168305,
      "grad_norm": 0.9978113760767837,
      "learning_rate": 3.579713293237194e-06,
      "loss": 0.1909,
      "step": 748
    },
    {
      "epoch": 0.39043461262787515,
      "grad_norm": 1.0577765356298126,
      "learning_rate": 3.5758238031439306e-06,
      "loss": 0.2048,
      "step": 749
    },
    {
      "epoch": 0.39095588714406726,
      "grad_norm": 1.0270887257797516,
      "learning_rate": 3.5719311145677057e-06,
      "loss": 0.195,
      "step": 750
    },
    {
      "epoch": 0.3914771616602593,
      "grad_norm": 1.0201002842834375,
      "learning_rate": 3.5680352390816942e-06,
      "loss": 0.2078,
      "step": 751
    },
    {
      "epoch": 0.3919984361764514,
      "grad_norm": 1.069239425622886,
      "learning_rate": 3.5641361882685487e-06,
      "loss": 0.2055,
      "step": 752
    },
    {
      "epoch": 0.3925197106926435,
      "grad_norm": 0.999820669699393,
      "learning_rate": 3.5602339737203593e-06,
      "loss": 0.2047,
      "step": 753
    },
    {
      "epoch": 0.3930409852088356,
      "grad_norm": 1.0700111846850282,
      "learning_rate": 3.5563286070386237e-06,
      "loss": 0.2145,
      "step": 754
    },
    {
      "epoch": 0.3935622597250277,
      "grad_norm": 1.1382464786486353,
      "learning_rate": 3.5524200998342095e-06,
      "loss": 0.2122,
      "step": 755
    },
    {
      "epoch": 0.3940835342412198,
      "grad_norm": 1.072823077451666,
      "learning_rate": 3.5485084637273225e-06,
      "loss": 0.1982,
      "step": 756
    },
    {
      "epoch": 0.3946048087574119,
      "grad_norm": 1.0198723117083601,
      "learning_rate": 3.5445937103474713e-06,
      "loss": 0.2047,
      "step": 757
    },
    {
      "epoch": 0.39512608327360393,
      "grad_norm": 0.9377945180547989,
      "learning_rate": 3.5406758513334316e-06,
      "loss": 0.194,
      "step": 758
    },
    {
      "epoch": 0.39564735778979604,
      "grad_norm": 1.054102417249867,
      "learning_rate": 3.536754898333211e-06,
      "loss": 0.2125,
      "step": 759
    },
    {
      "epoch": 0.39616863230598814,
      "grad_norm": 1.0932943063554317,
      "learning_rate": 3.532830863004018e-06,
      "loss": 0.2018,
      "step": 760
    },
    {
      "epoch": 0.39668990682218025,
      "grad_norm": 1.0453261209434133,
      "learning_rate": 3.5289037570122246e-06,
      "loss": 0.1793,
      "step": 761
    },
    {
      "epoch": 0.3972111813383723,
      "grad_norm": 1.085250928485593,
      "learning_rate": 3.5249735920333312e-06,
      "loss": 0.2215,
      "step": 762
    },
    {
      "epoch": 0.3977324558545644,
      "grad_norm": 1.0207448526376532,
      "learning_rate": 3.521040379751933e-06,
      "loss": 0.212,
      "step": 763
    },
    {
      "epoch": 0.3982537303707565,
      "grad_norm": 1.06545180546958,
      "learning_rate": 3.517104131861685e-06,
      "loss": 0.2093,
      "step": 764
    },
    {
      "epoch": 0.3987750048869486,
      "grad_norm": 1.015376416999948,
      "learning_rate": 3.513164860065267e-06,
      "loss": 0.2047,
      "step": 765
    },
    {
      "epoch": 0.39929627940314066,
      "grad_norm": 1.0885519954151794,
      "learning_rate": 3.509222576074349e-06,
      "loss": 0.2164,
      "step": 766
    },
    {
      "epoch": 0.39981755391933277,
      "grad_norm": 1.0221092011588429,
      "learning_rate": 3.5052772916095584e-06,
      "loss": 0.197,
      "step": 767
    },
    {
      "epoch": 0.4003388284355249,
      "grad_norm": 1.0074564985446492,
      "learning_rate": 3.50132901840044e-06,
      "loss": 0.1848,
      "step": 768
    },
    {
      "epoch": 0.4008601029517169,
      "grad_norm": 1.124944009301868,
      "learning_rate": 3.4973777681854265e-06,
      "loss": 0.227,
      "step": 769
    },
    {
      "epoch": 0.40138137746790903,
      "grad_norm": 0.9933224411295778,
      "learning_rate": 3.4934235527118e-06,
      "loss": 0.2007,
      "step": 770
    },
    {
      "epoch": 0.40190265198410113,
      "grad_norm": 1.1140553396131367,
      "learning_rate": 3.4894663837356607e-06,
      "loss": 0.2062,
      "step": 771
    },
    {
      "epoch": 0.40242392650029324,
      "grad_norm": 1.07900121401784,
      "learning_rate": 3.485506273021887e-06,
      "loss": 0.2174,
      "step": 772
    },
    {
      "epoch": 0.4029452010164853,
      "grad_norm": 1.0396822700390023,
      "learning_rate": 3.4815432323441043e-06,
      "loss": 0.2265,
      "step": 773
    },
    {
      "epoch": 0.4034664755326774,
      "grad_norm": 1.0063749840646985,
      "learning_rate": 3.47757727348465e-06,
      "loss": 0.2058,
      "step": 774
    },
    {
      "epoch": 0.4039877500488695,
      "grad_norm": 1.0351010851739715,
      "learning_rate": 3.4736084082345355e-06,
      "loss": 0.2093,
      "step": 775
    },
    {
      "epoch": 0.40450902456506155,
      "grad_norm": 0.9589611297953635,
      "learning_rate": 3.4696366483934156e-06,
      "loss": 0.2004,
      "step": 776
    },
    {
      "epoch": 0.40503029908125365,
      "grad_norm": 1.052474173402005,
      "learning_rate": 3.465662005769548e-06,
      "loss": 0.205,
      "step": 777
    },
    {
      "epoch": 0.40555157359744576,
      "grad_norm": 0.9712723540245317,
      "learning_rate": 3.461684492179763e-06,
      "loss": 0.195,
      "step": 778
    },
    {
      "epoch": 0.40607284811363786,
      "grad_norm": 1.0335824947872163,
      "learning_rate": 3.4577041194494253e-06,
      "loss": 0.1955,
      "step": 779
    },
    {
      "epoch": 0.4065941226298299,
      "grad_norm": 0.9782579425499615,
      "learning_rate": 3.4537208994124015e-06,
      "loss": 0.1829,
      "step": 780
    },
    {
      "epoch": 0.407115397146022,
      "grad_norm": 1.011734636363808,
      "learning_rate": 3.449734843911022e-06,
      "loss": 0.2034,
      "step": 781
    },
    {
      "epoch": 0.4076366716622141,
      "grad_norm": 0.9975323909301046,
      "learning_rate": 3.4457459647960477e-06,
      "loss": 0.208,
      "step": 782
    },
    {
      "epoch": 0.40815794617840623,
      "grad_norm": 1.004285481609286,
      "learning_rate": 3.441754273926634e-06,
      "loss": 0.1993,
      "step": 783
    },
    {
      "epoch": 0.4086792206945983,
      "grad_norm": 1.0152260185286432,
      "learning_rate": 3.4377597831702962e-06,
      "loss": 0.202,
      "step": 784
    },
    {
      "epoch": 0.4092004952107904,
      "grad_norm": 1.029762874421123,
      "learning_rate": 3.433762504402874e-06,
      "loss": 0.1923,
      "step": 785
    },
    {
      "epoch": 0.4097217697269825,
      "grad_norm": 1.0259363482780564,
      "learning_rate": 3.429762449508495e-06,
      "loss": 0.1877,
      "step": 786
    },
    {
      "epoch": 0.41024304424317454,
      "grad_norm": 0.9708063683587468,
      "learning_rate": 3.425759630379541e-06,
      "loss": 0.1939,
      "step": 787
    },
    {
      "epoch": 0.41076431875936664,
      "grad_norm": 1.0125745012513352,
      "learning_rate": 3.421754058916612e-06,
      "loss": 0.1932,
      "step": 788
    },
    {
      "epoch": 0.41128559327555875,
      "grad_norm": 1.0539642303638106,
      "learning_rate": 3.4177457470284916e-06,
      "loss": 0.2186,
      "step": 789
    },
    {
      "epoch": 0.41180686779175085,
      "grad_norm": 1.0106759073395906,
      "learning_rate": 3.4137347066321097e-06,
      "loss": 0.2025,
      "step": 790
    },
    {
      "epoch": 0.4123281423079429,
      "grad_norm": 1.047283205204026,
      "learning_rate": 3.4097209496525087e-06,
      "loss": 0.2136,
      "step": 791
    },
    {
      "epoch": 0.412849416824135,
      "grad_norm": 1.0288785017939832,
      "learning_rate": 3.4057044880228064e-06,
      "loss": 0.1987,
      "step": 792
    },
    {
      "epoch": 0.4133706913403271,
      "grad_norm": 1.100836923285089,
      "learning_rate": 3.4016853336841638e-06,
      "loss": 0.2165,
      "step": 793
    },
    {
      "epoch": 0.41389196585651916,
      "grad_norm": 1.07422690447585,
      "learning_rate": 3.397663498585747e-06,
      "loss": 0.2091,
      "step": 794
    },
    {
      "epoch": 0.41441324037271127,
      "grad_norm": 0.9652362696809133,
      "learning_rate": 3.39363899468469e-06,
      "loss": 0.1996,
      "step": 795
    },
    {
      "epoch": 0.41493451488890337,
      "grad_norm": 0.9534402042629027,
      "learning_rate": 3.3896118339460635e-06,
      "loss": 0.1855,
      "step": 796
    },
    {
      "epoch": 0.4154557894050955,
      "grad_norm": 1.0497663522609355,
      "learning_rate": 3.385582028342837e-06,
      "loss": 0.2087,
      "step": 797
    },
    {
      "epoch": 0.4159770639212875,
      "grad_norm": 1.0863225389912796,
      "learning_rate": 3.3815495898558424e-06,
      "loss": 0.2102,
      "step": 798
    },
    {
      "epoch": 0.41649833843747963,
      "grad_norm": 0.9880252641524678,
      "learning_rate": 3.377514530473739e-06,
      "loss": 0.1834,
      "step": 799
    },
    {
      "epoch": 0.41701961295367174,
      "grad_norm": 1.0933466253103774,
      "learning_rate": 3.3734768621929805e-06,
      "loss": 0.1974,
      "step": 800
    },
    {
      "epoch": 0.41754088746986384,
      "grad_norm": 1.0467600398315684,
      "learning_rate": 3.369436597017774e-06,
      "loss": 0.1978,
      "step": 801
    },
    {
      "epoch": 0.4180621619860559,
      "grad_norm": 1.0397173376484643,
      "learning_rate": 3.3653937469600483e-06,
      "loss": 0.2013,
      "step": 802
    },
    {
      "epoch": 0.418583436502248,
      "grad_norm": 0.9815507411522406,
      "learning_rate": 3.361348324039419e-06,
      "loss": 0.1954,
      "step": 803
    },
    {
      "epoch": 0.4191047110184401,
      "grad_norm": 0.9774944041763228,
      "learning_rate": 3.3573003402831487e-06,
      "loss": 0.1963,
      "step": 804
    },
    {
      "epoch": 0.41962598553463215,
      "grad_norm": 1.1198416721294486,
      "learning_rate": 3.353249807726115e-06,
      "loss": 0.2094,
      "step": 805
    },
    {
      "epoch": 0.42014726005082426,
      "grad_norm": 1.0790995136876083,
      "learning_rate": 3.349196738410771e-06,
      "loss": 0.2095,
      "step": 806
    },
    {
      "epoch": 0.42066853456701636,
      "grad_norm": 1.0111985446162226,
      "learning_rate": 3.3451411443871145e-06,
      "loss": 0.2022,
      "step": 807
    },
    {
      "epoch": 0.42118980908320847,
      "grad_norm": 1.1014100834244736,
      "learning_rate": 3.341083037712649e-06,
      "loss": 0.2184,
      "step": 808
    },
    {
      "epoch": 0.4217110835994005,
      "grad_norm": 1.001885982166375,
      "learning_rate": 3.337022430452346e-06,
      "loss": 0.1967,
      "step": 809
    },
    {
      "epoch": 0.4222323581155926,
      "grad_norm": 0.9377950692781114,
      "learning_rate": 3.3329593346786125e-06,
      "loss": 0.1877,
      "step": 810
    },
    {
      "epoch": 0.4227536326317847,
      "grad_norm": 1.096886871024171,
      "learning_rate": 3.328893762471255e-06,
      "loss": 0.2083,
      "step": 811
    },
    {
      "epoch": 0.42327490714797683,
      "grad_norm": 1.0227673076035053,
      "learning_rate": 3.324825725917442e-06,
      "loss": 0.1969,
      "step": 812
    },
    {
      "epoch": 0.4237961816641689,
      "grad_norm": 1.129454857460604,
      "learning_rate": 3.320755237111669e-06,
      "loss": 0.2024,
      "step": 813
    },
    {
      "epoch": 0.424317456180361,
      "grad_norm": 1.0491617495855319,
      "learning_rate": 3.316682308155721e-06,
      "loss": 0.2061,
      "step": 814
    },
    {
      "epoch": 0.4248387306965531,
      "grad_norm": 1.1595335723790747,
      "learning_rate": 3.312606951158638e-06,
      "loss": 0.2134,
      "step": 815
    },
    {
      "epoch": 0.42536000521274514,
      "grad_norm": 1.1081813559550924,
      "learning_rate": 3.308529178236679e-06,
      "loss": 0.1918,
      "step": 816
    },
    {
      "epoch": 0.42588127972893725,
      "grad_norm": 1.0946308608610222,
      "learning_rate": 3.304449001513287e-06,
      "loss": 0.1989,
      "step": 817
    },
    {
      "epoch": 0.42640255424512935,
      "grad_norm": 1.075048101820695,
      "learning_rate": 3.3003664331190487e-06,
      "loss": 0.1989,
      "step": 818
    },
    {
      "epoch": 0.42692382876132146,
      "grad_norm": 1.0734187102713864,
      "learning_rate": 3.296281485191665e-06,
      "loss": 0.1978,
      "step": 819
    },
    {
      "epoch": 0.4274451032775135,
      "grad_norm": 1.130789104429466,
      "learning_rate": 3.292194169875908e-06,
      "loss": 0.215,
      "step": 820
    },
    {
      "epoch": 0.4279663777937056,
      "grad_norm": 1.0291523352304348,
      "learning_rate": 3.2881044993235893e-06,
      "loss": 0.2011,
      "step": 821
    },
    {
      "epoch": 0.4284876523098977,
      "grad_norm": 1.0057993810989465,
      "learning_rate": 3.284012485693524e-06,
      "loss": 0.2068,
      "step": 822
    },
    {
      "epoch": 0.42900892682608976,
      "grad_norm": 1.046396865883086,
      "learning_rate": 3.279918141151492e-06,
      "loss": 0.2105,
      "step": 823
    },
    {
      "epoch": 0.42953020134228187,
      "grad_norm": 0.9922656623291862,
      "learning_rate": 3.2758214778702026e-06,
      "loss": 0.1845,
      "step": 824
    },
    {
      "epoch": 0.430051475858474,
      "grad_norm": 1.0565367939693868,
      "learning_rate": 3.2717225080292598e-06,
      "loss": 0.2034,
      "step": 825
    },
    {
      "epoch": 0.4305727503746661,
      "grad_norm": 1.058708383674746,
      "learning_rate": 3.2676212438151256e-06,
      "loss": 0.1993,
      "step": 826
    },
    {
      "epoch": 0.43109402489085813,
      "grad_norm": 0.9815918898679389,
      "learning_rate": 3.2635176974210824e-06,
      "loss": 0.1974,
      "step": 827
    },
    {
      "epoch": 0.43161529940705023,
      "grad_norm": 1.1067842143201054,
      "learning_rate": 3.2594118810471982e-06,
      "loss": 0.2066,
      "step": 828
    },
    {
      "epoch": 0.43213657392324234,
      "grad_norm": 1.0059867992897973,
      "learning_rate": 3.2553038069002885e-06,
      "loss": 0.2067,
      "step": 829
    },
    {
      "epoch": 0.43265784843943444,
      "grad_norm": 1.0777338938551524,
      "learning_rate": 3.2511934871938825e-06,
      "loss": 0.2033,
      "step": 830
    },
    {
      "epoch": 0.4331791229556265,
      "grad_norm": 1.104533496737417,
      "learning_rate": 3.247080934148186e-06,
      "loss": 0.199,
      "step": 831
    },
    {
      "epoch": 0.4337003974718186,
      "grad_norm": 1.0144271815214954,
      "learning_rate": 3.242966159990044e-06,
      "loss": 0.201,
      "step": 832
    },
    {
      "epoch": 0.4342216719880107,
      "grad_norm": 1.0028965037997253,
      "learning_rate": 3.238849176952904e-06,
      "loss": 0.1937,
      "step": 833
    },
    {
      "epoch": 0.43474294650420275,
      "grad_norm": 1.0804865893646987,
      "learning_rate": 3.2347299972767824e-06,
      "loss": 0.1908,
      "step": 834
    },
    {
      "epoch": 0.43526422102039486,
      "grad_norm": 1.1742256638405077,
      "learning_rate": 3.230608633208225e-06,
      "loss": 0.1956,
      "step": 835
    },
    {
      "epoch": 0.43578549553658696,
      "grad_norm": 1.0960493756055618,
      "learning_rate": 3.226485097000273e-06,
      "loss": 0.195,
      "step": 836
    },
    {
      "epoch": 0.43630677005277907,
      "grad_norm": 1.1756701100327795,
      "learning_rate": 3.2223594009124247e-06,
      "loss": 0.2202,
      "step": 837
    },
    {
      "epoch": 0.4368280445689711,
      "grad_norm": 0.962954960634096,
      "learning_rate": 3.2182315572105995e-06,
      "loss": 0.1916,
      "step": 838
    },
    {
      "epoch": 0.4373493190851632,
      "grad_norm": 1.096035560134522,
      "learning_rate": 3.2141015781671025e-06,
      "loss": 0.2037,
      "step": 839
    },
    {
      "epoch": 0.43787059360135533,
      "grad_norm": 1.1022461563067543,
      "learning_rate": 3.209969476060587e-06,
      "loss": 0.1969,
      "step": 840
    },
    {
      "epoch": 0.4383918681175474,
      "grad_norm": 1.060989135942518,
      "learning_rate": 3.2058352631760198e-06,
      "loss": 0.216,
      "step": 841
    },
    {
      "epoch": 0.4389131426337395,
      "grad_norm": 1.0795040679735655,
      "learning_rate": 3.2016989518046397e-06,
      "loss": 0.1972,
      "step": 842
    },
    {
      "epoch": 0.4394344171499316,
      "grad_norm": 1.01155187615578,
      "learning_rate": 3.1975605542439276e-06,
      "loss": 0.1958,
      "step": 843
    },
    {
      "epoch": 0.4399556916661237,
      "grad_norm": 0.960593686621856,
      "learning_rate": 3.1934200827975654e-06,
      "loss": 0.2063,
      "step": 844
    },
    {
      "epoch": 0.44047696618231574,
      "grad_norm": 1.052348833707456,
      "learning_rate": 3.1892775497754014e-06,
      "loss": 0.1977,
      "step": 845
    },
    {
      "epoch": 0.44099824069850785,
      "grad_norm": 1.122152577144792,
      "learning_rate": 3.1851329674934116e-06,
      "loss": 0.2075,
      "step": 846
    },
    {
      "epoch": 0.44151951521469995,
      "grad_norm": 1.0257719745756821,
      "learning_rate": 3.1809863482736663e-06,
      "loss": 0.1982,
      "step": 847
    },
    {
      "epoch": 0.44204078973089206,
      "grad_norm": 0.9494321503502252,
      "learning_rate": 3.176837704444291e-06,
      "loss": 0.1961,
      "step": 848
    },
    {
      "epoch": 0.4425620642470841,
      "grad_norm": 1.1099523946512315,
      "learning_rate": 3.1726870483394312e-06,
      "loss": 0.2122,
      "step": 849
    },
    {
      "epoch": 0.4430833387632762,
      "grad_norm": 1.073386369078645,
      "learning_rate": 3.168534392299214e-06,
      "loss": 0.2044,
      "step": 850
    },
    {
      "epoch": 0.4436046132794683,
      "grad_norm": 0.9859176069316815,
      "learning_rate": 3.1643797486697116e-06,
      "loss": 0.1938,
      "step": 851
    },
    {
      "epoch": 0.44412588779566037,
      "grad_norm": 0.9386602709194345,
      "learning_rate": 3.1602231298029074e-06,
      "loss": 0.1754,
      "step": 852
    },
    {
      "epoch": 0.4446471623118525,
      "grad_norm": 0.9948808105890651,
      "learning_rate": 3.1560645480566566e-06,
      "loss": 0.1934,
      "step": 853
    },
    {
      "epoch": 0.4451684368280446,
      "grad_norm": 1.0814129445631278,
      "learning_rate": 3.15190401579465e-06,
      "loss": 0.2066,
      "step": 854
    },
    {
      "epoch": 0.4456897113442367,
      "grad_norm": 0.9719582900510382,
      "learning_rate": 3.1477415453863772e-06,
      "loss": 0.184,
      "step": 855
    },
    {
      "epoch": 0.44621098586042873,
      "grad_norm": 0.9329010427938074,
      "learning_rate": 3.143577149207091e-06,
      "loss": 0.1899,
      "step": 856
    },
    {
      "epoch": 0.44673226037662084,
      "grad_norm": 0.9921945412172163,
      "learning_rate": 3.139410839637767e-06,
      "loss": 0.1962,
      "step": 857
    },
    {
      "epoch": 0.44725353489281294,
      "grad_norm": 1.0485901926630037,
      "learning_rate": 3.135242629065073e-06,
      "loss": 0.2018,
      "step": 858
    },
    {
      "epoch": 0.447774809409005,
      "grad_norm": 1.0379569766914887,
      "learning_rate": 3.131072529881326e-06,
      "loss": 0.2001,
      "step": 859
    },
    {
      "epoch": 0.4482960839251971,
      "grad_norm": 1.0305130275174221,
      "learning_rate": 3.126900554484459e-06,
      "loss": 0.1917,
      "step": 860
    },
    {
      "epoch": 0.4488173584413892,
      "grad_norm": 1.0437881640236402,
      "learning_rate": 3.122726715277983e-06,
      "loss": 0.1945,
      "step": 861
    },
    {
      "epoch": 0.4493386329575813,
      "grad_norm": 0.9799689803224048,
      "learning_rate": 3.1185510246709487e-06,
      "loss": 0.1887,
      "step": 862
    },
    {
      "epoch": 0.44985990747377336,
      "grad_norm": 1.0533266466186548,
      "learning_rate": 3.1143734950779155e-06,
      "loss": 0.201,
      "step": 863
    },
    {
      "epoch": 0.45038118198996546,
      "grad_norm": 1.004136035506477,
      "learning_rate": 3.1101941389189045e-06,
      "loss": 0.1978,
      "step": 864
    },
    {
      "epoch": 0.45090245650615757,
      "grad_norm": 1.0401781545267712,
      "learning_rate": 3.106012968619371e-06,
      "loss": 0.2033,
      "step": 865
    },
    {
      "epoch": 0.4514237310223497,
      "grad_norm": 1.0734661861751236,
      "learning_rate": 3.1018299966101624e-06,
      "loss": 0.2013,
      "step": 866
    },
    {
      "epoch": 0.4519450055385417,
      "grad_norm": 1.0052701698748918,
      "learning_rate": 3.097645235327483e-06,
      "loss": 0.187,
      "step": 867
    },
    {
      "epoch": 0.4524662800547338,
      "grad_norm": 1.0362538608122422,
      "learning_rate": 3.0934586972128574e-06,
      "loss": 0.197,
      "step": 868
    },
    {
      "epoch": 0.45298755457092593,
      "grad_norm": 1.060588394413901,
      "learning_rate": 3.0892703947130914e-06,
      "loss": 0.184,
      "step": 869
    },
    {
      "epoch": 0.453508829087118,
      "grad_norm": 1.1062613356935505,
      "learning_rate": 3.085080340280239e-06,
      "loss": 0.1827,
      "step": 870
    },
    {
      "epoch": 0.4540301036033101,
      "grad_norm": 0.983203748560144,
      "learning_rate": 3.0808885463715584e-06,
      "loss": 0.1935,
      "step": 871
    },
    {
      "epoch": 0.4545513781195022,
      "grad_norm": 0.9737122112212313,
      "learning_rate": 3.076695025449484e-06,
      "loss": 0.1976,
      "step": 872
    },
    {
      "epoch": 0.4550726526356943,
      "grad_norm": 1.0014155379159426,
      "learning_rate": 3.072499789981582e-06,
      "loss": 0.1962,
      "step": 873
    },
    {
      "epoch": 0.45559392715188635,
      "grad_norm": 1.0341459320964297,
      "learning_rate": 3.068302852440517e-06,
      "loss": 0.2036,
      "step": 874
    },
    {
      "epoch": 0.45611520166807845,
      "grad_norm": 1.0894399997153617,
      "learning_rate": 3.064104225304013e-06,
      "loss": 0.2023,
      "step": 875
    },
    {
      "epoch": 0.45663647618427056,
      "grad_norm": 1.0494724208473707,
      "learning_rate": 3.059903921054818e-06,
      "loss": 0.2098,
      "step": 876
    },
    {
      "epoch": 0.4571577507004626,
      "grad_norm": 0.9897158428149795,
      "learning_rate": 3.0557019521806667e-06,
      "loss": 0.1967,
      "step": 877
    },
    {
      "epoch": 0.4576790252166547,
      "grad_norm": 1.0264984042834973,
      "learning_rate": 3.0514983311742426e-06,
      "loss": 0.1978,
      "step": 878
    },
    {
      "epoch": 0.4582002997328468,
      "grad_norm": 1.0461074830254988,
      "learning_rate": 3.04729307053314e-06,
      "loss": 0.2004,
      "step": 879
    },
    {
      "epoch": 0.4587215742490389,
      "grad_norm": 1.051395189628469,
      "learning_rate": 3.0430861827598277e-06,
      "loss": 0.2061,
      "step": 880
    },
    {
      "epoch": 0.45924284876523097,
      "grad_norm": 0.9590732108822055,
      "learning_rate": 3.0388776803616138e-06,
      "loss": 0.1846,
      "step": 881
    },
    {
      "epoch": 0.4597641232814231,
      "grad_norm": 1.0023302855138927,
      "learning_rate": 3.034667575850607e-06,
      "loss": 0.186,
      "step": 882
    },
    {
      "epoch": 0.4602853977976152,
      "grad_norm": 1.0649714751792867,
      "learning_rate": 3.0304558817436767e-06,
      "loss": 0.2038,
      "step": 883
    },
    {
      "epoch": 0.4608066723138073,
      "grad_norm": 1.0031794140387544,
      "learning_rate": 3.02624261056242e-06,
      "loss": 0.1934,
      "step": 884
    },
    {
      "epoch": 0.46132794682999934,
      "grad_norm": 0.9583181588579257,
      "learning_rate": 3.0220277748331223e-06,
      "loss": 0.1846,
      "step": 885
    },
    {
      "epoch": 0.46184922134619144,
      "grad_norm": 1.10019469721715,
      "learning_rate": 3.017811387086721e-06,
      "loss": 0.1889,
      "step": 886
    },
    {
      "epoch": 0.46237049586238355,
      "grad_norm": 1.0282877540715634,
      "learning_rate": 3.013593459858767e-06,
      "loss": 0.176,
      "step": 887
    },
    {
      "epoch": 0.4628917703785756,
      "grad_norm": 0.9807399162917946,
      "learning_rate": 3.0093740056893882e-06,
      "loss": 0.1887,
      "step": 888
    },
    {
      "epoch": 0.4634130448947677,
      "grad_norm": 0.9486313009420617,
      "learning_rate": 3.005153037123253e-06,
      "loss": 0.1909,
      "step": 889
    },
    {
      "epoch": 0.4639343194109598,
      "grad_norm": 1.1228149064440365,
      "learning_rate": 3.000930566709531e-06,
      "loss": 0.1968,
      "step": 890
    },
    {
      "epoch": 0.4644555939271519,
      "grad_norm": 0.9838279802105361,
      "learning_rate": 2.996706607001858e-06,
      "loss": 0.1984,
      "step": 891
    },
    {
      "epoch": 0.46497686844334396,
      "grad_norm": 1.0285867670220825,
      "learning_rate": 2.9924811705582966e-06,
      "loss": 0.1898,
      "step": 892
    },
    {
      "epoch": 0.46549814295953607,
      "grad_norm": 1.0748194071023658,
      "learning_rate": 2.988254269941302e-06,
      "loss": 0.2047,
      "step": 893
    },
    {
      "epoch": 0.46601941747572817,
      "grad_norm": 0.9802690541168912,
      "learning_rate": 2.984025917717678e-06,
      "loss": 0.1903,
      "step": 894
    },
    {
      "epoch": 0.4665406919919202,
      "grad_norm": 1.0156827732220404,
      "learning_rate": 2.979796126458548e-06,
      "loss": 0.1996,
      "step": 895
    },
    {
      "epoch": 0.4670619665081123,
      "grad_norm": 1.036042762325025,
      "learning_rate": 2.975564908739313e-06,
      "loss": 0.1927,
      "step": 896
    },
    {
      "epoch": 0.46758324102430443,
      "grad_norm": 0.9790007561286332,
      "learning_rate": 2.9713322771396147e-06,
      "loss": 0.1947,
      "step": 897
    },
    {
      "epoch": 0.46810451554049654,
      "grad_norm": 0.9961185982546604,
      "learning_rate": 2.967098244243297e-06,
      "loss": 0.1938,
      "step": 898
    },
    {
      "epoch": 0.4686257900566886,
      "grad_norm": 0.9990853654605351,
      "learning_rate": 2.962862822638372e-06,
      "loss": 0.2085,
      "step": 899
    },
    {
      "epoch": 0.4691470645728807,
      "grad_norm": 1.0057041227395545,
      "learning_rate": 2.95862602491698e-06,
      "loss": 0.177,
      "step": 900
    },
    {
      "epoch": 0.4696683390890728,
      "grad_norm": 1.0330314786302779,
      "learning_rate": 2.9543878636753514e-06,
      "loss": 0.2002,
      "step": 901
    },
    {
      "epoch": 0.4701896136052649,
      "grad_norm": 1.0126061115045026,
      "learning_rate": 2.950148351513771e-06,
      "loss": 0.207,
      "step": 902
    },
    {
      "epoch": 0.47071088812145695,
      "grad_norm": 1.012115005303386,
      "learning_rate": 2.9459075010365406e-06,
      "loss": 0.1902,
      "step": 903
    },
    {
      "epoch": 0.47123216263764905,
      "grad_norm": 1.0790139527402576,
      "learning_rate": 2.9416653248519404e-06,
      "loss": 0.1977,
      "step": 904
    },
    {
      "epoch": 0.47175343715384116,
      "grad_norm": 0.9547667716226275,
      "learning_rate": 2.9374218355721925e-06,
      "loss": 0.1864,
      "step": 905
    },
    {
      "epoch": 0.4722747116700332,
      "grad_norm": 0.9569343156592153,
      "learning_rate": 2.933177045813421e-06,
      "loss": 0.1872,
      "step": 906
    },
    {
      "epoch": 0.4727959861862253,
      "grad_norm": 1.0609203952713406,
      "learning_rate": 2.9289309681956194e-06,
      "loss": 0.2047,
      "step": 907
    },
    {
      "epoch": 0.4733172607024174,
      "grad_norm": 1.0431365337991538,
      "learning_rate": 2.924683615342607e-06,
      "loss": 0.1969,
      "step": 908
    },
    {
      "epoch": 0.4738385352186095,
      "grad_norm": 1.0527870765085736,
      "learning_rate": 2.920434999881998e-06,
      "loss": 0.2129,
      "step": 909
    },
    {
      "epoch": 0.4743598097348016,
      "grad_norm": 0.9369500309949982,
      "learning_rate": 2.9161851344451563e-06,
      "loss": 0.1817,
      "step": 910
    },
    {
      "epoch": 0.4748810842509937,
      "grad_norm": 0.9738920669778763,
      "learning_rate": 2.9119340316671663e-06,
      "loss": 0.1979,
      "step": 911
    },
    {
      "epoch": 0.4754023587671858,
      "grad_norm": 0.9329665064906625,
      "learning_rate": 2.9076817041867863e-06,
      "loss": 0.1922,
      "step": 912
    },
    {
      "epoch": 0.47592363328337783,
      "grad_norm": 0.9597372603106097,
      "learning_rate": 2.9034281646464197e-06,
      "loss": 0.1732,
      "step": 913
    },
    {
      "epoch": 0.47644490779956994,
      "grad_norm": 1.0198109368789043,
      "learning_rate": 2.8991734256920723e-06,
      "loss": 0.2053,
      "step": 914
    },
    {
      "epoch": 0.47696618231576204,
      "grad_norm": 0.950279779039853,
      "learning_rate": 2.894917499973315e-06,
      "loss": 0.1949,
      "step": 915
    },
    {
      "epoch": 0.47748745683195415,
      "grad_norm": 1.061855943908999,
      "learning_rate": 2.890660400143248e-06,
      "loss": 0.1928,
      "step": 916
    },
    {
      "epoch": 0.4780087313481462,
      "grad_norm": 1.094038115550398,
      "learning_rate": 2.8864021388584606e-06,
      "loss": 0.2108,
      "step": 917
    },
    {
      "epoch": 0.4785300058643383,
      "grad_norm": 0.962986987019024,
      "learning_rate": 2.882142728778997e-06,
      "loss": 0.2022,
      "step": 918
    },
    {
      "epoch": 0.4790512803805304,
      "grad_norm": 0.9886755913465457,
      "learning_rate": 2.877882182568317e-06,
      "loss": 0.1911,
      "step": 919
    },
    {
      "epoch": 0.4795725548967225,
      "grad_norm": 0.9420184069269022,
      "learning_rate": 2.873620512893257e-06,
      "loss": 0.1821,
      "step": 920
    },
    {
      "epoch": 0.48009382941291456,
      "grad_norm": 0.9907105831800497,
      "learning_rate": 2.8693577324239925e-06,
      "loss": 0.1901,
      "step": 921
    },
    {
      "epoch": 0.48061510392910667,
      "grad_norm": 1.1281804493203011,
      "learning_rate": 2.865093853834004e-06,
      "loss": 0.1941,
      "step": 922
    },
    {
      "epoch": 0.4811363784452988,
      "grad_norm": 1.0324726305012064,
      "learning_rate": 2.8608288898000356e-06,
      "loss": 0.2062,
      "step": 923
    },
    {
      "epoch": 0.4816576529614908,
      "grad_norm": 0.9998082294529183,
      "learning_rate": 2.8565628530020584e-06,
      "loss": 0.1969,
      "step": 924
    },
    {
      "epoch": 0.48217892747768293,
      "grad_norm": 0.9410194841828444,
      "learning_rate": 2.8522957561232323e-06,
      "loss": 0.1889,
      "step": 925
    },
    {
      "epoch": 0.48270020199387503,
      "grad_norm": 0.9981084233938012,
      "learning_rate": 2.84802761184987e-06,
      "loss": 0.1923,
      "step": 926
    },
    {
      "epoch": 0.48322147651006714,
      "grad_norm": 1.009744428574062,
      "learning_rate": 2.8437584328713976e-06,
      "loss": 0.2002,
      "step": 927
    },
    {
      "epoch": 0.4837427510262592,
      "grad_norm": 1.0191370091693872,
      "learning_rate": 2.8394882318803174e-06,
      "loss": 0.1917,
      "step": 928
    },
    {
      "epoch": 0.4842640255424513,
      "grad_norm": 0.9907623471199559,
      "learning_rate": 2.835217021572171e-06,
      "loss": 0.1932,
      "step": 929
    },
    {
      "epoch": 0.4847853000586434,
      "grad_norm": 0.9493877146289591,
      "learning_rate": 2.8309448146454993e-06,
      "loss": 0.1861,
      "step": 930
    },
    {
      "epoch": 0.48530657457483545,
      "grad_norm": 0.9851626213583532,
      "learning_rate": 2.8266716238018065e-06,
      "loss": 0.1877,
      "step": 931
    },
    {
      "epoch": 0.48582784909102755,
      "grad_norm": 1.0361940174011144,
      "learning_rate": 2.822397461745524e-06,
      "loss": 0.2025,
      "step": 932
    },
    {
      "epoch": 0.48634912360721966,
      "grad_norm": 1.0421742004935162,
      "learning_rate": 2.8181223411839686e-06,
      "loss": 0.2009,
      "step": 933
    },
    {
      "epoch": 0.48687039812341176,
      "grad_norm": 1.0040799179404327,
      "learning_rate": 2.8138462748273072e-06,
      "loss": 0.1994,
      "step": 934
    },
    {
      "epoch": 0.4873916726396038,
      "grad_norm": 0.9213913695963329,
      "learning_rate": 2.8095692753885177e-06,
      "loss": 0.1752,
      "step": 935
    },
    {
      "epoch": 0.4879129471557959,
      "grad_norm": 1.1012263545067054,
      "learning_rate": 2.805291355583355e-06,
      "loss": 0.2082,
      "step": 936
    },
    {
      "epoch": 0.488434221671988,
      "grad_norm": 1.0309678892259206,
      "learning_rate": 2.8010125281303076e-06,
      "loss": 0.1888,
      "step": 937
    },
    {
      "epoch": 0.4889554961881801,
      "grad_norm": 0.9436212370983429,
      "learning_rate": 2.7967328057505637e-06,
      "loss": 0.1876,
      "step": 938
    },
    {
      "epoch": 0.4894767707043722,
      "grad_norm": 1.0588743350687662,
      "learning_rate": 2.792452201167971e-06,
      "loss": 0.194,
      "step": 939
    },
    {
      "epoch": 0.4899980452205643,
      "grad_norm": 1.0231800414996428,
      "learning_rate": 2.7881707271090018e-06,
      "loss": 0.2057,
      "step": 940
    },
    {
      "epoch": 0.4905193197367564,
      "grad_norm": 1.0270862744387954,
      "learning_rate": 2.7838883963027118e-06,
      "loss": 0.2023,
      "step": 941
    },
    {
      "epoch": 0.49104059425294844,
      "grad_norm": 1.0566276299374076,
      "learning_rate": 2.779605221480706e-06,
      "loss": 0.2014,
      "step": 942
    },
    {
      "epoch": 0.49156186876914054,
      "grad_norm": 1.0569768214284592,
      "learning_rate": 2.7753212153770947e-06,
      "loss": 0.1861,
      "step": 943
    },
    {
      "epoch": 0.49208314328533265,
      "grad_norm": 0.9787126303296814,
      "learning_rate": 2.7710363907284643e-06,
      "loss": 0.1863,
      "step": 944
    },
    {
      "epoch": 0.49260441780152475,
      "grad_norm": 0.9767411490508309,
      "learning_rate": 2.766750760273831e-06,
      "loss": 0.188,
      "step": 945
    },
    {
      "epoch": 0.4931256923177168,
      "grad_norm": 1.0939658351889692,
      "learning_rate": 2.76246433675461e-06,
      "loss": 0.1951,
      "step": 946
    },
    {
      "epoch": 0.4936469668339089,
      "grad_norm": 1.172523118347166,
      "learning_rate": 2.7581771329145713e-06,
      "loss": 0.203,
      "step": 947
    },
    {
      "epoch": 0.494168241350101,
      "grad_norm": 1.0061442496305424,
      "learning_rate": 2.7538891614998074e-06,
      "loss": 0.2,
      "step": 948
    },
    {
      "epoch": 0.49468951586629306,
      "grad_norm": 1.041944208690642,
      "learning_rate": 2.749600435258691e-06,
      "loss": 0.197,
      "step": 949
    },
    {
      "epoch": 0.49521079038248517,
      "grad_norm": 1.0130983804244988,
      "learning_rate": 2.745310966941839e-06,
      "loss": 0.1904,
      "step": 950
    },
    {
      "epoch": 0.49573206489867727,
      "grad_norm": 1.0351990108914926,
      "learning_rate": 2.741020769302077e-06,
      "loss": 0.1853,
      "step": 951
    },
    {
      "epoch": 0.4962533394148694,
      "grad_norm": 1.0210238359665258,
      "learning_rate": 2.7367298550943954e-06,
      "loss": 0.1875,
      "step": 952
    },
    {
      "epoch": 0.4967746139310614,
      "grad_norm": 0.9970390003149875,
      "learning_rate": 2.7324382370759174e-06,
      "loss": 0.1855,
      "step": 953
    },
    {
      "epoch": 0.49729588844725353,
      "grad_norm": 1.0879513566840098,
      "learning_rate": 2.7281459280058563e-06,
      "loss": 0.2108,
      "step": 954
    },
    {
      "epoch": 0.49781716296344564,
      "grad_norm": 1.1430957619253195,
      "learning_rate": 2.7238529406454834e-06,
      "loss": 0.1819,
      "step": 955
    },
    {
      "epoch": 0.49833843747963774,
      "grad_norm": 0.9771932290793619,
      "learning_rate": 2.719559287758085e-06,
      "loss": 0.1796,
      "step": 956
    },
    {
      "epoch": 0.4988597119958298,
      "grad_norm": 1.0436141215860535,
      "learning_rate": 2.7152649821089245e-06,
      "loss": 0.1943,
      "step": 957
    },
    {
      "epoch": 0.4993809865120219,
      "grad_norm": 0.9892708407682,
      "learning_rate": 2.7109700364652075e-06,
      "loss": 0.1905,
      "step": 958
    },
    {
      "epoch": 0.499902261028214,
      "grad_norm": 0.9916434111409589,
      "learning_rate": 2.7066744635960423e-06,
      "loss": 0.1858,
      "step": 959
    },
    {
      "epoch": 0.5004235355444061,
      "grad_norm": 0.9748778629471573,
      "learning_rate": 2.702378276272402e-06,
      "loss": 0.2101,
      "step": 960
    },
    {
      "epoch": 0.5009448100605982,
      "grad_norm": 1.0520038494784436,
      "learning_rate": 2.6980814872670863e-06,
      "loss": 0.1945,
      "step": 961
    },
    {
      "epoch": 0.5014660845767902,
      "grad_norm": 1.019315172372774,
      "learning_rate": 2.6937841093546842e-06,
      "loss": 0.1931,
      "step": 962
    },
    {
      "epoch": 0.5019873590929823,
      "grad_norm": 1.0290239491636743,
      "learning_rate": 2.6894861553115337e-06,
      "loss": 0.2038,
      "step": 963
    },
    {
      "epoch": 0.5025086336091744,
      "grad_norm": 0.9753619324088938,
      "learning_rate": 2.6851876379156884e-06,
      "loss": 0.1858,
      "step": 964
    },
    {
      "epoch": 0.5030299081253665,
      "grad_norm": 1.073883143404841,
      "learning_rate": 2.680888569946874e-06,
      "loss": 0.2225,
      "step": 965
    },
    {
      "epoch": 0.5035511826415586,
      "grad_norm": 0.9492011107722095,
      "learning_rate": 2.6765889641864562e-06,
      "loss": 0.1881,
      "step": 966
    },
    {
      "epoch": 0.5040724571577507,
      "grad_norm": 0.9673199856485143,
      "learning_rate": 2.6722888334173974e-06,
      "loss": 0.1936,
      "step": 967
    },
    {
      "epoch": 0.5045937316739428,
      "grad_norm": 1.0076521636183045,
      "learning_rate": 2.6679881904242198e-06,
      "loss": 0.2029,
      "step": 968
    },
    {
      "epoch": 0.5051150061901348,
      "grad_norm": 0.9686161019023214,
      "learning_rate": 2.663687047992972e-06,
      "loss": 0.1841,
      "step": 969
    },
    {
      "epoch": 0.5056362807063269,
      "grad_norm": 1.0250906895196519,
      "learning_rate": 2.6593854189111857e-06,
      "loss": 0.1981,
      "step": 970
    },
    {
      "epoch": 0.506157555222519,
      "grad_norm": 0.9616510423370317,
      "learning_rate": 2.655083315967838e-06,
      "loss": 0.1869,
      "step": 971
    },
    {
      "epoch": 0.5066788297387111,
      "grad_norm": 0.9930176109805637,
      "learning_rate": 2.650780751953316e-06,
      "loss": 0.1873,
      "step": 972
    },
    {
      "epoch": 0.5072001042549033,
      "grad_norm": 0.9967805328060277,
      "learning_rate": 2.646477739659378e-06,
      "loss": 0.1952,
      "step": 973
    },
    {
      "epoch": 0.5077213787710954,
      "grad_norm": 0.9871931304412768,
      "learning_rate": 2.6421742918791155e-06,
      "loss": 0.1983,
      "step": 974
    },
    {
      "epoch": 0.5082426532872875,
      "grad_norm": 1.0323661862589764,
      "learning_rate": 2.6378704214069133e-06,
      "loss": 0.1804,
      "step": 975
    },
    {
      "epoch": 0.5087639278034795,
      "grad_norm": 1.0229802891072548,
      "learning_rate": 2.633566141038413e-06,
      "loss": 0.2017,
      "step": 976
    },
    {
      "epoch": 0.5092852023196716,
      "grad_norm": 1.0240567212255771,
      "learning_rate": 2.629261463570476e-06,
      "loss": 0.1983,
      "step": 977
    },
    {
      "epoch": 0.5098064768358637,
      "grad_norm": 1.0186229069169201,
      "learning_rate": 2.6249564018011437e-06,
      "loss": 0.1878,
      "step": 978
    },
    {
      "epoch": 0.5103277513520558,
      "grad_norm": 0.9399329028559811,
      "learning_rate": 2.6206509685296e-06,
      "loss": 0.1957,
      "step": 979
    },
    {
      "epoch": 0.5108490258682479,
      "grad_norm": 1.0070801784398393,
      "learning_rate": 2.6163451765561324e-06,
      "loss": 0.1963,
      "step": 980
    },
    {
      "epoch": 0.51137030038444,
      "grad_norm": 0.9756463630914503,
      "learning_rate": 2.6120390386820975e-06,
      "loss": 0.1999,
      "step": 981
    },
    {
      "epoch": 0.5118915749006321,
      "grad_norm": 1.0959003656589381,
      "learning_rate": 2.607732567709877e-06,
      "loss": 0.1931,
      "step": 982
    },
    {
      "epoch": 0.5124128494168242,
      "grad_norm": 1.0262562081196058,
      "learning_rate": 2.6034257764428456e-06,
      "loss": 0.1997,
      "step": 983
    },
    {
      "epoch": 0.5129341239330162,
      "grad_norm": 0.9511566278900171,
      "learning_rate": 2.5991186776853277e-06,
      "loss": 0.1849,
      "step": 984
    },
    {
      "epoch": 0.5134553984492083,
      "grad_norm": 0.958443517567957,
      "learning_rate": 2.594811284242565e-06,
      "loss": 0.1783,
      "step": 985
    },
    {
      "epoch": 0.5139766729654004,
      "grad_norm": 0.9842691568175002,
      "learning_rate": 2.590503608920672e-06,
      "loss": 0.1944,
      "step": 986
    },
    {
      "epoch": 0.5144979474815925,
      "grad_norm": 1.0055481610107908,
      "learning_rate": 2.5861956645266036e-06,
      "loss": 0.1949,
      "step": 987
    },
    {
      "epoch": 0.5150192219977846,
      "grad_norm": 0.9527635866518458,
      "learning_rate": 2.581887463868114e-06,
      "loss": 0.1887,
      "step": 988
    },
    {
      "epoch": 0.5155404965139767,
      "grad_norm": 1.025332524858524,
      "learning_rate": 2.57757901975372e-06,
      "loss": 0.1996,
      "step": 989
    },
    {
      "epoch": 0.5160617710301688,
      "grad_norm": 1.0680557620822233,
      "learning_rate": 2.5732703449926595e-06,
      "loss": 0.2055,
      "step": 990
    },
    {
      "epoch": 0.5165830455463608,
      "grad_norm": 1.0380369075475906,
      "learning_rate": 2.568961452394859e-06,
      "loss": 0.1906,
      "step": 991
    },
    {
      "epoch": 0.5171043200625529,
      "grad_norm": 0.9456688282630512,
      "learning_rate": 2.564652354770892e-06,
      "loss": 0.1865,
      "step": 992
    },
    {
      "epoch": 0.517625594578745,
      "grad_norm": 1.0840975960140231,
      "learning_rate": 2.560343064931941e-06,
      "loss": 0.2152,
      "step": 993
    },
    {
      "epoch": 0.5181468690949371,
      "grad_norm": 0.9754317560093736,
      "learning_rate": 2.5560335956897603e-06,
      "loss": 0.1862,
      "step": 994
    },
    {
      "epoch": 0.5186681436111292,
      "grad_norm": 0.975666731751929,
      "learning_rate": 2.551723959856637e-06,
      "loss": 0.1889,
      "step": 995
    },
    {
      "epoch": 0.5191894181273213,
      "grad_norm": 0.9498987479759115,
      "learning_rate": 2.5474141702453536e-06,
      "loss": 0.1894,
      "step": 996
    },
    {
      "epoch": 0.5197106926435134,
      "grad_norm": 1.0287537604691805,
      "learning_rate": 2.543104239669152e-06,
      "loss": 0.2087,
      "step": 997
    },
    {
      "epoch": 0.5202319671597054,
      "grad_norm": 0.9942554093772231,
      "learning_rate": 2.5387941809416895e-06,
      "loss": 0.1795,
      "step": 998
    },
    {
      "epoch": 0.5207532416758975,
      "grad_norm": 1.0290989991698256,
      "learning_rate": 2.5344840068770076e-06,
      "loss": 0.1873,
      "step": 999
    },
    {
      "epoch": 0.5212745161920896,
      "grad_norm": 1.062500241588878,
      "learning_rate": 2.530173730289488e-06,
      "loss": 0.1946,
      "step": 1000
    },
    {
      "epoch": 0.5212745161920896,
      "eval_loss": 0.19480524957180023,
      "eval_runtime": 52.2734,
      "eval_samples_per_second": 23.741,
      "eval_steps_per_second": 2.984,
      "step": 1000
    },
    {
      "epoch": 0.5217957907082817,
      "grad_norm": 1.0383174062236034,
      "learning_rate": 2.5258633639938195e-06,
      "loss": 0.1995,
      "step": 1001
    },
    {
      "epoch": 0.5223170652244739,
      "grad_norm": 1.0619346174205255,
      "learning_rate": 2.521552920804956e-06,
      "loss": 0.2085,
      "step": 1002
    },
    {
      "epoch": 0.522838339740666,
      "grad_norm": 1.1587513971294219,
      "learning_rate": 2.5172424135380817e-06,
      "loss": 0.2047,
      "step": 1003
    },
    {
      "epoch": 0.5233596142568581,
      "grad_norm": 1.0254346556986893,
      "learning_rate": 2.51293185500857e-06,
      "loss": 0.1905,
      "step": 1004
    },
    {
      "epoch": 0.5238808887730501,
      "grad_norm": 1.007732855163628,
      "learning_rate": 2.5086212580319457e-06,
      "loss": 0.1995,
      "step": 1005
    },
    {
      "epoch": 0.5244021632892422,
      "grad_norm": 1.0065215395102691,
      "learning_rate": 2.50431063542385e-06,
      "loss": 0.1931,
      "step": 1006
    },
    {
      "epoch": 0.5249234378054343,
      "grad_norm": 1.1028563144125332,
      "learning_rate": 2.5e-06,
      "loss": 0.2001,
      "step": 1007
    },
    {
      "epoch": 0.5254447123216264,
      "grad_norm": 1.0547991596381279,
      "learning_rate": 2.49568936457615e-06,
      "loss": 0.197,
      "step": 1008
    },
    {
      "epoch": 0.5259659868378185,
      "grad_norm": 1.0501645268059348,
      "learning_rate": 2.491378741968055e-06,
      "loss": 0.1898,
      "step": 1009
    },
    {
      "epoch": 0.5264872613540106,
      "grad_norm": 1.0453727558531773,
      "learning_rate": 2.487068144991431e-06,
      "loss": 0.1935,
      "step": 1010
    },
    {
      "epoch": 0.5270085358702027,
      "grad_norm": 0.9784268100081941,
      "learning_rate": 2.4827575864619183e-06,
      "loss": 0.2068,
      "step": 1011
    },
    {
      "epoch": 0.5275298103863947,
      "grad_norm": 0.9489702371591282,
      "learning_rate": 2.4784470791950442e-06,
      "loss": 0.1952,
      "step": 1012
    },
    {
      "epoch": 0.5280510849025868,
      "grad_norm": 1.066317094816582,
      "learning_rate": 2.4741366360061813e-06,
      "loss": 0.1911,
      "step": 1013
    },
    {
      "epoch": 0.5285723594187789,
      "grad_norm": 1.0310100518300547,
      "learning_rate": 2.4698262697105128e-06,
      "loss": 0.1986,
      "step": 1014
    },
    {
      "epoch": 0.529093633934971,
      "grad_norm": 0.983376330407415,
      "learning_rate": 2.4655159931229932e-06,
      "loss": 0.1794,
      "step": 1015
    },
    {
      "epoch": 0.5296149084511631,
      "grad_norm": 0.9973472750235451,
      "learning_rate": 2.461205819058311e-06,
      "loss": 0.2025,
      "step": 1016
    },
    {
      "epoch": 0.5301361829673552,
      "grad_norm": 0.9963200987254586,
      "learning_rate": 2.4568957603308494e-06,
      "loss": 0.1891,
      "step": 1017
    },
    {
      "epoch": 0.5306574574835473,
      "grad_norm": 1.03843991512085,
      "learning_rate": 2.452585829754647e-06,
      "loss": 0.1957,
      "step": 1018
    },
    {
      "epoch": 0.5311787319997394,
      "grad_norm": 1.0284403625647769,
      "learning_rate": 2.448276040143364e-06,
      "loss": 0.212,
      "step": 1019
    },
    {
      "epoch": 0.5317000065159314,
      "grad_norm": 1.0375123811085925,
      "learning_rate": 2.4439664043102414e-06,
      "loss": 0.2078,
      "step": 1020
    },
    {
      "epoch": 0.5322212810321235,
      "grad_norm": 1.0154917946726163,
      "learning_rate": 2.43965693506806e-06,
      "loss": 0.1905,
      "step": 1021
    },
    {
      "epoch": 0.5327425555483156,
      "grad_norm": 1.0771383672005177,
      "learning_rate": 2.4353476452291086e-06,
      "loss": 0.1882,
      "step": 1022
    },
    {
      "epoch": 0.5332638300645077,
      "grad_norm": 1.0576240219047033,
      "learning_rate": 2.431038547605142e-06,
      "loss": 0.1753,
      "step": 1023
    },
    {
      "epoch": 0.5337851045806998,
      "grad_norm": 1.0111187689876087,
      "learning_rate": 2.4267296550073413e-06,
      "loss": 0.1985,
      "step": 1024
    },
    {
      "epoch": 0.5343063790968919,
      "grad_norm": 1.0250830102181434,
      "learning_rate": 2.4224209802462818e-06,
      "loss": 0.1889,
      "step": 1025
    },
    {
      "epoch": 0.534827653613084,
      "grad_norm": 1.0333860347256845,
      "learning_rate": 2.4181125361318868e-06,
      "loss": 0.1883,
      "step": 1026
    },
    {
      "epoch": 0.535348928129276,
      "grad_norm": 1.052906861325362,
      "learning_rate": 2.413804335473397e-06,
      "loss": 0.1869,
      "step": 1027
    },
    {
      "epoch": 0.5358702026454681,
      "grad_norm": 1.0145844359267704,
      "learning_rate": 2.409496391079329e-06,
      "loss": 0.1856,
      "step": 1028
    },
    {
      "epoch": 0.5363914771616602,
      "grad_norm": 1.0461114420942073,
      "learning_rate": 2.4051887157574356e-06,
      "loss": 0.1971,
      "step": 1029
    },
    {
      "epoch": 0.5369127516778524,
      "grad_norm": 0.975138981928165,
      "learning_rate": 2.4008813223146723e-06,
      "loss": 0.1774,
      "step": 1030
    },
    {
      "epoch": 0.5374340261940445,
      "grad_norm": 1.079976404735243,
      "learning_rate": 2.3965742235571557e-06,
      "loss": 0.2078,
      "step": 1031
    },
    {
      "epoch": 0.5379553007102366,
      "grad_norm": 1.0856031527562608,
      "learning_rate": 2.3922674322901236e-06,
      "loss": 0.2026,
      "step": 1032
    },
    {
      "epoch": 0.5384765752264287,
      "grad_norm": 1.0705003957337433,
      "learning_rate": 2.387960961317903e-06,
      "loss": 0.2027,
      "step": 1033
    },
    {
      "epoch": 0.5389978497426207,
      "grad_norm": 1.0248865414814123,
      "learning_rate": 2.383654823443868e-06,
      "loss": 0.1992,
      "step": 1034
    },
    {
      "epoch": 0.5395191242588128,
      "grad_norm": 1.0290121022581171,
      "learning_rate": 2.3793490314704005e-06,
      "loss": 0.1914,
      "step": 1035
    },
    {
      "epoch": 0.5400403987750049,
      "grad_norm": 1.1076737668760843,
      "learning_rate": 2.3750435981988576e-06,
      "loss": 0.2115,
      "step": 1036
    },
    {
      "epoch": 0.540561673291197,
      "grad_norm": 1.0214795198933087,
      "learning_rate": 2.3707385364295245e-06,
      "loss": 0.2009,
      "step": 1037
    },
    {
      "epoch": 0.5410829478073891,
      "grad_norm": 0.9830881085492219,
      "learning_rate": 2.366433858961587e-06,
      "loss": 0.195,
      "step": 1038
    },
    {
      "epoch": 0.5416042223235812,
      "grad_norm": 1.0182317451712537,
      "learning_rate": 2.362129578593088e-06,
      "loss": 0.1926,
      "step": 1039
    },
    {
      "epoch": 0.5421254968397733,
      "grad_norm": 1.0927879125581443,
      "learning_rate": 2.3578257081208853e-06,
      "loss": 0.1947,
      "step": 1040
    },
    {
      "epoch": 0.5426467713559653,
      "grad_norm": 1.1605323377220922,
      "learning_rate": 2.3535222603406223e-06,
      "loss": 0.2107,
      "step": 1041
    },
    {
      "epoch": 0.5431680458721574,
      "grad_norm": 0.994353490468453,
      "learning_rate": 2.3492192480466845e-06,
      "loss": 0.1857,
      "step": 1042
    },
    {
      "epoch": 0.5436893203883495,
      "grad_norm": 0.9673106256287484,
      "learning_rate": 2.344916684032163e-06,
      "loss": 0.1784,
      "step": 1043
    },
    {
      "epoch": 0.5442105949045416,
      "grad_norm": 1.0775673476430663,
      "learning_rate": 2.3406145810888143e-06,
      "loss": 0.1993,
      "step": 1044
    },
    {
      "epoch": 0.5447318694207337,
      "grad_norm": 1.0071909867214548,
      "learning_rate": 2.3363129520070286e-06,
      "loss": 0.1858,
      "step": 1045
    },
    {
      "epoch": 0.5452531439369258,
      "grad_norm": 1.0486545569572066,
      "learning_rate": 2.3320118095757806e-06,
      "loss": 0.2105,
      "step": 1046
    },
    {
      "epoch": 0.5457744184531179,
      "grad_norm": 1.081028929707012,
      "learning_rate": 2.327711166582604e-06,
      "loss": 0.1985,
      "step": 1047
    },
    {
      "epoch": 0.5462956929693099,
      "grad_norm": 1.037851463032111,
      "learning_rate": 2.3234110358135446e-06,
      "loss": 0.1802,
      "step": 1048
    },
    {
      "epoch": 0.546816967485502,
      "grad_norm": 0.9955744873005768,
      "learning_rate": 2.319111430053126e-06,
      "loss": 0.19,
      "step": 1049
    },
    {
      "epoch": 0.5473382420016941,
      "grad_norm": 0.9625839485821711,
      "learning_rate": 2.3148123620843132e-06,
      "loss": 0.1903,
      "step": 1050
    },
    {
      "epoch": 0.5478595165178862,
      "grad_norm": 0.98467636090744,
      "learning_rate": 2.310513844688467e-06,
      "loss": 0.1858,
      "step": 1051
    },
    {
      "epoch": 0.5483807910340783,
      "grad_norm": 1.0085351128724585,
      "learning_rate": 2.306215890645316e-06,
      "loss": 0.1864,
      "step": 1052
    },
    {
      "epoch": 0.5489020655502704,
      "grad_norm": 0.997147524038755,
      "learning_rate": 2.3019185127329145e-06,
      "loss": 0.1871,
      "step": 1053
    },
    {
      "epoch": 0.5494233400664625,
      "grad_norm": 1.0495981467979099,
      "learning_rate": 2.2976217237275983e-06,
      "loss": 0.1896,
      "step": 1054
    },
    {
      "epoch": 0.5499446145826546,
      "grad_norm": 0.9845529574139438,
      "learning_rate": 2.293325536403958e-06,
      "loss": 0.2015,
      "step": 1055
    },
    {
      "epoch": 0.5504658890988466,
      "grad_norm": 1.052070335880451,
      "learning_rate": 2.2890299635347933e-06,
      "loss": 0.1978,
      "step": 1056
    },
    {
      "epoch": 0.5509871636150387,
      "grad_norm": 1.0192406532778389,
      "learning_rate": 2.2847350178910763e-06,
      "loss": 0.199,
      "step": 1057
    },
    {
      "epoch": 0.5515084381312308,
      "grad_norm": 0.9255815167749964,
      "learning_rate": 2.2804407122419165e-06,
      "loss": 0.1743,
      "step": 1058
    },
    {
      "epoch": 0.552029712647423,
      "grad_norm": 0.9425920036365858,
      "learning_rate": 2.276147059354517e-06,
      "loss": 0.1915,
      "step": 1059
    },
    {
      "epoch": 0.5525509871636151,
      "grad_norm": 0.9756852726978495,
      "learning_rate": 2.271854071994144e-06,
      "loss": 0.1851,
      "step": 1060
    },
    {
      "epoch": 0.5530722616798072,
      "grad_norm": 1.0319509062414018,
      "learning_rate": 2.2675617629240842e-06,
      "loss": 0.2026,
      "step": 1061
    },
    {
      "epoch": 0.5535935361959993,
      "grad_norm": 1.0469736880808636,
      "learning_rate": 2.2632701449056054e-06,
      "loss": 0.1985,
      "step": 1062
    },
    {
      "epoch": 0.5541148107121913,
      "grad_norm": 1.0154988381567833,
      "learning_rate": 2.258979230697923e-06,
      "loss": 0.184,
      "step": 1063
    },
    {
      "epoch": 0.5546360852283834,
      "grad_norm": 1.0706994099193394,
      "learning_rate": 2.2546890330581616e-06,
      "loss": 0.1861,
      "step": 1064
    },
    {
      "epoch": 0.5551573597445755,
      "grad_norm": 1.0465560677474248,
      "learning_rate": 2.25039956474131e-06,
      "loss": 0.1878,
      "step": 1065
    },
    {
      "epoch": 0.5556786342607676,
      "grad_norm": 1.029253640779406,
      "learning_rate": 2.246110838500194e-06,
      "loss": 0.1914,
      "step": 1066
    },
    {
      "epoch": 0.5561999087769597,
      "grad_norm": 1.0753625496287706,
      "learning_rate": 2.241822867085429e-06,
      "loss": 0.1935,
      "step": 1067
    },
    {
      "epoch": 0.5567211832931518,
      "grad_norm": 1.0894709655635972,
      "learning_rate": 2.2375356632453906e-06,
      "loss": 0.1893,
      "step": 1068
    },
    {
      "epoch": 0.5572424578093439,
      "grad_norm": 1.1026529187740728,
      "learning_rate": 2.2332492397261695e-06,
      "loss": 0.1951,
      "step": 1069
    },
    {
      "epoch": 0.5577637323255359,
      "grad_norm": 0.9641199594525882,
      "learning_rate": 2.2289636092715365e-06,
      "loss": 0.1834,
      "step": 1070
    },
    {
      "epoch": 0.558285006841728,
      "grad_norm": 1.0402753633451525,
      "learning_rate": 2.2246787846229057e-06,
      "loss": 0.1834,
      "step": 1071
    },
    {
      "epoch": 0.5588062813579201,
      "grad_norm": 1.081780488844143,
      "learning_rate": 2.2203947785192954e-06,
      "loss": 0.1878,
      "step": 1072
    },
    {
      "epoch": 0.5593275558741122,
      "grad_norm": 1.0234654215462535,
      "learning_rate": 2.2161116036972886e-06,
      "loss": 0.1943,
      "step": 1073
    },
    {
      "epoch": 0.5598488303903043,
      "grad_norm": 1.0411803692422346,
      "learning_rate": 2.2118292728909986e-06,
      "loss": 0.2069,
      "step": 1074
    },
    {
      "epoch": 0.5603701049064964,
      "grad_norm": 0.9713387589599493,
      "learning_rate": 2.2075477988320295e-06,
      "loss": 0.1849,
      "step": 1075
    },
    {
      "epoch": 0.5608913794226885,
      "grad_norm": 1.0246645842699202,
      "learning_rate": 2.2032671942494367e-06,
      "loss": 0.1844,
      "step": 1076
    },
    {
      "epoch": 0.5614126539388805,
      "grad_norm": 0.9611141873928662,
      "learning_rate": 2.1989874718696936e-06,
      "loss": 0.1926,
      "step": 1077
    },
    {
      "epoch": 0.5619339284550726,
      "grad_norm": 1.0452694949791048,
      "learning_rate": 2.194708644416646e-06,
      "loss": 0.1971,
      "step": 1078
    },
    {
      "epoch": 0.5624552029712647,
      "grad_norm": 0.96241502887155,
      "learning_rate": 2.1904307246114827e-06,
      "loss": 0.1917,
      "step": 1079
    },
    {
      "epoch": 0.5629764774874568,
      "grad_norm": 1.0410217584016626,
      "learning_rate": 2.1861537251726944e-06,
      "loss": 0.2003,
      "step": 1080
    },
    {
      "epoch": 0.5634977520036489,
      "grad_norm": 0.9990407192557624,
      "learning_rate": 2.1818776588160323e-06,
      "loss": 0.1872,
      "step": 1081
    },
    {
      "epoch": 0.564019026519841,
      "grad_norm": 1.0543177739241163,
      "learning_rate": 2.1776025382544765e-06,
      "loss": 0.2082,
      "step": 1082
    },
    {
      "epoch": 0.5645403010360331,
      "grad_norm": 1.0179741092006622,
      "learning_rate": 2.173328376198194e-06,
      "loss": 0.1894,
      "step": 1083
    },
    {
      "epoch": 0.5650615755522251,
      "grad_norm": 1.1269521168293783,
      "learning_rate": 2.1690551853545016e-06,
      "loss": 0.2155,
      "step": 1084
    },
    {
      "epoch": 0.5655828500684172,
      "grad_norm": 0.9797289046446224,
      "learning_rate": 2.1647829784278294e-06,
      "loss": 0.1908,
      "step": 1085
    },
    {
      "epoch": 0.5661041245846093,
      "grad_norm": 1.0049914572735903,
      "learning_rate": 2.1605117681196834e-06,
      "loss": 0.1878,
      "step": 1086
    },
    {
      "epoch": 0.5666253991008015,
      "grad_norm": 1.0936325945307788,
      "learning_rate": 2.1562415671286032e-06,
      "loss": 0.2054,
      "step": 1087
    },
    {
      "epoch": 0.5671466736169936,
      "grad_norm": 1.0392537049431012,
      "learning_rate": 2.151972388150131e-06,
      "loss": 0.2041,
      "step": 1088
    },
    {
      "epoch": 0.5676679481331857,
      "grad_norm": 1.0644432398734358,
      "learning_rate": 2.1477042438767685e-06,
      "loss": 0.2039,
      "step": 1089
    },
    {
      "epoch": 0.5681892226493778,
      "grad_norm": 0.9914071825710259,
      "learning_rate": 2.1434371469979424e-06,
      "loss": 0.1968,
      "step": 1090
    },
    {
      "epoch": 0.5687104971655699,
      "grad_norm": 1.0280714916865434,
      "learning_rate": 2.1391711101999656e-06,
      "loss": 0.2044,
      "step": 1091
    },
    {
      "epoch": 0.5692317716817619,
      "grad_norm": 1.015323078566454,
      "learning_rate": 2.1349061461659966e-06,
      "loss": 0.1994,
      "step": 1092
    },
    {
      "epoch": 0.569753046197954,
      "grad_norm": 1.0029348080456495,
      "learning_rate": 2.130642267576008e-06,
      "loss": 0.1848,
      "step": 1093
    },
    {
      "epoch": 0.5702743207141461,
      "grad_norm": 1.0256878184254634,
      "learning_rate": 2.1263794871067443e-06,
      "loss": 0.1999,
      "step": 1094
    },
    {
      "epoch": 0.5707955952303382,
      "grad_norm": 0.8964762595915654,
      "learning_rate": 2.1221178174316833e-06,
      "loss": 0.1693,
      "step": 1095
    },
    {
      "epoch": 0.5713168697465303,
      "grad_norm": 0.9850586416445359,
      "learning_rate": 2.117857271221003e-06,
      "loss": 0.1859,
      "step": 1096
    },
    {
      "epoch": 0.5718381442627224,
      "grad_norm": 1.052259025949159,
      "learning_rate": 2.11359786114154e-06,
      "loss": 0.1982,
      "step": 1097
    },
    {
      "epoch": 0.5723594187789145,
      "grad_norm": 0.9277809484196436,
      "learning_rate": 2.1093395998567527e-06,
      "loss": 0.1798,
      "step": 1098
    },
    {
      "epoch": 0.5728806932951065,
      "grad_norm": 0.983532854304064,
      "learning_rate": 2.1050825000266862e-06,
      "loss": 0.1765,
      "step": 1099
    },
    {
      "epoch": 0.5734019678112986,
      "grad_norm": 0.9631298651135378,
      "learning_rate": 2.1008265743079286e-06,
      "loss": 0.1958,
      "step": 1100
    },
    {
      "epoch": 0.5739232423274907,
      "grad_norm": 1.1033429405561814,
      "learning_rate": 2.0965718353535807e-06,
      "loss": 0.2055,
      "step": 1101
    },
    {
      "epoch": 0.5744445168436828,
      "grad_norm": 1.112349427593408,
      "learning_rate": 2.0923182958132146e-06,
      "loss": 0.2047,
      "step": 1102
    },
    {
      "epoch": 0.5749657913598749,
      "grad_norm": 0.9959652770009602,
      "learning_rate": 2.0880659683328346e-06,
      "loss": 0.187,
      "step": 1103
    },
    {
      "epoch": 0.575487065876067,
      "grad_norm": 1.0564422259924786,
      "learning_rate": 2.0838148655548433e-06,
      "loss": 0.2004,
      "step": 1104
    },
    {
      "epoch": 0.5760083403922591,
      "grad_norm": 1.0130265251843205,
      "learning_rate": 2.0795650001180028e-06,
      "loss": 0.191,
      "step": 1105
    },
    {
      "epoch": 0.5765296149084511,
      "grad_norm": 0.9988574601298932,
      "learning_rate": 2.0753163846573933e-06,
      "loss": 0.1997,
      "step": 1106
    },
    {
      "epoch": 0.5770508894246432,
      "grad_norm": 1.0298772663531115,
      "learning_rate": 2.0710690318043814e-06,
      "loss": 0.1913,
      "step": 1107
    },
    {
      "epoch": 0.5775721639408353,
      "grad_norm": 0.9523722397681231,
      "learning_rate": 2.0668229541865796e-06,
      "loss": 0.1857,
      "step": 1108
    },
    {
      "epoch": 0.5780934384570274,
      "grad_norm": 0.9797070321569168,
      "learning_rate": 2.0625781644278083e-06,
      "loss": 0.1943,
      "step": 1109
    },
    {
      "epoch": 0.5786147129732195,
      "grad_norm": 0.9838905532587987,
      "learning_rate": 2.058334675148061e-06,
      "loss": 0.1885,
      "step": 1110
    },
    {
      "epoch": 0.5791359874894116,
      "grad_norm": 0.9885106191955585,
      "learning_rate": 2.05409249896346e-06,
      "loss": 0.186,
      "step": 1111
    },
    {
      "epoch": 0.5796572620056037,
      "grad_norm": 0.9871469852945931,
      "learning_rate": 2.049851648486229e-06,
      "loss": 0.1935,
      "step": 1112
    },
    {
      "epoch": 0.5801785365217957,
      "grad_norm": 0.9966312159771141,
      "learning_rate": 2.04561213632465e-06,
      "loss": 0.1774,
      "step": 1113
    },
    {
      "epoch": 0.5806998110379878,
      "grad_norm": 1.0016101012425886,
      "learning_rate": 2.041373975083021e-06,
      "loss": 0.1978,
      "step": 1114
    },
    {
      "epoch": 0.58122108555418,
      "grad_norm": 1.0772865636871705,
      "learning_rate": 2.037137177361628e-06,
      "loss": 0.2117,
      "step": 1115
    },
    {
      "epoch": 0.581742360070372,
      "grad_norm": 1.031443556302358,
      "learning_rate": 2.0329017557567034e-06,
      "loss": 0.1997,
      "step": 1116
    },
    {
      "epoch": 0.5822636345865642,
      "grad_norm": 0.9954142013373319,
      "learning_rate": 2.028667722860386e-06,
      "loss": 0.195,
      "step": 1117
    },
    {
      "epoch": 0.5827849091027563,
      "grad_norm": 1.08950454301798,
      "learning_rate": 2.024435091260687e-06,
      "loss": 0.1965,
      "step": 1118
    },
    {
      "epoch": 0.5833061836189484,
      "grad_norm": 1.09820177135685,
      "learning_rate": 2.0202038735414532e-06,
      "loss": 0.2055,
      "step": 1119
    },
    {
      "epoch": 0.5838274581351404,
      "grad_norm": 1.0309431487311989,
      "learning_rate": 2.0159740822823233e-06,
      "loss": 0.1887,
      "step": 1120
    },
    {
      "epoch": 0.5843487326513325,
      "grad_norm": 0.9988968546634137,
      "learning_rate": 2.0117457300586996e-06,
      "loss": 0.1761,
      "step": 1121
    },
    {
      "epoch": 0.5848700071675246,
      "grad_norm": 1.0316307028818723,
      "learning_rate": 2.007518829441704e-06,
      "loss": 0.1849,
      "step": 1122
    },
    {
      "epoch": 0.5853912816837167,
      "grad_norm": 1.083940789434207,
      "learning_rate": 2.0032933929981425e-06,
      "loss": 0.1974,
      "step": 1123
    },
    {
      "epoch": 0.5859125561999088,
      "grad_norm": 1.0055559369202263,
      "learning_rate": 1.9990694332904705e-06,
      "loss": 0.1903,
      "step": 1124
    },
    {
      "epoch": 0.5864338307161009,
      "grad_norm": 1.1098092742660377,
      "learning_rate": 1.9948469628767475e-06,
      "loss": 0.197,
      "step": 1125
    },
    {
      "epoch": 0.586955105232293,
      "grad_norm": 0.9883818948839239,
      "learning_rate": 1.990625994310612e-06,
      "loss": 0.1956,
      "step": 1126
    },
    {
      "epoch": 0.5874763797484851,
      "grad_norm": 0.9848938775649817,
      "learning_rate": 1.986406540141234e-06,
      "loss": 0.1961,
      "step": 1127
    },
    {
      "epoch": 0.5879976542646771,
      "grad_norm": 1.0458643023845622,
      "learning_rate": 1.98218861291328e-06,
      "loss": 0.2008,
      "step": 1128
    },
    {
      "epoch": 0.5885189287808692,
      "grad_norm": 1.007080661415992,
      "learning_rate": 1.977972225166878e-06,
      "loss": 0.1744,
      "step": 1129
    },
    {
      "epoch": 0.5890402032970613,
      "grad_norm": 0.9341279253901913,
      "learning_rate": 1.973757389437581e-06,
      "loss": 0.1786,
      "step": 1130
    },
    {
      "epoch": 0.5895614778132534,
      "grad_norm": 1.0046180442150527,
      "learning_rate": 1.9695441182563237e-06,
      "loss": 0.201,
      "step": 1131
    },
    {
      "epoch": 0.5900827523294455,
      "grad_norm": 1.0791793317015532,
      "learning_rate": 1.965332424149394e-06,
      "loss": 0.1978,
      "step": 1132
    },
    {
      "epoch": 0.5906040268456376,
      "grad_norm": 0.9524534239736143,
      "learning_rate": 1.9611223196383866e-06,
      "loss": 0.1875,
      "step": 1133
    },
    {
      "epoch": 0.5911253013618297,
      "grad_norm": 0.9955184698085288,
      "learning_rate": 1.956913817240173e-06,
      "loss": 0.199,
      "step": 1134
    },
    {
      "epoch": 0.5916465758780217,
      "grad_norm": 0.9558706784940049,
      "learning_rate": 1.9527069294668617e-06,
      "loss": 0.1784,
      "step": 1135
    },
    {
      "epoch": 0.5921678503942138,
      "grad_norm": 0.9597385949296013,
      "learning_rate": 1.9485016688257578e-06,
      "loss": 0.1849,
      "step": 1136
    },
    {
      "epoch": 0.5926891249104059,
      "grad_norm": 1.0197141801144376,
      "learning_rate": 1.9442980478193332e-06,
      "loss": 0.1965,
      "step": 1137
    },
    {
      "epoch": 0.593210399426598,
      "grad_norm": 1.0512916670594386,
      "learning_rate": 1.9400960789451827e-06,
      "loss": 0.1944,
      "step": 1138
    },
    {
      "epoch": 0.5937316739427901,
      "grad_norm": 1.0086948626775847,
      "learning_rate": 1.935895774695988e-06,
      "loss": 0.1932,
      "step": 1139
    },
    {
      "epoch": 0.5942529484589822,
      "grad_norm": 1.008689795216177,
      "learning_rate": 1.9316971475594835e-06,
      "loss": 0.187,
      "step": 1140
    },
    {
      "epoch": 0.5947742229751743,
      "grad_norm": 0.9794699211499784,
      "learning_rate": 1.9275002100184186e-06,
      "loss": 0.1755,
      "step": 1141
    },
    {
      "epoch": 0.5952954974913663,
      "grad_norm": 0.969416416794242,
      "learning_rate": 1.9233049745505167e-06,
      "loss": 0.1822,
      "step": 1142
    },
    {
      "epoch": 0.5958167720075584,
      "grad_norm": 1.081025759730669,
      "learning_rate": 1.919111453628442e-06,
      "loss": 0.2018,
      "step": 1143
    },
    {
      "epoch": 0.5963380465237506,
      "grad_norm": 0.9710783893843369,
      "learning_rate": 1.914919659719762e-06,
      "loss": 0.1943,
      "step": 1144
    },
    {
      "epoch": 0.5968593210399427,
      "grad_norm": 1.0048716650665384,
      "learning_rate": 1.9107296052869086e-06,
      "loss": 0.1993,
      "step": 1145
    },
    {
      "epoch": 0.5973805955561348,
      "grad_norm": 1.0270218106767095,
      "learning_rate": 1.9065413027871437e-06,
      "loss": 0.2039,
      "step": 1146
    },
    {
      "epoch": 0.5979018700723269,
      "grad_norm": 0.9785024434468242,
      "learning_rate": 1.902354764672518e-06,
      "loss": 0.1835,
      "step": 1147
    },
    {
      "epoch": 0.598423144588519,
      "grad_norm": 0.974720285983276,
      "learning_rate": 1.8981700033898387e-06,
      "loss": 0.1916,
      "step": 1148
    },
    {
      "epoch": 0.598944419104711,
      "grad_norm": 0.9702604223153065,
      "learning_rate": 1.8939870313806302e-06,
      "loss": 0.1709,
      "step": 1149
    },
    {
      "epoch": 0.5994656936209031,
      "grad_norm": 0.989097265484285,
      "learning_rate": 1.8898058610810963e-06,
      "loss": 0.1976,
      "step": 1150
    },
    {
      "epoch": 0.5999869681370952,
      "grad_norm": 0.9488843498461761,
      "learning_rate": 1.8856265049220852e-06,
      "loss": 0.1763,
      "step": 1151
    },
    {
      "epoch": 0.6005082426532873,
      "grad_norm": 0.9596820229292397,
      "learning_rate": 1.8814489753290517e-06,
      "loss": 0.1905,
      "step": 1152
    },
    {
      "epoch": 0.6010295171694794,
      "grad_norm": 1.0384689717698112,
      "learning_rate": 1.8772732847220182e-06,
      "loss": 0.1805,
      "step": 1153
    },
    {
      "epoch": 0.6015507916856715,
      "grad_norm": 1.0290105611946982,
      "learning_rate": 1.873099445515542e-06,
      "loss": 0.1988,
      "step": 1154
    },
    {
      "epoch": 0.6020720662018636,
      "grad_norm": 1.025034490780247,
      "learning_rate": 1.868927470118675e-06,
      "loss": 0.1873,
      "step": 1155
    },
    {
      "epoch": 0.6025933407180556,
      "grad_norm": 1.025733485025965,
      "learning_rate": 1.8647573709349275e-06,
      "loss": 0.19,
      "step": 1156
    },
    {
      "epoch": 0.6031146152342477,
      "grad_norm": 0.9609127405778992,
      "learning_rate": 1.860589160362234e-06,
      "loss": 0.1855,
      "step": 1157
    },
    {
      "epoch": 0.6036358897504398,
      "grad_norm": 0.995953626450909,
      "learning_rate": 1.8564228507929099e-06,
      "loss": 0.1885,
      "step": 1158
    },
    {
      "epoch": 0.6041571642666319,
      "grad_norm": 0.9578995736358235,
      "learning_rate": 1.852258454613623e-06,
      "loss": 0.1839,
      "step": 1159
    },
    {
      "epoch": 0.604678438782824,
      "grad_norm": 1.039678404554969,
      "learning_rate": 1.8480959842053508e-06,
      "loss": 0.1871,
      "step": 1160
    },
    {
      "epoch": 0.6051997132990161,
      "grad_norm": 1.0276613326638602,
      "learning_rate": 1.843935451943344e-06,
      "loss": 0.1863,
      "step": 1161
    },
    {
      "epoch": 0.6057209878152082,
      "grad_norm": 1.0166903319135137,
      "learning_rate": 1.839776870197093e-06,
      "loss": 0.1895,
      "step": 1162
    },
    {
      "epoch": 0.6062422623314003,
      "grad_norm": 1.0187451093130595,
      "learning_rate": 1.8356202513302896e-06,
      "loss": 0.1851,
      "step": 1163
    },
    {
      "epoch": 0.6067635368475923,
      "grad_norm": 1.0097296058564171,
      "learning_rate": 1.831465607700787e-06,
      "loss": 0.1845,
      "step": 1164
    },
    {
      "epoch": 0.6072848113637844,
      "grad_norm": 1.0328200061175161,
      "learning_rate": 1.8273129516605698e-06,
      "loss": 0.1938,
      "step": 1165
    },
    {
      "epoch": 0.6078060858799765,
      "grad_norm": 1.052763044966393,
      "learning_rate": 1.8231622955557094e-06,
      "loss": 0.1871,
      "step": 1166
    },
    {
      "epoch": 0.6083273603961686,
      "grad_norm": 0.9365124362744572,
      "learning_rate": 1.819013651726334e-06,
      "loss": 0.183,
      "step": 1167
    },
    {
      "epoch": 0.6088486349123607,
      "grad_norm": 0.9824999481894984,
      "learning_rate": 1.8148670325065893e-06,
      "loss": 0.192,
      "step": 1168
    },
    {
      "epoch": 0.6093699094285528,
      "grad_norm": 1.0403686506787184,
      "learning_rate": 1.8107224502245997e-06,
      "loss": 0.1932,
      "step": 1169
    },
    {
      "epoch": 0.609891183944745,
      "grad_norm": 0.9876137785006327,
      "learning_rate": 1.806579917202435e-06,
      "loss": 0.1843,
      "step": 1170
    },
    {
      "epoch": 0.610412458460937,
      "grad_norm": 1.0298525415804813,
      "learning_rate": 1.802439445756073e-06,
      "loss": 0.1998,
      "step": 1171
    },
    {
      "epoch": 0.610933732977129,
      "grad_norm": 1.0240435196918738,
      "learning_rate": 1.7983010481953605e-06,
      "loss": 0.2012,
      "step": 1172
    },
    {
      "epoch": 0.6114550074933212,
      "grad_norm": 0.9783351242362962,
      "learning_rate": 1.7941647368239806e-06,
      "loss": 0.1767,
      "step": 1173
    },
    {
      "epoch": 0.6119762820095133,
      "grad_norm": 1.075799158466481,
      "learning_rate": 1.7900305239394134e-06,
      "loss": 0.1962,
      "step": 1174
    },
    {
      "epoch": 0.6124975565257054,
      "grad_norm": 1.0272437821538145,
      "learning_rate": 1.785898421832898e-06,
      "loss": 0.1991,
      "step": 1175
    },
    {
      "epoch": 0.6130188310418975,
      "grad_norm": 0.9289145434032438,
      "learning_rate": 1.7817684427894016e-06,
      "loss": 0.175,
      "step": 1176
    },
    {
      "epoch": 0.6135401055580896,
      "grad_norm": 1.0215094809266112,
      "learning_rate": 1.7776405990875761e-06,
      "loss": 0.2054,
      "step": 1177
    },
    {
      "epoch": 0.6140613800742816,
      "grad_norm": 1.006967541928481,
      "learning_rate": 1.7735149029997273e-06,
      "loss": 0.1742,
      "step": 1178
    },
    {
      "epoch": 0.6145826545904737,
      "grad_norm": 0.9941973792423793,
      "learning_rate": 1.7693913667917757e-06,
      "loss": 0.1904,
      "step": 1179
    },
    {
      "epoch": 0.6151039291066658,
      "grad_norm": 1.0478055612749158,
      "learning_rate": 1.7652700027232184e-06,
      "loss": 0.19,
      "step": 1180
    },
    {
      "epoch": 0.6156252036228579,
      "grad_norm": 0.9806980882496268,
      "learning_rate": 1.7611508230470963e-06,
      "loss": 0.1918,
      "step": 1181
    },
    {
      "epoch": 0.61614647813905,
      "grad_norm": 0.9472007679518236,
      "learning_rate": 1.7570338400099569e-06,
      "loss": 0.1811,
      "step": 1182
    },
    {
      "epoch": 0.6166677526552421,
      "grad_norm": 1.0323067946116886,
      "learning_rate": 1.7529190658518142e-06,
      "loss": 0.1946,
      "step": 1183
    },
    {
      "epoch": 0.6171890271714342,
      "grad_norm": 0.9486288320757825,
      "learning_rate": 1.7488065128061187e-06,
      "loss": 0.184,
      "step": 1184
    },
    {
      "epoch": 0.6177103016876262,
      "grad_norm": 0.9846912548005768,
      "learning_rate": 1.7446961930997126e-06,
      "loss": 0.1832,
      "step": 1185
    },
    {
      "epoch": 0.6182315762038183,
      "grad_norm": 1.0191390866118366,
      "learning_rate": 1.7405881189528024e-06,
      "loss": 0.2049,
      "step": 1186
    },
    {
      "epoch": 0.6187528507200104,
      "grad_norm": 1.0498041064818562,
      "learning_rate": 1.7364823025789184e-06,
      "loss": 0.195,
      "step": 1187
    },
    {
      "epoch": 0.6192741252362025,
      "grad_norm": 1.004045862782643,
      "learning_rate": 1.732378756184875e-06,
      "loss": 0.1893,
      "step": 1188
    },
    {
      "epoch": 0.6197953997523946,
      "grad_norm": 0.9919733197061383,
      "learning_rate": 1.7282774919707406e-06,
      "loss": 0.1851,
      "step": 1189
    },
    {
      "epoch": 0.6203166742685867,
      "grad_norm": 1.0767337921705271,
      "learning_rate": 1.7241785221297984e-06,
      "loss": 0.1955,
      "step": 1190
    },
    {
      "epoch": 0.6208379487847788,
      "grad_norm": 1.0235629614034516,
      "learning_rate": 1.7200818588485088e-06,
      "loss": 0.1835,
      "step": 1191
    },
    {
      "epoch": 0.6213592233009708,
      "grad_norm": 1.0357338095764963,
      "learning_rate": 1.715987514306476e-06,
      "loss": 0.192,
      "step": 1192
    },
    {
      "epoch": 0.6218804978171629,
      "grad_norm": 1.0642681157967024,
      "learning_rate": 1.7118955006764116e-06,
      "loss": 0.1832,
      "step": 1193
    },
    {
      "epoch": 0.622401772333355,
      "grad_norm": 0.9887558170762537,
      "learning_rate": 1.707805830124093e-06,
      "loss": 0.1851,
      "step": 1194
    },
    {
      "epoch": 0.6229230468495471,
      "grad_norm": 0.9869295102295927,
      "learning_rate": 1.703718514808336e-06,
      "loss": 0.1854,
      "step": 1195
    },
    {
      "epoch": 0.6234443213657392,
      "grad_norm": 1.0395086127510296,
      "learning_rate": 1.6996335668809515e-06,
      "loss": 0.1872,
      "step": 1196
    },
    {
      "epoch": 0.6239655958819313,
      "grad_norm": 1.001948089481037,
      "learning_rate": 1.6955509984867136e-06,
      "loss": 0.2012,
      "step": 1197
    },
    {
      "epoch": 0.6244868703981235,
      "grad_norm": 1.0298879796915752,
      "learning_rate": 1.691470821763322e-06,
      "loss": 0.1918,
      "step": 1198
    },
    {
      "epoch": 0.6250081449143156,
      "grad_norm": 1.0513920418408271,
      "learning_rate": 1.6873930488413628e-06,
      "loss": 0.2038,
      "step": 1199
    },
    {
      "epoch": 0.6255294194305075,
      "grad_norm": 0.9574789209372537,
      "learning_rate": 1.6833176918442796e-06,
      "loss": 0.172,
      "step": 1200
    },
    {
      "epoch": 0.6260506939466997,
      "grad_norm": 0.9811702658841635,
      "learning_rate": 1.6792447628883318e-06,
      "loss": 0.1931,
      "step": 1201
    },
    {
      "epoch": 0.6265719684628918,
      "grad_norm": 0.9521761233434269,
      "learning_rate": 1.6751742740825583e-06,
      "loss": 0.1859,
      "step": 1202
    },
    {
      "epoch": 0.6270932429790839,
      "grad_norm": 0.9755265815503429,
      "learning_rate": 1.6711062375287451e-06,
      "loss": 0.1886,
      "step": 1203
    },
    {
      "epoch": 0.627614517495276,
      "grad_norm": 0.9957314201963082,
      "learning_rate": 1.6670406653213883e-06,
      "loss": 0.1879,
      "step": 1204
    },
    {
      "epoch": 0.6281357920114681,
      "grad_norm": 0.985722885510226,
      "learning_rate": 1.6629775695476551e-06,
      "loss": 0.1888,
      "step": 1205
    },
    {
      "epoch": 0.6286570665276602,
      "grad_norm": 0.9638328517939708,
      "learning_rate": 1.6589169622873524e-06,
      "loss": 0.1768,
      "step": 1206
    },
    {
      "epoch": 0.6291783410438522,
      "grad_norm": 1.0202343212469378,
      "learning_rate": 1.654858855612886e-06,
      "loss": 0.1854,
      "step": 1207
    },
    {
      "epoch": 0.6296996155600443,
      "grad_norm": 1.0969397772782907,
      "learning_rate": 1.6508032615892294e-06,
      "loss": 0.1941,
      "step": 1208
    },
    {
      "epoch": 0.6302208900762364,
      "grad_norm": 1.1023320539818071,
      "learning_rate": 1.6467501922738862e-06,
      "loss": 0.2016,
      "step": 1209
    },
    {
      "epoch": 0.6307421645924285,
      "grad_norm": 1.0065554440090716,
      "learning_rate": 1.6426996597168517e-06,
      "loss": 0.2032,
      "step": 1210
    },
    {
      "epoch": 0.6312634391086206,
      "grad_norm": 0.9810314552742739,
      "learning_rate": 1.6386516759605813e-06,
      "loss": 0.2042,
      "step": 1211
    },
    {
      "epoch": 0.6317847136248127,
      "grad_norm": 1.0224511831624452,
      "learning_rate": 1.6346062530399525e-06,
      "loss": 0.1806,
      "step": 1212
    },
    {
      "epoch": 0.6323059881410048,
      "grad_norm": 1.0894152231671272,
      "learning_rate": 1.6305634029822267e-06,
      "loss": 0.1953,
      "step": 1213
    },
    {
      "epoch": 0.6328272626571968,
      "grad_norm": 1.0801698994360478,
      "learning_rate": 1.6265231378070197e-06,
      "loss": 0.1896,
      "step": 1214
    },
    {
      "epoch": 0.6333485371733889,
      "grad_norm": 1.0133080206136096,
      "learning_rate": 1.6224854695262616e-06,
      "loss": 0.1931,
      "step": 1215
    },
    {
      "epoch": 0.633869811689581,
      "grad_norm": 1.0103400995536298,
      "learning_rate": 1.6184504101441584e-06,
      "loss": 0.1796,
      "step": 1216
    },
    {
      "epoch": 0.6343910862057731,
      "grad_norm": 0.9640891691753638,
      "learning_rate": 1.614417971657164e-06,
      "loss": 0.1785,
      "step": 1217
    },
    {
      "epoch": 0.6349123607219652,
      "grad_norm": 1.0045197351637396,
      "learning_rate": 1.6103881660539369e-06,
      "loss": 0.1989,
      "step": 1218
    },
    {
      "epoch": 0.6354336352381573,
      "grad_norm": 1.0001304471152574,
      "learning_rate": 1.6063610053153106e-06,
      "loss": 0.1953,
      "step": 1219
    },
    {
      "epoch": 0.6359549097543494,
      "grad_norm": 1.046094461507422,
      "learning_rate": 1.6023365014142544e-06,
      "loss": 0.1887,
      "step": 1220
    },
    {
      "epoch": 0.6364761842705414,
      "grad_norm": 1.0184247756475113,
      "learning_rate": 1.5983146663158368e-06,
      "loss": 0.1912,
      "step": 1221
    },
    {
      "epoch": 0.6369974587867335,
      "grad_norm": 0.9979630692334167,
      "learning_rate": 1.5942955119771942e-06,
      "loss": 0.1931,
      "step": 1222
    },
    {
      "epoch": 0.6375187333029256,
      "grad_norm": 1.0610360668546952,
      "learning_rate": 1.5902790503474928e-06,
      "loss": 0.2011,
      "step": 1223
    },
    {
      "epoch": 0.6380400078191177,
      "grad_norm": 1.0520291292243582,
      "learning_rate": 1.586265293367891e-06,
      "loss": 0.1898,
      "step": 1224
    },
    {
      "epoch": 0.6385612823353098,
      "grad_norm": 1.1004751904858854,
      "learning_rate": 1.5822542529715084e-06,
      "loss": 0.193,
      "step": 1225
    },
    {
      "epoch": 0.639082556851502,
      "grad_norm": 0.9426792149238837,
      "learning_rate": 1.578245941083389e-06,
      "loss": 0.1713,
      "step": 1226
    },
    {
      "epoch": 0.639603831367694,
      "grad_norm": 1.0373871973414508,
      "learning_rate": 1.57424036962046e-06,
      "loss": 0.1879,
      "step": 1227
    },
    {
      "epoch": 0.640125105883886,
      "grad_norm": 0.9881065573845652,
      "learning_rate": 1.5702375504915062e-06,
      "loss": 0.1802,
      "step": 1228
    },
    {
      "epoch": 0.6406463804000782,
      "grad_norm": 1.0347702489584836,
      "learning_rate": 1.5662374955971268e-06,
      "loss": 0.1995,
      "step": 1229
    },
    {
      "epoch": 0.6411676549162703,
      "grad_norm": 1.0019968601970073,
      "learning_rate": 1.562240216829704e-06,
      "loss": 0.1839,
      "step": 1230
    },
    {
      "epoch": 0.6416889294324624,
      "grad_norm": 1.0410722029026276,
      "learning_rate": 1.5582457260733664e-06,
      "loss": 0.2073,
      "step": 1231
    },
    {
      "epoch": 0.6422102039486545,
      "grad_norm": 0.9942687899851895,
      "learning_rate": 1.5542540352039525e-06,
      "loss": 0.1994,
      "step": 1232
    },
    {
      "epoch": 0.6427314784648466,
      "grad_norm": 1.0482151628235123,
      "learning_rate": 1.550265156088978e-06,
      "loss": 0.1987,
      "step": 1233
    },
    {
      "epoch": 0.6432527529810387,
      "grad_norm": 0.9939264455923658,
      "learning_rate": 1.5462791005875994e-06,
      "loss": 0.1968,
      "step": 1234
    },
    {
      "epoch": 0.6437740274972308,
      "grad_norm": 1.0121688496454913,
      "learning_rate": 1.542295880550575e-06,
      "loss": 0.1856,
      "step": 1235
    },
    {
      "epoch": 0.6442953020134228,
      "grad_norm": 0.9801231914876962,
      "learning_rate": 1.5383155078202377e-06,
      "loss": 0.1789,
      "step": 1236
    },
    {
      "epoch": 0.6448165765296149,
      "grad_norm": 0.950154507161884,
      "learning_rate": 1.534337994230453e-06,
      "loss": 0.1833,
      "step": 1237
    },
    {
      "epoch": 0.645337851045807,
      "grad_norm": 1.065207980128952,
      "learning_rate": 1.5303633516065852e-06,
      "loss": 0.2029,
      "step": 1238
    },
    {
      "epoch": 0.6458591255619991,
      "grad_norm": 1.0117872135317376,
      "learning_rate": 1.5263915917654654e-06,
      "loss": 0.1895,
      "step": 1239
    },
    {
      "epoch": 0.6463804000781912,
      "grad_norm": 0.9920726473045625,
      "learning_rate": 1.5224227265153512e-06,
      "loss": 0.1843,
      "step": 1240
    },
    {
      "epoch": 0.6469016745943833,
      "grad_norm": 0.9517933928073986,
      "learning_rate": 1.5184567676558965e-06,
      "loss": 0.177,
      "step": 1241
    },
    {
      "epoch": 0.6474229491105754,
      "grad_norm": 1.0080575561411724,
      "learning_rate": 1.5144937269781142e-06,
      "loss": 0.1902,
      "step": 1242
    },
    {
      "epoch": 0.6479442236267674,
      "grad_norm": 1.0107231835819686,
      "learning_rate": 1.5105336162643403e-06,
      "loss": 0.1879,
      "step": 1243
    },
    {
      "epoch": 0.6484654981429595,
      "grad_norm": 1.0925930567014153,
      "learning_rate": 1.5065764472882e-06,
      "loss": 0.1991,
      "step": 1244
    },
    {
      "epoch": 0.6489867726591516,
      "grad_norm": 1.0814704323195263,
      "learning_rate": 1.5026222318145745e-06,
      "loss": 0.1776,
      "step": 1245
    },
    {
      "epoch": 0.6495080471753437,
      "grad_norm": 1.0148219928164357,
      "learning_rate": 1.4986709815995604e-06,
      "loss": 0.1811,
      "step": 1246
    },
    {
      "epoch": 0.6500293216915358,
      "grad_norm": 0.9554614027381633,
      "learning_rate": 1.494722708390442e-06,
      "loss": 0.1829,
      "step": 1247
    },
    {
      "epoch": 0.6505505962077279,
      "grad_norm": 1.0642386717671874,
      "learning_rate": 1.4907774239256517e-06,
      "loss": 0.2026,
      "step": 1248
    },
    {
      "epoch": 0.65107187072392,
      "grad_norm": 1.0345865561234546,
      "learning_rate": 1.486835139934734e-06,
      "loss": 0.1875,
      "step": 1249
    },
    {
      "epoch": 0.651593145240112,
      "grad_norm": 0.9737287029963221,
      "learning_rate": 1.4828958681383163e-06,
      "loss": 0.1806,
      "step": 1250
    },
    {
      "epoch": 0.6521144197563041,
      "grad_norm": 1.0602613532722558,
      "learning_rate": 1.4789596202480678e-06,
      "loss": 0.1954,
      "step": 1251
    },
    {
      "epoch": 0.6526356942724962,
      "grad_norm": 1.0691301072685224,
      "learning_rate": 1.475026407966669e-06,
      "loss": 0.1919,
      "step": 1252
    },
    {
      "epoch": 0.6531569687886883,
      "grad_norm": 0.9584321444229563,
      "learning_rate": 1.4710962429877763e-06,
      "loss": 0.1794,
      "step": 1253
    },
    {
      "epoch": 0.6536782433048804,
      "grad_norm": 0.9266776818848917,
      "learning_rate": 1.4671691369959826e-06,
      "loss": 0.1858,
      "step": 1254
    },
    {
      "epoch": 0.6541995178210726,
      "grad_norm": 0.9960819403397485,
      "learning_rate": 1.4632451016667899e-06,
      "loss": 0.1879,
      "step": 1255
    },
    {
      "epoch": 0.6547207923372647,
      "grad_norm": 0.9468971247993424,
      "learning_rate": 1.4593241486665705e-06,
      "loss": 0.17,
      "step": 1256
    },
    {
      "epoch": 0.6552420668534567,
      "grad_norm": 1.007504071715992,
      "learning_rate": 1.4554062896525295e-06,
      "loss": 0.1882,
      "step": 1257
    },
    {
      "epoch": 0.6557633413696488,
      "grad_norm": 1.0514437808565542,
      "learning_rate": 1.4514915362726773e-06,
      "loss": 0.196,
      "step": 1258
    },
    {
      "epoch": 0.6562846158858409,
      "grad_norm": 0.950780660352539,
      "learning_rate": 1.447579900165792e-06,
      "loss": 0.1737,
      "step": 1259
    },
    {
      "epoch": 0.656805890402033,
      "grad_norm": 0.996415800183811,
      "learning_rate": 1.4436713929613771e-06,
      "loss": 0.1839,
      "step": 1260
    },
    {
      "epoch": 0.6573271649182251,
      "grad_norm": 0.9380132689198648,
      "learning_rate": 1.4397660262796411e-06,
      "loss": 0.1874,
      "step": 1261
    },
    {
      "epoch": 0.6578484394344172,
      "grad_norm": 0.9607994317507901,
      "learning_rate": 1.4358638117314521e-06,
      "loss": 0.1735,
      "step": 1262
    },
    {
      "epoch": 0.6583697139506093,
      "grad_norm": 1.0481774247515332,
      "learning_rate": 1.4319647609183058e-06,
      "loss": 0.1877,
      "step": 1263
    },
    {
      "epoch": 0.6588909884668013,
      "grad_norm": 1.0605765278321324,
      "learning_rate": 1.428068885432296e-06,
      "loss": 0.1832,
      "step": 1264
    },
    {
      "epoch": 0.6594122629829934,
      "grad_norm": 0.935028645038226,
      "learning_rate": 1.4241761968560703e-06,
      "loss": 0.1707,
      "step": 1265
    },
    {
      "epoch": 0.6599335374991855,
      "grad_norm": 1.088512403659691,
      "learning_rate": 1.4202867067628068e-06,
      "loss": 0.1935,
      "step": 1266
    },
    {
      "epoch": 0.6604548120153776,
      "grad_norm": 1.1050024584482743,
      "learning_rate": 1.4164004267161718e-06,
      "loss": 0.1972,
      "step": 1267
    },
    {
      "epoch": 0.6609760865315697,
      "grad_norm": 1.082962753309766,
      "learning_rate": 1.4125173682702869e-06,
      "loss": 0.2,
      "step": 1268
    },
    {
      "epoch": 0.6614973610477618,
      "grad_norm": 1.0872379502189071,
      "learning_rate": 1.4086375429696987e-06,
      "loss": 0.184,
      "step": 1269
    },
    {
      "epoch": 0.6620186355639539,
      "grad_norm": 1.0164027128987314,
      "learning_rate": 1.4047609623493396e-06,
      "loss": 0.1728,
      "step": 1270
    },
    {
      "epoch": 0.662539910080146,
      "grad_norm": 1.027205814045248,
      "learning_rate": 1.400887637934495e-06,
      "loss": 0.1812,
      "step": 1271
    },
    {
      "epoch": 0.663061184596338,
      "grad_norm": 0.9821499317534202,
      "learning_rate": 1.3970175812407705e-06,
      "loss": 0.1807,
      "step": 1272
    },
    {
      "epoch": 0.6635824591125301,
      "grad_norm": 0.9665906056557528,
      "learning_rate": 1.393150803774055e-06,
      "loss": 0.1776,
      "step": 1273
    },
    {
      "epoch": 0.6641037336287222,
      "grad_norm": 1.035799794747972,
      "learning_rate": 1.3892873170304913e-06,
      "loss": 0.192,
      "step": 1274
    },
    {
      "epoch": 0.6646250081449143,
      "grad_norm": 1.0872770993767031,
      "learning_rate": 1.3854271324964353e-06,
      "loss": 0.2011,
      "step": 1275
    },
    {
      "epoch": 0.6651462826611064,
      "grad_norm": 0.9635813935447365,
      "learning_rate": 1.3815702616484262e-06,
      "loss": 0.1862,
      "step": 1276
    },
    {
      "epoch": 0.6656675571772985,
      "grad_norm": 1.002332685225423,
      "learning_rate": 1.3777167159531532e-06,
      "loss": 0.1731,
      "step": 1277
    },
    {
      "epoch": 0.6661888316934906,
      "grad_norm": 0.9888613603357296,
      "learning_rate": 1.3738665068674176e-06,
      "loss": 0.1947,
      "step": 1278
    },
    {
      "epoch": 0.6667101062096826,
      "grad_norm": 0.9740172319261662,
      "learning_rate": 1.3700196458381012e-06,
      "loss": 0.1804,
      "step": 1279
    },
    {
      "epoch": 0.6672313807258747,
      "grad_norm": 1.0305686003951755,
      "learning_rate": 1.3661761443021316e-06,
      "loss": 0.1943,
      "step": 1280
    },
    {
      "epoch": 0.6677526552420668,
      "grad_norm": 0.9511757362279438,
      "learning_rate": 1.36233601368645e-06,
      "loss": 0.1842,
      "step": 1281
    },
    {
      "epoch": 0.6682739297582589,
      "grad_norm": 1.013556966120342,
      "learning_rate": 1.3584992654079742e-06,
      "loss": 0.1867,
      "step": 1282
    },
    {
      "epoch": 0.668795204274451,
      "grad_norm": 0.9860617236699549,
      "learning_rate": 1.3546659108735666e-06,
      "loss": 0.178,
      "step": 1283
    },
    {
      "epoch": 0.6693164787906432,
      "grad_norm": 1.0204468647005256,
      "learning_rate": 1.3508359614799998e-06,
      "loss": 0.1867,
      "step": 1284
    },
    {
      "epoch": 0.6698377533068353,
      "grad_norm": 1.005403169634692,
      "learning_rate": 1.3470094286139213e-06,
      "loss": 0.2016,
      "step": 1285
    },
    {
      "epoch": 0.6703590278230273,
      "grad_norm": 0.9842739318151207,
      "learning_rate": 1.3431863236518242e-06,
      "loss": 0.1878,
      "step": 1286
    },
    {
      "epoch": 0.6708803023392194,
      "grad_norm": 0.9967671972856613,
      "learning_rate": 1.3393666579600078e-06,
      "loss": 0.1778,
      "step": 1287
    },
    {
      "epoch": 0.6714015768554115,
      "grad_norm": 1.0979134267108357,
      "learning_rate": 1.3355504428945464e-06,
      "loss": 0.1897,
      "step": 1288
    },
    {
      "epoch": 0.6719228513716036,
      "grad_norm": 1.0286269914897095,
      "learning_rate": 1.3317376898012573e-06,
      "loss": 0.1864,
      "step": 1289
    },
    {
      "epoch": 0.6724441258877957,
      "grad_norm": 1.1079206380471165,
      "learning_rate": 1.3279284100156633e-06,
      "loss": 0.2057,
      "step": 1290
    },
    {
      "epoch": 0.6729654004039878,
      "grad_norm": 0.9777662583461869,
      "learning_rate": 1.32412261486296e-06,
      "loss": 0.1848,
      "step": 1291
    },
    {
      "epoch": 0.6734866749201799,
      "grad_norm": 0.9233332390149452,
      "learning_rate": 1.3203203156579875e-06,
      "loss": 0.1712,
      "step": 1292
    },
    {
      "epoch": 0.6740079494363719,
      "grad_norm": 0.9886048566457575,
      "learning_rate": 1.316521523705185e-06,
      "loss": 0.1836,
      "step": 1293
    },
    {
      "epoch": 0.674529223952564,
      "grad_norm": 0.9565948337244332,
      "learning_rate": 1.3127262502985722e-06,
      "loss": 0.1837,
      "step": 1294
    },
    {
      "epoch": 0.6750504984687561,
      "grad_norm": 1.0257962365139226,
      "learning_rate": 1.3089345067217025e-06,
      "loss": 0.1881,
      "step": 1295
    },
    {
      "epoch": 0.6755717729849482,
      "grad_norm": 0.9891148901824458,
      "learning_rate": 1.3051463042476358e-06,
      "loss": 0.1729,
      "step": 1296
    },
    {
      "epoch": 0.6760930475011403,
      "grad_norm": 0.9216274264750317,
      "learning_rate": 1.301361654138909e-06,
      "loss": 0.1734,
      "step": 1297
    },
    {
      "epoch": 0.6766143220173324,
      "grad_norm": 1.1038391115681303,
      "learning_rate": 1.297580567647489e-06,
      "loss": 0.1936,
      "step": 1298
    },
    {
      "epoch": 0.6771355965335245,
      "grad_norm": 1.0276297679110642,
      "learning_rate": 1.2938030560147558e-06,
      "loss": 0.1824,
      "step": 1299
    },
    {
      "epoch": 0.6776568710497165,
      "grad_norm": 1.04852485176481,
      "learning_rate": 1.2900291304714568e-06,
      "loss": 0.1937,
      "step": 1300
    },
    {
      "epoch": 0.6781781455659086,
      "grad_norm": 1.0424297328699568,
      "learning_rate": 1.2862588022376782e-06,
      "loss": 0.1999,
      "step": 1301
    },
    {
      "epoch": 0.6786994200821007,
      "grad_norm": 0.9759091830528064,
      "learning_rate": 1.2824920825228132e-06,
      "loss": 0.1801,
      "step": 1302
    },
    {
      "epoch": 0.6792206945982928,
      "grad_norm": 1.0620848679730106,
      "learning_rate": 1.278728982525525e-06,
      "loss": 0.2058,
      "step": 1303
    },
    {
      "epoch": 0.6797419691144849,
      "grad_norm": 1.0090625327508094,
      "learning_rate": 1.2749695134337149e-06,
      "loss": 0.1842,
      "step": 1304
    },
    {
      "epoch": 0.680263243630677,
      "grad_norm": 0.9242576575771541,
      "learning_rate": 1.2712136864244917e-06,
      "loss": 0.1608,
      "step": 1305
    },
    {
      "epoch": 0.6807845181468691,
      "grad_norm": 0.9933430990539174,
      "learning_rate": 1.2674615126641313e-06,
      "loss": 0.1855,
      "step": 1306
    },
    {
      "epoch": 0.6813057926630612,
      "grad_norm": 0.9469833404696006,
      "learning_rate": 1.2637130033080541e-06,
      "loss": 0.1778,
      "step": 1307
    },
    {
      "epoch": 0.6818270671792532,
      "grad_norm": 1.014407572677528,
      "learning_rate": 1.2599681695007822e-06,
      "loss": 0.1908,
      "step": 1308
    },
    {
      "epoch": 0.6823483416954453,
      "grad_norm": 1.0462359805351626,
      "learning_rate": 1.2562270223759104e-06,
      "loss": 0.1896,
      "step": 1309
    },
    {
      "epoch": 0.6828696162116374,
      "grad_norm": 0.9820482321193924,
      "learning_rate": 1.2524895730560755e-06,
      "loss": 0.188,
      "step": 1310
    },
    {
      "epoch": 0.6833908907278295,
      "grad_norm": 0.9857078784821011,
      "learning_rate": 1.2487558326529177e-06,
      "loss": 0.1936,
      "step": 1311
    },
    {
      "epoch": 0.6839121652440217,
      "grad_norm": 0.9560315476350769,
      "learning_rate": 1.2450258122670508e-06,
      "loss": 0.1765,
      "step": 1312
    },
    {
      "epoch": 0.6844334397602138,
      "grad_norm": 0.9715559877348052,
      "learning_rate": 1.2412995229880295e-06,
      "loss": 0.1826,
      "step": 1313
    },
    {
      "epoch": 0.6849547142764059,
      "grad_norm": 1.036183134119075,
      "learning_rate": 1.2375769758943148e-06,
      "loss": 0.1991,
      "step": 1314
    },
    {
      "epoch": 0.6854759887925979,
      "grad_norm": 1.0175228043953535,
      "learning_rate": 1.2338581820532442e-06,
      "loss": 0.1942,
      "step": 1315
    },
    {
      "epoch": 0.68599726330879,
      "grad_norm": 0.9369199956509254,
      "learning_rate": 1.2301431525209942e-06,
      "loss": 0.1787,
      "step": 1316
    },
    {
      "epoch": 0.6865185378249821,
      "grad_norm": 0.9634845863264452,
      "learning_rate": 1.2264318983425498e-06,
      "loss": 0.1861,
      "step": 1317
    },
    {
      "epoch": 0.6870398123411742,
      "grad_norm": 0.9397809158825329,
      "learning_rate": 1.222724430551674e-06,
      "loss": 0.1753,
      "step": 1318
    },
    {
      "epoch": 0.6875610868573663,
      "grad_norm": 0.9675975576563132,
      "learning_rate": 1.2190207601708707e-06,
      "loss": 0.1778,
      "step": 1319
    },
    {
      "epoch": 0.6880823613735584,
      "grad_norm": 1.0333022331875106,
      "learning_rate": 1.215320898211354e-06,
      "loss": 0.1705,
      "step": 1320
    },
    {
      "epoch": 0.6886036358897505,
      "grad_norm": 0.9675011004283072,
      "learning_rate": 1.2116248556730151e-06,
      "loss": 0.1853,
      "step": 1321
    },
    {
      "epoch": 0.6891249104059425,
      "grad_norm": 1.2644809920198241,
      "learning_rate": 1.207932643544392e-06,
      "loss": 0.1869,
      "step": 1322
    },
    {
      "epoch": 0.6896461849221346,
      "grad_norm": 0.9943646922901057,
      "learning_rate": 1.2042442728026325e-06,
      "loss": 0.1817,
      "step": 1323
    },
    {
      "epoch": 0.6901674594383267,
      "grad_norm": 1.0673514818907925,
      "learning_rate": 1.200559754413464e-06,
      "loss": 0.2027,
      "step": 1324
    },
    {
      "epoch": 0.6906887339545188,
      "grad_norm": 1.0287772854680546,
      "learning_rate": 1.1968790993311613e-06,
      "loss": 0.1875,
      "step": 1325
    },
    {
      "epoch": 0.6912100084707109,
      "grad_norm": 1.0051001942800528,
      "learning_rate": 1.193202318498512e-06,
      "loss": 0.1863,
      "step": 1326
    },
    {
      "epoch": 0.691731282986903,
      "grad_norm": 0.9998296525007605,
      "learning_rate": 1.1895294228467886e-06,
      "loss": 0.1888,
      "step": 1327
    },
    {
      "epoch": 0.6922525575030951,
      "grad_norm": 1.0858913140394213,
      "learning_rate": 1.1858604232957096e-06,
      "loss": 0.1849,
      "step": 1328
    },
    {
      "epoch": 0.6927738320192871,
      "grad_norm": 1.0530841322620519,
      "learning_rate": 1.18219533075341e-06,
      "loss": 0.1911,
      "step": 1329
    },
    {
      "epoch": 0.6932951065354792,
      "grad_norm": 0.9766766830693403,
      "learning_rate": 1.1785341561164135e-06,
      "loss": 0.1748,
      "step": 1330
    },
    {
      "epoch": 0.6938163810516713,
      "grad_norm": 0.9730885083934125,
      "learning_rate": 1.1748769102695882e-06,
      "loss": 0.1812,
      "step": 1331
    },
    {
      "epoch": 0.6943376555678634,
      "grad_norm": 0.9801328173252677,
      "learning_rate": 1.1712236040861278e-06,
      "loss": 0.1847,
      "step": 1332
    },
    {
      "epoch": 0.6948589300840555,
      "grad_norm": 1.0299850610113763,
      "learning_rate": 1.1675742484275132e-06,
      "loss": 0.1868,
      "step": 1333
    },
    {
      "epoch": 0.6953802046002476,
      "grad_norm": 0.9980716124589071,
      "learning_rate": 1.1639288541434745e-06,
      "loss": 0.1892,
      "step": 1334
    },
    {
      "epoch": 0.6959014791164397,
      "grad_norm": 0.9331877883209707,
      "learning_rate": 1.160287432071971e-06,
      "loss": 0.1736,
      "step": 1335
    },
    {
      "epoch": 0.6964227536326317,
      "grad_norm": 0.9134410030075388,
      "learning_rate": 1.1566499930391484e-06,
      "loss": 0.1714,
      "step": 1336
    },
    {
      "epoch": 0.6969440281488238,
      "grad_norm": 1.0091365360794284,
      "learning_rate": 1.153016547859311e-06,
      "loss": 0.1859,
      "step": 1337
    },
    {
      "epoch": 0.6974653026650159,
      "grad_norm": 1.060667791878475,
      "learning_rate": 1.1493871073348926e-06,
      "loss": 0.1839,
      "step": 1338
    },
    {
      "epoch": 0.697986577181208,
      "grad_norm": 0.9791411501305487,
      "learning_rate": 1.1457616822564145e-06,
      "loss": 0.1862,
      "step": 1339
    },
    {
      "epoch": 0.6985078516974002,
      "grad_norm": 0.9968145078955214,
      "learning_rate": 1.1421402834024662e-06,
      "loss": 0.1773,
      "step": 1340
    },
    {
      "epoch": 0.6990291262135923,
      "grad_norm": 0.9968965667803882,
      "learning_rate": 1.1385229215396638e-06,
      "loss": 0.1854,
      "step": 1341
    },
    {
      "epoch": 0.6995504007297844,
      "grad_norm": 0.9841396870056438,
      "learning_rate": 1.1349096074226205e-06,
      "loss": 0.1813,
      "step": 1342
    },
    {
      "epoch": 0.7000716752459765,
      "grad_norm": 1.0163737558040131,
      "learning_rate": 1.1313003517939189e-06,
      "loss": 0.1893,
      "step": 1343
    },
    {
      "epoch": 0.7005929497621685,
      "grad_norm": 0.9657072281019743,
      "learning_rate": 1.127695165384072e-06,
      "loss": 0.1781,
      "step": 1344
    },
    {
      "epoch": 0.7011142242783606,
      "grad_norm": 1.0059536067719912,
      "learning_rate": 1.1240940589114953e-06,
      "loss": 0.1937,
      "step": 1345
    },
    {
      "epoch": 0.7016354987945527,
      "grad_norm": 1.0220451516359383,
      "learning_rate": 1.1204970430824782e-06,
      "loss": 0.188,
      "step": 1346
    },
    {
      "epoch": 0.7021567733107448,
      "grad_norm": 1.2826564607452682,
      "learning_rate": 1.1169041285911411e-06,
      "loss": 0.1978,
      "step": 1347
    },
    {
      "epoch": 0.7026780478269369,
      "grad_norm": 0.9652970950610329,
      "learning_rate": 1.113315326119418e-06,
      "loss": 0.1869,
      "step": 1348
    },
    {
      "epoch": 0.703199322343129,
      "grad_norm": 1.0123161734090071,
      "learning_rate": 1.1097306463370142e-06,
      "loss": 0.1939,
      "step": 1349
    },
    {
      "epoch": 0.7037205968593211,
      "grad_norm": 0.9817702021580154,
      "learning_rate": 1.1061500999013771e-06,
      "loss": 0.1764,
      "step": 1350
    },
    {
      "epoch": 0.7042418713755131,
      "grad_norm": 0.9729692552593922,
      "learning_rate": 1.1025736974576693e-06,
      "loss": 0.1818,
      "step": 1351
    },
    {
      "epoch": 0.7047631458917052,
      "grad_norm": 0.9431936624148214,
      "learning_rate": 1.0990014496387296e-06,
      "loss": 0.1705,
      "step": 1352
    },
    {
      "epoch": 0.7052844204078973,
      "grad_norm": 1.0042776662576223,
      "learning_rate": 1.0954333670650461e-06,
      "loss": 0.1925,
      "step": 1353
    },
    {
      "epoch": 0.7058056949240894,
      "grad_norm": 1.0977586396369685,
      "learning_rate": 1.0918694603447221e-06,
      "loss": 0.1837,
      "step": 1354
    },
    {
      "epoch": 0.7063269694402815,
      "grad_norm": 1.0168679210111984,
      "learning_rate": 1.0883097400734494e-06,
      "loss": 0.1962,
      "step": 1355
    },
    {
      "epoch": 0.7068482439564736,
      "grad_norm": 1.0292920840070696,
      "learning_rate": 1.0847542168344695e-06,
      "loss": 0.1801,
      "step": 1356
    },
    {
      "epoch": 0.7073695184726657,
      "grad_norm": 1.0207478308048166,
      "learning_rate": 1.0812029011985472e-06,
      "loss": 0.1976,
      "step": 1357
    },
    {
      "epoch": 0.7078907929888577,
      "grad_norm": 1.0183283198559208,
      "learning_rate": 1.0776558037239385e-06,
      "loss": 0.1817,
      "step": 1358
    },
    {
      "epoch": 0.7084120675050498,
      "grad_norm": 1.0247674891113325,
      "learning_rate": 1.0741129349563567e-06,
      "loss": 0.1893,
      "step": 1359
    },
    {
      "epoch": 0.7089333420212419,
      "grad_norm": 0.9495105171569985,
      "learning_rate": 1.0705743054289464e-06,
      "loss": 0.1691,
      "step": 1360
    },
    {
      "epoch": 0.709454616537434,
      "grad_norm": 0.997066256080277,
      "learning_rate": 1.0670399256622455e-06,
      "loss": 0.1811,
      "step": 1361
    },
    {
      "epoch": 0.7099758910536261,
      "grad_norm": 1.0041061371700566,
      "learning_rate": 1.0635098061641572e-06,
      "loss": 0.1951,
      "step": 1362
    },
    {
      "epoch": 0.7104971655698182,
      "grad_norm": 0.9971517329809013,
      "learning_rate": 1.0599839574299217e-06,
      "loss": 0.1877,
      "step": 1363
    },
    {
      "epoch": 0.7110184400860103,
      "grad_norm": 0.933411114504205,
      "learning_rate": 1.056462389942079e-06,
      "loss": 0.1644,
      "step": 1364
    },
    {
      "epoch": 0.7115397146022023,
      "grad_norm": 1.0074540227246198,
      "learning_rate": 1.052945114170441e-06,
      "loss": 0.1759,
      "step": 1365
    },
    {
      "epoch": 0.7120609891183944,
      "grad_norm": 0.9771710556827581,
      "learning_rate": 1.0494321405720627e-06,
      "loss": 0.1829,
      "step": 1366
    },
    {
      "epoch": 0.7125822636345865,
      "grad_norm": 1.0241608798558146,
      "learning_rate": 1.045923479591204e-06,
      "loss": 0.1684,
      "step": 1367
    },
    {
      "epoch": 0.7131035381507786,
      "grad_norm": 1.0344869658445373,
      "learning_rate": 1.042419141659308e-06,
      "loss": 0.1912,
      "step": 1368
    },
    {
      "epoch": 0.7136248126669708,
      "grad_norm": 1.0488873275095165,
      "learning_rate": 1.038919137194962e-06,
      "loss": 0.1941,
      "step": 1369
    },
    {
      "epoch": 0.7141460871831629,
      "grad_norm": 0.9613500411786476,
      "learning_rate": 1.0354234766038696e-06,
      "loss": 0.1791,
      "step": 1370
    },
    {
      "epoch": 0.714667361699355,
      "grad_norm": 0.9873531305786635,
      "learning_rate": 1.0319321702788234e-06,
      "loss": 0.1962,
      "step": 1371
    },
    {
      "epoch": 0.715188636215547,
      "grad_norm": 1.0233494605971423,
      "learning_rate": 1.028445228599665e-06,
      "loss": 0.1905,
      "step": 1372
    },
    {
      "epoch": 0.7157099107317391,
      "grad_norm": 1.0121219803666412,
      "learning_rate": 1.0249626619332642e-06,
      "loss": 0.1848,
      "step": 1373
    },
    {
      "epoch": 0.7162311852479312,
      "grad_norm": 1.0829372012716103,
      "learning_rate": 1.0214844806334817e-06,
      "loss": 0.1794,
      "step": 1374
    },
    {
      "epoch": 0.7167524597641233,
      "grad_norm": 1.022954391315338,
      "learning_rate": 1.0180106950411397e-06,
      "loss": 0.1918,
      "step": 1375
    },
    {
      "epoch": 0.7172737342803154,
      "grad_norm": 0.9591593243680496,
      "learning_rate": 1.0145413154839945e-06,
      "loss": 0.1701,
      "step": 1376
    },
    {
      "epoch": 0.7177950087965075,
      "grad_norm": 1.0345938222382716,
      "learning_rate": 1.0110763522767e-06,
      "loss": 0.1876,
      "step": 1377
    },
    {
      "epoch": 0.7183162833126996,
      "grad_norm": 0.9570590329423115,
      "learning_rate": 1.0076158157207801e-06,
      "loss": 0.1768,
      "step": 1378
    },
    {
      "epoch": 0.7188375578288917,
      "grad_norm": 1.0197234887133053,
      "learning_rate": 1.0041597161046025e-06,
      "loss": 0.1765,
      "step": 1379
    },
    {
      "epoch": 0.7193588323450837,
      "grad_norm": 1.0473051909557438,
      "learning_rate": 1.0007080637033359e-06,
      "loss": 0.192,
      "step": 1380
    },
    {
      "epoch": 0.7198801068612758,
      "grad_norm": 0.9992399137729582,
      "learning_rate": 9.972608687789346e-07,
      "loss": 0.1785,
      "step": 1381
    },
    {
      "epoch": 0.7204013813774679,
      "grad_norm": 0.968324575355683,
      "learning_rate": 9.938181415800966e-07,
      "loss": 0.1815,
      "step": 1382
    },
    {
      "epoch": 0.72092265589366,
      "grad_norm": 0.9070465078823997,
      "learning_rate": 9.903798923422369e-07,
      "loss": 0.1688,
      "step": 1383
    },
    {
      "epoch": 0.7214439304098521,
      "grad_norm": 1.0229224399463046,
      "learning_rate": 9.869461312874603e-07,
      "loss": 0.1782,
      "step": 1384
    },
    {
      "epoch": 0.7219652049260442,
      "grad_norm": 0.983129325811245,
      "learning_rate": 9.835168686245252e-07,
      "loss": 0.1847,
      "step": 1385
    },
    {
      "epoch": 0.7224864794422363,
      "grad_norm": 0.9887481799923892,
      "learning_rate": 9.800921145488171e-07,
      "loss": 0.1933,
      "step": 1386
    },
    {
      "epoch": 0.7230077539584283,
      "grad_norm": 1.0022753881190445,
      "learning_rate": 9.766718792423159e-07,
      "loss": 0.1772,
      "step": 1387
    },
    {
      "epoch": 0.7235290284746204,
      "grad_norm": 0.9848977927448296,
      "learning_rate": 9.732561728735698e-07,
      "loss": 0.1914,
      "step": 1388
    },
    {
      "epoch": 0.7240503029908125,
      "grad_norm": 1.0227529951142291,
      "learning_rate": 9.698450055976596e-07,
      "loss": 0.1804,
      "step": 1389
    },
    {
      "epoch": 0.7245715775070046,
      "grad_norm": 0.9645113903876174,
      "learning_rate": 9.664383875561726e-07,
      "loss": 0.1887,
      "step": 1390
    },
    {
      "epoch": 0.7250928520231967,
      "grad_norm": 0.9861088156398282,
      "learning_rate": 9.630363288771689e-07,
      "loss": 0.1806,
      "step": 1391
    },
    {
      "epoch": 0.7256141265393888,
      "grad_norm": 0.9742197798488345,
      "learning_rate": 9.596388396751567e-07,
      "loss": 0.1752,
      "step": 1392
    },
    {
      "epoch": 0.7261354010555809,
      "grad_norm": 1.0016811389435136,
      "learning_rate": 9.562459300510562e-07,
      "loss": 0.1963,
      "step": 1393
    },
    {
      "epoch": 0.7266566755717729,
      "grad_norm": 0.9925381490501447,
      "learning_rate": 9.528576100921736e-07,
      "loss": 0.1914,
      "step": 1394
    },
    {
      "epoch": 0.727177950087965,
      "grad_norm": 0.974247719423833,
      "learning_rate": 9.494738898721681e-07,
      "loss": 0.1888,
      "step": 1395
    },
    {
      "epoch": 0.7276992246041571,
      "grad_norm": 0.937358269563298,
      "learning_rate": 9.460947794510264e-07,
      "loss": 0.199,
      "step": 1396
    },
    {
      "epoch": 0.7282204991203493,
      "grad_norm": 0.9831205201102619,
      "learning_rate": 9.427202888750278e-07,
      "loss": 0.1906,
      "step": 1397
    },
    {
      "epoch": 0.7287417736365414,
      "grad_norm": 0.9354371339815937,
      "learning_rate": 9.393504281767163e-07,
      "loss": 0.1728,
      "step": 1398
    },
    {
      "epoch": 0.7292630481527335,
      "grad_norm": 0.9256419264427953,
      "learning_rate": 9.359852073748746e-07,
      "loss": 0.1757,
      "step": 1399
    },
    {
      "epoch": 0.7297843226689256,
      "grad_norm": 0.9729037248406467,
      "learning_rate": 9.326246364744845e-07,
      "loss": 0.1774,
      "step": 1400
    },
    {
      "epoch": 0.7303055971851176,
      "grad_norm": 0.9810766776155514,
      "learning_rate": 9.292687254667096e-07,
      "loss": 0.1655,
      "step": 1401
    },
    {
      "epoch": 0.7308268717013097,
      "grad_norm": 1.0321001659719977,
      "learning_rate": 9.259174843288557e-07,
      "loss": 0.1858,
      "step": 1402
    },
    {
      "epoch": 0.7313481462175018,
      "grad_norm": 0.9208300723484822,
      "learning_rate": 9.225709230243455e-07,
      "loss": 0.1643,
      "step": 1403
    },
    {
      "epoch": 0.7318694207336939,
      "grad_norm": 0.9530814587607274,
      "learning_rate": 9.192290515026903e-07,
      "loss": 0.1656,
      "step": 1404
    },
    {
      "epoch": 0.732390695249886,
      "grad_norm": 1.0161767095557832,
      "learning_rate": 9.158918796994543e-07,
      "loss": 0.1882,
      "step": 1405
    },
    {
      "epoch": 0.7329119697660781,
      "grad_norm": 0.9466451255581054,
      "learning_rate": 9.125594175362326e-07,
      "loss": 0.1671,
      "step": 1406
    },
    {
      "epoch": 0.7334332442822702,
      "grad_norm": 0.9745495623256282,
      "learning_rate": 9.092316749206198e-07,
      "loss": 0.1761,
      "step": 1407
    },
    {
      "epoch": 0.7339545187984622,
      "grad_norm": 0.9933921444812821,
      "learning_rate": 9.059086617461723e-07,
      "loss": 0.199,
      "step": 1408
    },
    {
      "epoch": 0.7344757933146543,
      "grad_norm": 0.9846422781950723,
      "learning_rate": 9.025903878923934e-07,
      "loss": 0.1887,
      "step": 1409
    },
    {
      "epoch": 0.7349970678308464,
      "grad_norm": 1.0878272099998363,
      "learning_rate": 8.992768632246907e-07,
      "loss": 0.211,
      "step": 1410
    },
    {
      "epoch": 0.7355183423470385,
      "grad_norm": 0.9673466071846137,
      "learning_rate": 8.959680975943541e-07,
      "loss": 0.1875,
      "step": 1411
    },
    {
      "epoch": 0.7360396168632306,
      "grad_norm": 1.0541385848385707,
      "learning_rate": 8.926641008385268e-07,
      "loss": 0.1819,
      "step": 1412
    },
    {
      "epoch": 0.7365608913794227,
      "grad_norm": 1.0655569570932815,
      "learning_rate": 8.893648827801685e-07,
      "loss": 0.1946,
      "step": 1413
    },
    {
      "epoch": 0.7370821658956148,
      "grad_norm": 0.9489331305334471,
      "learning_rate": 8.860704532280373e-07,
      "loss": 0.1786,
      "step": 1414
    },
    {
      "epoch": 0.7376034404118069,
      "grad_norm": 0.9626760271408069,
      "learning_rate": 8.827808219766513e-07,
      "loss": 0.1729,
      "step": 1415
    },
    {
      "epoch": 0.7381247149279989,
      "grad_norm": 0.9608225208461724,
      "learning_rate": 8.794959988062632e-07,
      "loss": 0.1749,
      "step": 1416
    },
    {
      "epoch": 0.738645989444191,
      "grad_norm": 0.9933305652995392,
      "learning_rate": 8.762159934828337e-07,
      "loss": 0.1777,
      "step": 1417
    },
    {
      "epoch": 0.7391672639603831,
      "grad_norm": 0.9939192332322994,
      "learning_rate": 8.729408157579968e-07,
      "loss": 0.1918,
      "step": 1418
    },
    {
      "epoch": 0.7396885384765752,
      "grad_norm": 1.018471566040822,
      "learning_rate": 8.696704753690344e-07,
      "loss": 0.1819,
      "step": 1419
    },
    {
      "epoch": 0.7402098129927673,
      "grad_norm": 1.0614581746374432,
      "learning_rate": 8.664049820388492e-07,
      "loss": 0.1968,
      "step": 1420
    },
    {
      "epoch": 0.7407310875089594,
      "grad_norm": 1.0222909779878884,
      "learning_rate": 8.631443454759283e-07,
      "loss": 0.1915,
      "step": 1421
    },
    {
      "epoch": 0.7412523620251515,
      "grad_norm": 1.0488852837468665,
      "learning_rate": 8.598885753743247e-07,
      "loss": 0.1858,
      "step": 1422
    },
    {
      "epoch": 0.7417736365413435,
      "grad_norm": 0.9724091270831652,
      "learning_rate": 8.566376814136201e-07,
      "loss": 0.1759,
      "step": 1423
    },
    {
      "epoch": 0.7422949110575356,
      "grad_norm": 1.0285374753801173,
      "learning_rate": 8.533916732588996e-07,
      "loss": 0.1965,
      "step": 1424
    },
    {
      "epoch": 0.7428161855737277,
      "grad_norm": 1.032068131509467,
      "learning_rate": 8.501505605607236e-07,
      "loss": 0.1918,
      "step": 1425
    },
    {
      "epoch": 0.7433374600899199,
      "grad_norm": 0.9604042665606899,
      "learning_rate": 8.469143529550968e-07,
      "loss": 0.1882,
      "step": 1426
    },
    {
      "epoch": 0.743858734606112,
      "grad_norm": 0.9805324536722502,
      "learning_rate": 8.436830600634416e-07,
      "loss": 0.1785,
      "step": 1427
    },
    {
      "epoch": 0.7443800091223041,
      "grad_norm": 0.9804859313469502,
      "learning_rate": 8.404566914925672e-07,
      "loss": 0.1726,
      "step": 1428
    },
    {
      "epoch": 0.7449012836384962,
      "grad_norm": 1.0411249495355863,
      "learning_rate": 8.372352568346453e-07,
      "loss": 0.1942,
      "step": 1429
    },
    {
      "epoch": 0.7454225581546882,
      "grad_norm": 1.065983987357002,
      "learning_rate": 8.340187656671767e-07,
      "loss": 0.1898,
      "step": 1430
    },
    {
      "epoch": 0.7459438326708803,
      "grad_norm": 0.997010556310911,
      "learning_rate": 8.308072275529652e-07,
      "loss": 0.1793,
      "step": 1431
    },
    {
      "epoch": 0.7464651071870724,
      "grad_norm": 0.9721936402384388,
      "learning_rate": 8.27600652040089e-07,
      "loss": 0.195,
      "step": 1432
    },
    {
      "epoch": 0.7469863817032645,
      "grad_norm": 0.9805292308865531,
      "learning_rate": 8.243990486618717e-07,
      "loss": 0.1837,
      "step": 1433
    },
    {
      "epoch": 0.7475076562194566,
      "grad_norm": 0.9925749768449906,
      "learning_rate": 8.212024269368565e-07,
      "loss": 0.1723,
      "step": 1434
    },
    {
      "epoch": 0.7480289307356487,
      "grad_norm": 0.9735123460764685,
      "learning_rate": 8.180107963687741e-07,
      "loss": 0.1743,
      "step": 1435
    },
    {
      "epoch": 0.7485502052518408,
      "grad_norm": 0.9538754854147554,
      "learning_rate": 8.148241664465157e-07,
      "loss": 0.1703,
      "step": 1436
    },
    {
      "epoch": 0.7490714797680328,
      "grad_norm": 0.9671414937233449,
      "learning_rate": 8.116425466441077e-07,
      "loss": 0.1702,
      "step": 1437
    },
    {
      "epoch": 0.7495927542842249,
      "grad_norm": 0.9358794599973769,
      "learning_rate": 8.084659464206787e-07,
      "loss": 0.1754,
      "step": 1438
    },
    {
      "epoch": 0.750114028800417,
      "grad_norm": 1.0328819574667487,
      "learning_rate": 8.052943752204339e-07,
      "loss": 0.1951,
      "step": 1439
    },
    {
      "epoch": 0.7506353033166091,
      "grad_norm": 0.9391446564466408,
      "learning_rate": 8.021278424726308e-07,
      "loss": 0.182,
      "step": 1440
    },
    {
      "epoch": 0.7511565778328012,
      "grad_norm": 0.9973077636582282,
      "learning_rate": 7.989663575915407e-07,
      "loss": 0.183,
      "step": 1441
    },
    {
      "epoch": 0.7516778523489933,
      "grad_norm": 0.994948001820505,
      "learning_rate": 7.958099299764332e-07,
      "loss": 0.1816,
      "step": 1442
    },
    {
      "epoch": 0.7521991268651854,
      "grad_norm": 0.9745039882273455,
      "learning_rate": 7.926585690115396e-07,
      "loss": 0.1849,
      "step": 1443
    },
    {
      "epoch": 0.7527204013813774,
      "grad_norm": 1.0585792440522037,
      "learning_rate": 7.895122840660272e-07,
      "loss": 0.1891,
      "step": 1444
    },
    {
      "epoch": 0.7532416758975695,
      "grad_norm": 1.0072933672269964,
      "learning_rate": 7.863710844939759e-07,
      "loss": 0.1813,
      "step": 1445
    },
    {
      "epoch": 0.7537629504137616,
      "grad_norm": 1.046799665395406,
      "learning_rate": 7.8323497963434e-07,
      "loss": 0.196,
      "step": 1446
    },
    {
      "epoch": 0.7542842249299537,
      "grad_norm": 0.9996936215535054,
      "learning_rate": 7.801039788109332e-07,
      "loss": 0.1818,
      "step": 1447
    },
    {
      "epoch": 0.7548054994461458,
      "grad_norm": 1.0324011507953594,
      "learning_rate": 7.769780913323916e-07,
      "loss": 0.1862,
      "step": 1448
    },
    {
      "epoch": 0.7553267739623379,
      "grad_norm": 0.9634685781762996,
      "learning_rate": 7.738573264921481e-07,
      "loss": 0.1837,
      "step": 1449
    },
    {
      "epoch": 0.75584804847853,
      "grad_norm": 0.9954983995056319,
      "learning_rate": 7.707416935684092e-07,
      "loss": 0.1841,
      "step": 1450
    },
    {
      "epoch": 0.7563693229947221,
      "grad_norm": 0.9465965431404901,
      "learning_rate": 7.676312018241211e-07,
      "loss": 0.182,
      "step": 1451
    },
    {
      "epoch": 0.7568905975109141,
      "grad_norm": 0.9844735426559792,
      "learning_rate": 7.645258605069444e-07,
      "loss": 0.1822,
      "step": 1452
    },
    {
      "epoch": 0.7574118720271062,
      "grad_norm": 0.9918860172082978,
      "learning_rate": 7.614256788492316e-07,
      "loss": 0.1748,
      "step": 1453
    },
    {
      "epoch": 0.7579331465432984,
      "grad_norm": 0.9681618564765927,
      "learning_rate": 7.583306660679888e-07,
      "loss": 0.1759,
      "step": 1454
    },
    {
      "epoch": 0.7584544210594905,
      "grad_norm": 1.0081967981460622,
      "learning_rate": 7.552408313648602e-07,
      "loss": 0.1874,
      "step": 1455
    },
    {
      "epoch": 0.7589756955756826,
      "grad_norm": 0.9932265000906999,
      "learning_rate": 7.521561839260927e-07,
      "loss": 0.1733,
      "step": 1456
    },
    {
      "epoch": 0.7594969700918747,
      "grad_norm": 1.037111313956083,
      "learning_rate": 7.49076732922511e-07,
      "loss": 0.1912,
      "step": 1457
    },
    {
      "epoch": 0.7600182446080668,
      "grad_norm": 0.9698497048428991,
      "learning_rate": 7.460024875094926e-07,
      "loss": 0.1791,
      "step": 1458
    },
    {
      "epoch": 0.7605395191242588,
      "grad_norm": 1.000100859624916,
      "learning_rate": 7.429334568269364e-07,
      "loss": 0.1951,
      "step": 1459
    },
    {
      "epoch": 0.7610607936404509,
      "grad_norm": 0.9698217611279528,
      "learning_rate": 7.398696499992386e-07,
      "loss": 0.1678,
      "step": 1460
    },
    {
      "epoch": 0.761582068156643,
      "grad_norm": 0.9813749452116728,
      "learning_rate": 7.368110761352631e-07,
      "loss": 0.1872,
      "step": 1461
    },
    {
      "epoch": 0.7621033426728351,
      "grad_norm": 0.9916765810612727,
      "learning_rate": 7.337577443283192e-07,
      "loss": 0.1775,
      "step": 1462
    },
    {
      "epoch": 0.7626246171890272,
      "grad_norm": 0.9499274567356653,
      "learning_rate": 7.307096636561281e-07,
      "loss": 0.1801,
      "step": 1463
    },
    {
      "epoch": 0.7631458917052193,
      "grad_norm": 0.981945431179099,
      "learning_rate": 7.276668431808004e-07,
      "loss": 0.1828,
      "step": 1464
    },
    {
      "epoch": 0.7636671662214114,
      "grad_norm": 0.9566783558292398,
      "learning_rate": 7.246292919488068e-07,
      "loss": 0.1752,
      "step": 1465
    },
    {
      "epoch": 0.7641884407376034,
      "grad_norm": 0.992884475254835,
      "learning_rate": 7.215970189909546e-07,
      "loss": 0.1814,
      "step": 1466
    },
    {
      "epoch": 0.7647097152537955,
      "grad_norm": 1.052003115861817,
      "learning_rate": 7.185700333223564e-07,
      "loss": 0.2006,
      "step": 1467
    },
    {
      "epoch": 0.7652309897699876,
      "grad_norm": 0.9624719382988022,
      "learning_rate": 7.155483439424058e-07,
      "loss": 0.1843,
      "step": 1468
    },
    {
      "epoch": 0.7657522642861797,
      "grad_norm": 1.0339494657208437,
      "learning_rate": 7.1253195983475e-07,
      "loss": 0.1867,
      "step": 1469
    },
    {
      "epoch": 0.7662735388023718,
      "grad_norm": 1.0270868784537983,
      "learning_rate": 7.09520889967265e-07,
      "loss": 0.1819,
      "step": 1470
    },
    {
      "epoch": 0.7667948133185639,
      "grad_norm": 0.9934453354538934,
      "learning_rate": 7.065151432920258e-07,
      "loss": 0.1841,
      "step": 1471
    },
    {
      "epoch": 0.767316087834756,
      "grad_norm": 1.1315817573018363,
      "learning_rate": 7.035147287452803e-07,
      "loss": 0.1995,
      "step": 1472
    },
    {
      "epoch": 0.767837362350948,
      "grad_norm": 1.018463046969739,
      "learning_rate": 7.005196552474275e-07,
      "loss": 0.1871,
      "step": 1473
    },
    {
      "epoch": 0.7683586368671401,
      "grad_norm": 1.0421420164623663,
      "learning_rate": 6.975299317029813e-07,
      "loss": 0.1922,
      "step": 1474
    },
    {
      "epoch": 0.7688799113833322,
      "grad_norm": 1.002838203747042,
      "learning_rate": 6.945455670005558e-07,
      "loss": 0.191,
      "step": 1475
    },
    {
      "epoch": 0.7694011858995243,
      "grad_norm": 0.9810312260984373,
      "learning_rate": 6.915665700128291e-07,
      "loss": 0.1786,
      "step": 1476
    },
    {
      "epoch": 0.7699224604157164,
      "grad_norm": 0.980439512239371,
      "learning_rate": 6.885929495965216e-07,
      "loss": 0.188,
      "step": 1477
    },
    {
      "epoch": 0.7704437349319085,
      "grad_norm": 1.022741021825978,
      "learning_rate": 6.856247145923708e-07,
      "loss": 0.1857,
      "step": 1478
    },
    {
      "epoch": 0.7709650094481006,
      "grad_norm": 0.9557018945651466,
      "learning_rate": 6.82661873825101e-07,
      "loss": 0.176,
      "step": 1479
    },
    {
      "epoch": 0.7714862839642926,
      "grad_norm": 1.0289243917403224,
      "learning_rate": 6.797044361033986e-07,
      "loss": 0.195,
      "step": 1480
    },
    {
      "epoch": 0.7720075584804847,
      "grad_norm": 1.0648986089824894,
      "learning_rate": 6.767524102198903e-07,
      "loss": 0.1812,
      "step": 1481
    },
    {
      "epoch": 0.7725288329966769,
      "grad_norm": 1.0620665115031733,
      "learning_rate": 6.738058049511067e-07,
      "loss": 0.2003,
      "step": 1482
    },
    {
      "epoch": 0.773050107512869,
      "grad_norm": 1.0327996964046404,
      "learning_rate": 6.708646290574694e-07,
      "loss": 0.1962,
      "step": 1483
    },
    {
      "epoch": 0.7735713820290611,
      "grad_norm": 1.0114282559427161,
      "learning_rate": 6.679288912832538e-07,
      "loss": 0.1913,
      "step": 1484
    },
    {
      "epoch": 0.7740926565452532,
      "grad_norm": 0.9816443250943625,
      "learning_rate": 6.649986003565678e-07,
      "loss": 0.1768,
      "step": 1485
    },
    {
      "epoch": 0.7746139310614453,
      "grad_norm": 0.9331501300978031,
      "learning_rate": 6.620737649893291e-07,
      "loss": 0.1678,
      "step": 1486
    },
    {
      "epoch": 0.7751352055776374,
      "grad_norm": 1.0566663982761528,
      "learning_rate": 6.591543938772302e-07,
      "loss": 0.1949,
      "step": 1487
    },
    {
      "epoch": 0.7756564800938294,
      "grad_norm": 0.967850497825642,
      "learning_rate": 6.562404956997229e-07,
      "loss": 0.1736,
      "step": 1488
    },
    {
      "epoch": 0.7761777546100215,
      "grad_norm": 1.0157020122014457,
      "learning_rate": 6.533320791199848e-07,
      "loss": 0.1779,
      "step": 1489
    },
    {
      "epoch": 0.7766990291262136,
      "grad_norm": 1.045591809870396,
      "learning_rate": 6.504291527848972e-07,
      "loss": 0.1834,
      "step": 1490
    },
    {
      "epoch": 0.7772203036424057,
      "grad_norm": 1.0434125948047916,
      "learning_rate": 6.475317253250196e-07,
      "loss": 0.1854,
      "step": 1491
    },
    {
      "epoch": 0.7777415781585978,
      "grad_norm": 0.9525166005684407,
      "learning_rate": 6.446398053545619e-07,
      "loss": 0.1679,
      "step": 1492
    },
    {
      "epoch": 0.7782628526747899,
      "grad_norm": 0.9809309122145046,
      "learning_rate": 6.417534014713594e-07,
      "loss": 0.182,
      "step": 1493
    },
    {
      "epoch": 0.778784127190982,
      "grad_norm": 1.0183445694309556,
      "learning_rate": 6.388725222568498e-07,
      "loss": 0.1922,
      "step": 1494
    },
    {
      "epoch": 0.779305401707174,
      "grad_norm": 1.030579937127978,
      "learning_rate": 6.359971762760442e-07,
      "loss": 0.182,
      "step": 1495
    },
    {
      "epoch": 0.7798266762233661,
      "grad_norm": 0.9551228877332707,
      "learning_rate": 6.331273720775028e-07,
      "loss": 0.1754,
      "step": 1496
    },
    {
      "epoch": 0.7803479507395582,
      "grad_norm": 1.0195208828277464,
      "learning_rate": 6.302631181933106e-07,
      "loss": 0.1863,
      "step": 1497
    },
    {
      "epoch": 0.7808692252557503,
      "grad_norm": 0.9720903989483768,
      "learning_rate": 6.274044231390503e-07,
      "loss": 0.1845,
      "step": 1498
    },
    {
      "epoch": 0.7813904997719424,
      "grad_norm": 0.9897645284030194,
      "learning_rate": 6.245512954137795e-07,
      "loss": 0.187,
      "step": 1499
    },
    {
      "epoch": 0.7819117742881345,
      "grad_norm": 0.9745175491361323,
      "learning_rate": 6.217037435000028e-07,
      "loss": 0.1811,
      "step": 1500
    },
    {
      "epoch": 0.7819117742881345,
      "eval_loss": 0.18243278563022614,
      "eval_runtime": 52.0221,
      "eval_samples_per_second": 23.855,
      "eval_steps_per_second": 2.999,
      "step": 1500
    },
    {
      "epoch": 0.7824330488043266,
      "grad_norm": 0.9624283120304148,
      "learning_rate": 6.18861775863647e-07,
      "loss": 0.1843,
      "step": 1501
    },
    {
      "epoch": 0.7829543233205186,
      "grad_norm": 0.9997178294787541,
      "learning_rate": 6.16025400954037e-07,
      "loss": 0.1844,
      "step": 1502
    },
    {
      "epoch": 0.7834755978367107,
      "grad_norm": 0.9690170298588008,
      "learning_rate": 6.131946272038719e-07,
      "loss": 0.1747,
      "step": 1503
    },
    {
      "epoch": 0.7839968723529028,
      "grad_norm": 0.9872530531107133,
      "learning_rate": 6.103694630291954e-07,
      "loss": 0.1889,
      "step": 1504
    },
    {
      "epoch": 0.7845181468690949,
      "grad_norm": 1.0002803187642446,
      "learning_rate": 6.07549916829375e-07,
      "loss": 0.1693,
      "step": 1505
    },
    {
      "epoch": 0.785039421385287,
      "grad_norm": 0.9866149594139076,
      "learning_rate": 6.047359969870773e-07,
      "loss": 0.1752,
      "step": 1506
    },
    {
      "epoch": 0.7855606959014791,
      "grad_norm": 1.0156582051001866,
      "learning_rate": 6.019277118682371e-07,
      "loss": 0.1932,
      "step": 1507
    },
    {
      "epoch": 0.7860819704176712,
      "grad_norm": 0.9848342276162435,
      "learning_rate": 5.991250698220416e-07,
      "loss": 0.185,
      "step": 1508
    },
    {
      "epoch": 0.7866032449338632,
      "grad_norm": 0.984391137727102,
      "learning_rate": 5.963280791808984e-07,
      "loss": 0.1962,
      "step": 1509
    },
    {
      "epoch": 0.7871245194500553,
      "grad_norm": 0.9483291063857775,
      "learning_rate": 5.935367482604124e-07,
      "loss": 0.1802,
      "step": 1510
    },
    {
      "epoch": 0.7876457939662475,
      "grad_norm": 1.0418162412838423,
      "learning_rate": 5.907510853593642e-07,
      "loss": 0.1857,
      "step": 1511
    },
    {
      "epoch": 0.7881670684824396,
      "grad_norm": 1.0343511985551024,
      "learning_rate": 5.879710987596818e-07,
      "loss": 0.1887,
      "step": 1512
    },
    {
      "epoch": 0.7886883429986317,
      "grad_norm": 0.9773227817505601,
      "learning_rate": 5.851967967264164e-07,
      "loss": 0.1718,
      "step": 1513
    },
    {
      "epoch": 0.7892096175148238,
      "grad_norm": 0.9692717128786674,
      "learning_rate": 5.824281875077212e-07,
      "loss": 0.1747,
      "step": 1514
    },
    {
      "epoch": 0.7897308920310159,
      "grad_norm": 0.9592152925255154,
      "learning_rate": 5.796652793348203e-07,
      "loss": 0.1698,
      "step": 1515
    },
    {
      "epoch": 0.7902521665472079,
      "grad_norm": 1.0199630611548223,
      "learning_rate": 5.769080804219929e-07,
      "loss": 0.1823,
      "step": 1516
    },
    {
      "epoch": 0.7907734410634,
      "grad_norm": 0.993001420164248,
      "learning_rate": 5.741565989665413e-07,
      "loss": 0.1914,
      "step": 1517
    },
    {
      "epoch": 0.7912947155795921,
      "grad_norm": 1.0044001865259375,
      "learning_rate": 5.714108431487692e-07,
      "loss": 0.1919,
      "step": 1518
    },
    {
      "epoch": 0.7918159900957842,
      "grad_norm": 1.009348827060207,
      "learning_rate": 5.68670821131961e-07,
      "loss": 0.1886,
      "step": 1519
    },
    {
      "epoch": 0.7923372646119763,
      "grad_norm": 1.0193106668548484,
      "learning_rate": 5.659365410623491e-07,
      "loss": 0.1932,
      "step": 1520
    },
    {
      "epoch": 0.7928585391281684,
      "grad_norm": 0.9849031420188077,
      "learning_rate": 5.63208011069099e-07,
      "loss": 0.1841,
      "step": 1521
    },
    {
      "epoch": 0.7933798136443605,
      "grad_norm": 0.9564397362264664,
      "learning_rate": 5.604852392642813e-07,
      "loss": 0.179,
      "step": 1522
    },
    {
      "epoch": 0.7939010881605526,
      "grad_norm": 0.9832377238334014,
      "learning_rate": 5.577682337428417e-07,
      "loss": 0.1882,
      "step": 1523
    },
    {
      "epoch": 0.7944223626767446,
      "grad_norm": 0.9964927270554703,
      "learning_rate": 5.550570025825888e-07,
      "loss": 0.185,
      "step": 1524
    },
    {
      "epoch": 0.7949436371929367,
      "grad_norm": 1.078579571117194,
      "learning_rate": 5.523515538441604e-07,
      "loss": 0.1885,
      "step": 1525
    },
    {
      "epoch": 0.7954649117091288,
      "grad_norm": 0.9423733562321697,
      "learning_rate": 5.496518955710023e-07,
      "loss": 0.1642,
      "step": 1526
    },
    {
      "epoch": 0.7959861862253209,
      "grad_norm": 0.9932650132284885,
      "learning_rate": 5.469580357893484e-07,
      "loss": 0.1834,
      "step": 1527
    },
    {
      "epoch": 0.796507460741513,
      "grad_norm": 0.9874453522435449,
      "learning_rate": 5.442699825081885e-07,
      "loss": 0.1781,
      "step": 1528
    },
    {
      "epoch": 0.7970287352577051,
      "grad_norm": 0.9898112554414138,
      "learning_rate": 5.415877437192535e-07,
      "loss": 0.1871,
      "step": 1529
    },
    {
      "epoch": 0.7975500097738972,
      "grad_norm": 1.0099309005097243,
      "learning_rate": 5.389113273969857e-07,
      "loss": 0.1916,
      "step": 1530
    },
    {
      "epoch": 0.7980712842900892,
      "grad_norm": 0.9194794514622183,
      "learning_rate": 5.36240741498516e-07,
      "loss": 0.1715,
      "step": 1531
    },
    {
      "epoch": 0.7985925588062813,
      "grad_norm": 0.9525100925930671,
      "learning_rate": 5.33575993963644e-07,
      "loss": 0.1677,
      "step": 1532
    },
    {
      "epoch": 0.7991138333224734,
      "grad_norm": 1.0095716989433727,
      "learning_rate": 5.309170927148088e-07,
      "loss": 0.173,
      "step": 1533
    },
    {
      "epoch": 0.7996351078386655,
      "grad_norm": 1.0361628682209612,
      "learning_rate": 5.282640456570692e-07,
      "loss": 0.19,
      "step": 1534
    },
    {
      "epoch": 0.8001563823548576,
      "grad_norm": 1.0032246075672515,
      "learning_rate": 5.256168606780784e-07,
      "loss": 0.1826,
      "step": 1535
    },
    {
      "epoch": 0.8006776568710497,
      "grad_norm": 0.9910274150150304,
      "learning_rate": 5.229755456480635e-07,
      "loss": 0.1661,
      "step": 1536
    },
    {
      "epoch": 0.8011989313872419,
      "grad_norm": 1.0188050211844513,
      "learning_rate": 5.203401084197973e-07,
      "loss": 0.1888,
      "step": 1537
    },
    {
      "epoch": 0.8017202059034338,
      "grad_norm": 0.9812489181895766,
      "learning_rate": 5.177105568285793e-07,
      "loss": 0.1745,
      "step": 1538
    },
    {
      "epoch": 0.802241480419626,
      "grad_norm": 1.0094080576230375,
      "learning_rate": 5.150868986922092e-07,
      "loss": 0.1841,
      "step": 1539
    },
    {
      "epoch": 0.8027627549358181,
      "grad_norm": 0.9390952886758952,
      "learning_rate": 5.124691418109673e-07,
      "loss": 0.1698,
      "step": 1540
    },
    {
      "epoch": 0.8032840294520102,
      "grad_norm": 1.0670290135128153,
      "learning_rate": 5.09857293967587e-07,
      "loss": 0.1911,
      "step": 1541
    },
    {
      "epoch": 0.8038053039682023,
      "grad_norm": 0.9622567515885702,
      "learning_rate": 5.072513629272352e-07,
      "loss": 0.1792,
      "step": 1542
    },
    {
      "epoch": 0.8043265784843944,
      "grad_norm": 1.0833879216384623,
      "learning_rate": 5.046513564374861e-07,
      "loss": 0.1918,
      "step": 1543
    },
    {
      "epoch": 0.8048478530005865,
      "grad_norm": 1.071394833175433,
      "learning_rate": 5.020572822283027e-07,
      "loss": 0.2082,
      "step": 1544
    },
    {
      "epoch": 0.8053691275167785,
      "grad_norm": 0.9465583168284061,
      "learning_rate": 4.994691480120086e-07,
      "loss": 0.1778,
      "step": 1545
    },
    {
      "epoch": 0.8058904020329706,
      "grad_norm": 0.9692484651932467,
      "learning_rate": 4.968869614832681e-07,
      "loss": 0.1616,
      "step": 1546
    },
    {
      "epoch": 0.8064116765491627,
      "grad_norm": 1.0420699332824392,
      "learning_rate": 4.943107303190639e-07,
      "loss": 0.1946,
      "step": 1547
    },
    {
      "epoch": 0.8069329510653548,
      "grad_norm": 0.9720960517462183,
      "learning_rate": 4.917404621786703e-07,
      "loss": 0.1729,
      "step": 1548
    },
    {
      "epoch": 0.8074542255815469,
      "grad_norm": 0.9837707057560047,
      "learning_rate": 4.891761647036364e-07,
      "loss": 0.1881,
      "step": 1549
    },
    {
      "epoch": 0.807975500097739,
      "grad_norm": 0.978722167032406,
      "learning_rate": 4.866178455177584e-07,
      "loss": 0.1739,
      "step": 1550
    },
    {
      "epoch": 0.8084967746139311,
      "grad_norm": 0.9409676553020354,
      "learning_rate": 4.840655122270582e-07,
      "loss": 0.1649,
      "step": 1551
    },
    {
      "epoch": 0.8090180491301231,
      "grad_norm": 0.9685240805958721,
      "learning_rate": 4.815191724197634e-07,
      "loss": 0.1782,
      "step": 1552
    },
    {
      "epoch": 0.8095393236463152,
      "grad_norm": 0.9901929023181778,
      "learning_rate": 4.78978833666281e-07,
      "loss": 0.171,
      "step": 1553
    },
    {
      "epoch": 0.8100605981625073,
      "grad_norm": 1.0737101425140607,
      "learning_rate": 4.7644450351917617e-07,
      "loss": 0.1779,
      "step": 1554
    },
    {
      "epoch": 0.8105818726786994,
      "grad_norm": 1.0148124979347264,
      "learning_rate": 4.7391618951315277e-07,
      "loss": 0.1831,
      "step": 1555
    },
    {
      "epoch": 0.8111031471948915,
      "grad_norm": 1.038320233035674,
      "learning_rate": 4.713938991650241e-07,
      "loss": 0.1961,
      "step": 1556
    },
    {
      "epoch": 0.8116244217110836,
      "grad_norm": 0.9697118349993694,
      "learning_rate": 4.688776399736991e-07,
      "loss": 0.1789,
      "step": 1557
    },
    {
      "epoch": 0.8121456962272757,
      "grad_norm": 0.9976052679774774,
      "learning_rate": 4.6636741942015314e-07,
      "loss": 0.1812,
      "step": 1558
    },
    {
      "epoch": 0.8126669707434678,
      "grad_norm": 0.9351359560800986,
      "learning_rate": 4.6386324496740855e-07,
      "loss": 0.174,
      "step": 1559
    },
    {
      "epoch": 0.8131882452596598,
      "grad_norm": 0.996276397945659,
      "learning_rate": 4.6136512406051495e-07,
      "loss": 0.1804,
      "step": 1560
    },
    {
      "epoch": 0.8137095197758519,
      "grad_norm": 1.0250044182372304,
      "learning_rate": 4.5887306412651933e-07,
      "loss": 0.1897,
      "step": 1561
    },
    {
      "epoch": 0.814230794292044,
      "grad_norm": 0.9542320294681652,
      "learning_rate": 4.563870725744543e-07,
      "loss": 0.186,
      "step": 1562
    },
    {
      "epoch": 0.8147520688082361,
      "grad_norm": 0.9837597440865188,
      "learning_rate": 4.539071567953077e-07,
      "loss": 0.1911,
      "step": 1563
    },
    {
      "epoch": 0.8152733433244282,
      "grad_norm": 0.981422035168465,
      "learning_rate": 4.514333241620045e-07,
      "loss": 0.1864,
      "step": 1564
    },
    {
      "epoch": 0.8157946178406204,
      "grad_norm": 0.9991362993949138,
      "learning_rate": 4.4896558202938505e-07,
      "loss": 0.1849,
      "step": 1565
    },
    {
      "epoch": 0.8163158923568125,
      "grad_norm": 0.9329728006170872,
      "learning_rate": 4.4650393773418123e-07,
      "loss": 0.1726,
      "step": 1566
    },
    {
      "epoch": 0.8168371668730044,
      "grad_norm": 1.0092147678909589,
      "learning_rate": 4.440483985949953e-07,
      "loss": 0.1772,
      "step": 1567
    },
    {
      "epoch": 0.8173584413891966,
      "grad_norm": 0.936495487087966,
      "learning_rate": 4.4159897191228096e-07,
      "loss": 0.1784,
      "step": 1568
    },
    {
      "epoch": 0.8178797159053887,
      "grad_norm": 1.032928505703867,
      "learning_rate": 4.3915566496831654e-07,
      "loss": 0.1889,
      "step": 1569
    },
    {
      "epoch": 0.8184009904215808,
      "grad_norm": 1.0075404329801207,
      "learning_rate": 4.367184850271872e-07,
      "loss": 0.1879,
      "step": 1570
    },
    {
      "epoch": 0.8189222649377729,
      "grad_norm": 0.9624770935173103,
      "learning_rate": 4.3428743933476207e-07,
      "loss": 0.1717,
      "step": 1571
    },
    {
      "epoch": 0.819443539453965,
      "grad_norm": 1.012318148309551,
      "learning_rate": 4.3186253511867227e-07,
      "loss": 0.1798,
      "step": 1572
    },
    {
      "epoch": 0.8199648139701571,
      "grad_norm": 1.0103015687227486,
      "learning_rate": 4.294437795882919e-07,
      "loss": 0.1855,
      "step": 1573
    },
    {
      "epoch": 0.8204860884863491,
      "grad_norm": 1.0291823813828462,
      "learning_rate": 4.2703117993471295e-07,
      "loss": 0.1857,
      "step": 1574
    },
    {
      "epoch": 0.8210073630025412,
      "grad_norm": 1.0164999339217138,
      "learning_rate": 4.2462474333072544e-07,
      "loss": 0.1813,
      "step": 1575
    },
    {
      "epoch": 0.8215286375187333,
      "grad_norm": 1.1952579229840181,
      "learning_rate": 4.222244769307965e-07,
      "loss": 0.1835,
      "step": 1576
    },
    {
      "epoch": 0.8220499120349254,
      "grad_norm": 1.036232662086274,
      "learning_rate": 4.198303878710508e-07,
      "loss": 0.1791,
      "step": 1577
    },
    {
      "epoch": 0.8225711865511175,
      "grad_norm": 0.9831689035420226,
      "learning_rate": 4.174424832692453e-07,
      "loss": 0.1799,
      "step": 1578
    },
    {
      "epoch": 0.8230924610673096,
      "grad_norm": 1.0187121088831226,
      "learning_rate": 4.1506077022475e-07,
      "loss": 0.1864,
      "step": 1579
    },
    {
      "epoch": 0.8236137355835017,
      "grad_norm": 0.9756861806562158,
      "learning_rate": 4.1268525581853015e-07,
      "loss": 0.1797,
      "step": 1580
    },
    {
      "epoch": 0.8241350100996937,
      "grad_norm": 0.9987252492688348,
      "learning_rate": 4.1031594711311686e-07,
      "loss": 0.1792,
      "step": 1581
    },
    {
      "epoch": 0.8246562846158858,
      "grad_norm": 1.0161464474871202,
      "learning_rate": 4.079528511525968e-07,
      "loss": 0.1886,
      "step": 1582
    },
    {
      "epoch": 0.8251775591320779,
      "grad_norm": 0.9675388822474331,
      "learning_rate": 4.0559597496258263e-07,
      "loss": 0.1861,
      "step": 1583
    },
    {
      "epoch": 0.82569883364827,
      "grad_norm": 0.9620631526320462,
      "learning_rate": 4.032453255501956e-07,
      "loss": 0.1828,
      "step": 1584
    },
    {
      "epoch": 0.8262201081644621,
      "grad_norm": 0.9659011244034107,
      "learning_rate": 4.0090090990404563e-07,
      "loss": 0.1763,
      "step": 1585
    },
    {
      "epoch": 0.8267413826806542,
      "grad_norm": 1.0028308830055879,
      "learning_rate": 3.9856273499420837e-07,
      "loss": 0.1872,
      "step": 1586
    },
    {
      "epoch": 0.8272626571968463,
      "grad_norm": 0.9981885758468215,
      "learning_rate": 3.962308077722049e-07,
      "loss": 0.1806,
      "step": 1587
    },
    {
      "epoch": 0.8277839317130383,
      "grad_norm": 0.9135455364846385,
      "learning_rate": 3.9390513517098413e-07,
      "loss": 0.1691,
      "step": 1588
    },
    {
      "epoch": 0.8283052062292304,
      "grad_norm": 0.9733994445538949,
      "learning_rate": 3.9158572410489537e-07,
      "loss": 0.153,
      "step": 1589
    },
    {
      "epoch": 0.8288264807454225,
      "grad_norm": 0.9694813007897026,
      "learning_rate": 3.8927258146967626e-07,
      "loss": 0.1703,
      "step": 1590
    },
    {
      "epoch": 0.8293477552616146,
      "grad_norm": 1.0508545379309457,
      "learning_rate": 3.8696571414242546e-07,
      "loss": 0.1929,
      "step": 1591
    },
    {
      "epoch": 0.8298690297778067,
      "grad_norm": 0.9910207028835948,
      "learning_rate": 3.846651289815856e-07,
      "loss": 0.1699,
      "step": 1592
    },
    {
      "epoch": 0.8303903042939988,
      "grad_norm": 0.970422630360107,
      "learning_rate": 3.823708328269232e-07,
      "loss": 0.183,
      "step": 1593
    },
    {
      "epoch": 0.830911578810191,
      "grad_norm": 1.0433065768511562,
      "learning_rate": 3.800828324995043e-07,
      "loss": 0.1776,
      "step": 1594
    },
    {
      "epoch": 0.8314328533263831,
      "grad_norm": 0.9840284379510608,
      "learning_rate": 3.7780113480167967e-07,
      "loss": 0.1698,
      "step": 1595
    },
    {
      "epoch": 0.831954127842575,
      "grad_norm": 1.0584347815355786,
      "learning_rate": 3.7552574651706253e-07,
      "loss": 0.187,
      "step": 1596
    },
    {
      "epoch": 0.8324754023587672,
      "grad_norm": 1.0497713676498555,
      "learning_rate": 3.7325667441050457e-07,
      "loss": 0.189,
      "step": 1597
    },
    {
      "epoch": 0.8329966768749593,
      "grad_norm": 0.9679001486261071,
      "learning_rate": 3.7099392522808253e-07,
      "loss": 0.1814,
      "step": 1598
    },
    {
      "epoch": 0.8335179513911514,
      "grad_norm": 0.969875902904243,
      "learning_rate": 3.6873750569707297e-07,
      "loss": 0.1837,
      "step": 1599
    },
    {
      "epoch": 0.8340392259073435,
      "grad_norm": 1.000730042756625,
      "learning_rate": 3.664874225259338e-07,
      "loss": 0.1926,
      "step": 1600
    },
    {
      "epoch": 0.8345605004235356,
      "grad_norm": 0.9799204124672941,
      "learning_rate": 3.6424368240428666e-07,
      "loss": 0.1805,
      "step": 1601
    },
    {
      "epoch": 0.8350817749397277,
      "grad_norm": 0.9720793205092597,
      "learning_rate": 3.62006292002893e-07,
      "loss": 0.1822,
      "step": 1602
    },
    {
      "epoch": 0.8356030494559197,
      "grad_norm": 0.9414624324828083,
      "learning_rate": 3.597752579736363e-07,
      "loss": 0.1773,
      "step": 1603
    },
    {
      "epoch": 0.8361243239721118,
      "grad_norm": 0.9704325170983187,
      "learning_rate": 3.57550586949503e-07,
      "loss": 0.1773,
      "step": 1604
    },
    {
      "epoch": 0.8366455984883039,
      "grad_norm": 1.0278784833228696,
      "learning_rate": 3.5533228554456095e-07,
      "loss": 0.1749,
      "step": 1605
    },
    {
      "epoch": 0.837166873004496,
      "grad_norm": 0.9698076444384477,
      "learning_rate": 3.5312036035394263e-07,
      "loss": 0.1668,
      "step": 1606
    },
    {
      "epoch": 0.8376881475206881,
      "grad_norm": 0.9865415755933876,
      "learning_rate": 3.5091481795382216e-07,
      "loss": 0.1908,
      "step": 1607
    },
    {
      "epoch": 0.8382094220368802,
      "grad_norm": 0.9841444286679596,
      "learning_rate": 3.487156649013973e-07,
      "loss": 0.175,
      "step": 1608
    },
    {
      "epoch": 0.8387306965530723,
      "grad_norm": 1.0613086922617954,
      "learning_rate": 3.4652290773487003e-07,
      "loss": 0.2001,
      "step": 1609
    },
    {
      "epoch": 0.8392519710692643,
      "grad_norm": 0.9987120923297828,
      "learning_rate": 3.44336552973428e-07,
      "loss": 0.1861,
      "step": 1610
    },
    {
      "epoch": 0.8397732455854564,
      "grad_norm": 0.9957523602327897,
      "learning_rate": 3.4215660711722276e-07,
      "loss": 0.1882,
      "step": 1611
    },
    {
      "epoch": 0.8402945201016485,
      "grad_norm": 1.0417753308685644,
      "learning_rate": 3.399830766473525e-07,
      "loss": 0.2053,
      "step": 1612
    },
    {
      "epoch": 0.8408157946178406,
      "grad_norm": 0.994487304090861,
      "learning_rate": 3.378159680258428e-07,
      "loss": 0.1869,
      "step": 1613
    },
    {
      "epoch": 0.8413370691340327,
      "grad_norm": 0.986003967741045,
      "learning_rate": 3.356552876956257e-07,
      "loss": 0.1716,
      "step": 1614
    },
    {
      "epoch": 0.8418583436502248,
      "grad_norm": 0.9510075016608932,
      "learning_rate": 3.3350104208052153e-07,
      "loss": 0.1794,
      "step": 1615
    },
    {
      "epoch": 0.8423796181664169,
      "grad_norm": 1.0680380304167463,
      "learning_rate": 3.3135323758522047e-07,
      "loss": 0.179,
      "step": 1616
    },
    {
      "epoch": 0.8429008926826089,
      "grad_norm": 0.9344897888121284,
      "learning_rate": 3.292118805952621e-07,
      "loss": 0.1666,
      "step": 1617
    },
    {
      "epoch": 0.843422167198801,
      "grad_norm": 0.9995816237383274,
      "learning_rate": 3.2707697747701913e-07,
      "loss": 0.1817,
      "step": 1618
    },
    {
      "epoch": 0.8439434417149931,
      "grad_norm": 0.9527368112369238,
      "learning_rate": 3.249485345776743e-07,
      "loss": 0.1818,
      "step": 1619
    },
    {
      "epoch": 0.8444647162311852,
      "grad_norm": 0.9563185247038841,
      "learning_rate": 3.228265582252041e-07,
      "loss": 0.1842,
      "step": 1620
    },
    {
      "epoch": 0.8449859907473773,
      "grad_norm": 0.9676686730390603,
      "learning_rate": 3.2071105472836207e-07,
      "loss": 0.1842,
      "step": 1621
    },
    {
      "epoch": 0.8455072652635695,
      "grad_norm": 0.945759746107448,
      "learning_rate": 3.186020303766532e-07,
      "loss": 0.1678,
      "step": 1622
    },
    {
      "epoch": 0.8460285397797616,
      "grad_norm": 0.9897130171138304,
      "learning_rate": 3.164994914403241e-07,
      "loss": 0.1692,
      "step": 1623
    },
    {
      "epoch": 0.8465498142959537,
      "grad_norm": 0.9858965780032267,
      "learning_rate": 3.1440344417033754e-07,
      "loss": 0.1798,
      "step": 1624
    },
    {
      "epoch": 0.8470710888121457,
      "grad_norm": 1.006916832075123,
      "learning_rate": 3.123138947983556e-07,
      "loss": 0.1724,
      "step": 1625
    },
    {
      "epoch": 0.8475923633283378,
      "grad_norm": 0.9659860247439446,
      "learning_rate": 3.102308495367243e-07,
      "loss": 0.173,
      "step": 1626
    },
    {
      "epoch": 0.8481136378445299,
      "grad_norm": 0.9804528945179373,
      "learning_rate": 3.081543145784499e-07,
      "loss": 0.1792,
      "step": 1627
    },
    {
      "epoch": 0.848634912360722,
      "grad_norm": 1.0081587481145602,
      "learning_rate": 3.060842960971844e-07,
      "loss": 0.1919,
      "step": 1628
    },
    {
      "epoch": 0.8491561868769141,
      "grad_norm": 1.0093772965493921,
      "learning_rate": 3.0402080024720724e-07,
      "loss": 0.1862,
      "step": 1629
    },
    {
      "epoch": 0.8496774613931062,
      "grad_norm": 0.9873162646732845,
      "learning_rate": 3.0196383316340226e-07,
      "loss": 0.1857,
      "step": 1630
    },
    {
      "epoch": 0.8501987359092983,
      "grad_norm": 0.9265922477417696,
      "learning_rate": 2.9991340096124644e-07,
      "loss": 0.1734,
      "step": 1631
    },
    {
      "epoch": 0.8507200104254903,
      "grad_norm": 1.0215134877772702,
      "learning_rate": 2.978695097367862e-07,
      "loss": 0.2025,
      "step": 1632
    },
    {
      "epoch": 0.8512412849416824,
      "grad_norm": 0.9613814433527633,
      "learning_rate": 2.958321655666219e-07,
      "loss": 0.1762,
      "step": 1633
    },
    {
      "epoch": 0.8517625594578745,
      "grad_norm": 0.9939851036657399,
      "learning_rate": 2.938013745078899e-07,
      "loss": 0.1817,
      "step": 1634
    },
    {
      "epoch": 0.8522838339740666,
      "grad_norm": 0.9361338440124887,
      "learning_rate": 2.917771425982413e-07,
      "loss": 0.1582,
      "step": 1635
    },
    {
      "epoch": 0.8528051084902587,
      "grad_norm": 0.9335310987532818,
      "learning_rate": 2.897594758558298e-07,
      "loss": 0.1726,
      "step": 1636
    },
    {
      "epoch": 0.8533263830064508,
      "grad_norm": 0.9993168548254366,
      "learning_rate": 2.8774838027928854e-07,
      "loss": 0.1754,
      "step": 1637
    },
    {
      "epoch": 0.8538476575226429,
      "grad_norm": 0.9662051256266582,
      "learning_rate": 2.857438618477135e-07,
      "loss": 0.1734,
      "step": 1638
    },
    {
      "epoch": 0.8543689320388349,
      "grad_norm": 0.9798016098663039,
      "learning_rate": 2.837459265206491e-07,
      "loss": 0.1781,
      "step": 1639
    },
    {
      "epoch": 0.854890206555027,
      "grad_norm": 1.083203468022416,
      "learning_rate": 2.817545802380653e-07,
      "loss": 0.2084,
      "step": 1640
    },
    {
      "epoch": 0.8554114810712191,
      "grad_norm": 0.9702767949680986,
      "learning_rate": 2.797698289203432e-07,
      "loss": 0.1814,
      "step": 1641
    },
    {
      "epoch": 0.8559327555874112,
      "grad_norm": 0.9864042489800868,
      "learning_rate": 2.7779167846825797e-07,
      "loss": 0.1742,
      "step": 1642
    },
    {
      "epoch": 0.8564540301036033,
      "grad_norm": 0.9525936742900207,
      "learning_rate": 2.7582013476295826e-07,
      "loss": 0.1826,
      "step": 1643
    },
    {
      "epoch": 0.8569753046197954,
      "grad_norm": 0.975678936997716,
      "learning_rate": 2.738552036659517e-07,
      "loss": 0.1838,
      "step": 1644
    },
    {
      "epoch": 0.8574965791359875,
      "grad_norm": 0.9472599686570292,
      "learning_rate": 2.718968910190853e-07,
      "loss": 0.1674,
      "step": 1645
    },
    {
      "epoch": 0.8580178536521795,
      "grad_norm": 0.959904368853958,
      "learning_rate": 2.6994520264452957e-07,
      "loss": 0.1753,
      "step": 1646
    },
    {
      "epoch": 0.8585391281683716,
      "grad_norm": 0.9094487422267438,
      "learning_rate": 2.6800014434476137e-07,
      "loss": 0.1621,
      "step": 1647
    },
    {
      "epoch": 0.8590604026845637,
      "grad_norm": 0.9597770843306208,
      "learning_rate": 2.6606172190254467e-07,
      "loss": 0.1722,
      "step": 1648
    },
    {
      "epoch": 0.8595816772007558,
      "grad_norm": 1.0318500304093965,
      "learning_rate": 2.6412994108091586e-07,
      "loss": 0.1856,
      "step": 1649
    },
    {
      "epoch": 0.860102951716948,
      "grad_norm": 1.0407206119375378,
      "learning_rate": 2.6220480762316396e-07,
      "loss": 0.1905,
      "step": 1650
    },
    {
      "epoch": 0.86062422623314,
      "grad_norm": 0.9806484699231431,
      "learning_rate": 2.6028632725281723e-07,
      "loss": 0.192,
      "step": 1651
    },
    {
      "epoch": 0.8611455007493322,
      "grad_norm": 1.0021777700354728,
      "learning_rate": 2.58374505673622e-07,
      "loss": 0.2076,
      "step": 1652
    },
    {
      "epoch": 0.8616667752655242,
      "grad_norm": 1.0056279393326901,
      "learning_rate": 2.56469348569528e-07,
      "loss": 0.1964,
      "step": 1653
    },
    {
      "epoch": 0.8621880497817163,
      "grad_norm": 0.9696793333836914,
      "learning_rate": 2.5457086160467187e-07,
      "loss": 0.1753,
      "step": 1654
    },
    {
      "epoch": 0.8627093242979084,
      "grad_norm": 0.9772770309895201,
      "learning_rate": 2.526790504233592e-07,
      "loss": 0.176,
      "step": 1655
    },
    {
      "epoch": 0.8632305988141005,
      "grad_norm": 0.9821643376455257,
      "learning_rate": 2.507939206500479e-07,
      "loss": 0.1791,
      "step": 1656
    },
    {
      "epoch": 0.8637518733302926,
      "grad_norm": 0.9905289226333893,
      "learning_rate": 2.4891547788933185e-07,
      "loss": 0.1931,
      "step": 1657
    },
    {
      "epoch": 0.8642731478464847,
      "grad_norm": 0.96401523470504,
      "learning_rate": 2.4704372772592326e-07,
      "loss": 0.1797,
      "step": 1658
    },
    {
      "epoch": 0.8647944223626768,
      "grad_norm": 0.9499449664889251,
      "learning_rate": 2.4517867572463884e-07,
      "loss": 0.1743,
      "step": 1659
    },
    {
      "epoch": 0.8653156968788689,
      "grad_norm": 0.9850999754121548,
      "learning_rate": 2.433203274303794e-07,
      "loss": 0.1689,
      "step": 1660
    },
    {
      "epoch": 0.8658369713950609,
      "grad_norm": 1.01772409958344,
      "learning_rate": 2.414686883681158e-07,
      "loss": 0.1753,
      "step": 1661
    },
    {
      "epoch": 0.866358245911253,
      "grad_norm": 0.9691772600585687,
      "learning_rate": 2.3962376404287365e-07,
      "loss": 0.1825,
      "step": 1662
    },
    {
      "epoch": 0.8668795204274451,
      "grad_norm": 1.0124591744101696,
      "learning_rate": 2.3778555993971164e-07,
      "loss": 0.1866,
      "step": 1663
    },
    {
      "epoch": 0.8674007949436372,
      "grad_norm": 0.9965123836345188,
      "learning_rate": 2.359540815237124e-07,
      "loss": 0.1811,
      "step": 1664
    },
    {
      "epoch": 0.8679220694598293,
      "grad_norm": 1.0446021288366747,
      "learning_rate": 2.3412933423996114e-07,
      "loss": 0.1909,
      "step": 1665
    },
    {
      "epoch": 0.8684433439760214,
      "grad_norm": 0.9880596622848659,
      "learning_rate": 2.3231132351353075e-07,
      "loss": 0.1809,
      "step": 1666
    },
    {
      "epoch": 0.8689646184922135,
      "grad_norm": 1.014446771053952,
      "learning_rate": 2.305000547494679e-07,
      "loss": 0.1878,
      "step": 1667
    },
    {
      "epoch": 0.8694858930084055,
      "grad_norm": 1.0142208291598611,
      "learning_rate": 2.2869553333277145e-07,
      "loss": 0.1852,
      "step": 1668
    },
    {
      "epoch": 0.8700071675245976,
      "grad_norm": 0.9751837375695704,
      "learning_rate": 2.2689776462838348e-07,
      "loss": 0.1733,
      "step": 1669
    },
    {
      "epoch": 0.8705284420407897,
      "grad_norm": 0.9939676929319553,
      "learning_rate": 2.251067539811691e-07,
      "loss": 0.1897,
      "step": 1670
    },
    {
      "epoch": 0.8710497165569818,
      "grad_norm": 0.9773698978757089,
      "learning_rate": 2.233225067158995e-07,
      "loss": 0.1735,
      "step": 1671
    },
    {
      "epoch": 0.8715709910731739,
      "grad_norm": 0.9841460047764211,
      "learning_rate": 2.2154502813724026e-07,
      "loss": 0.1938,
      "step": 1672
    },
    {
      "epoch": 0.872092265589366,
      "grad_norm": 1.02614898643778,
      "learning_rate": 2.1977432352973226e-07,
      "loss": 0.1958,
      "step": 1673
    },
    {
      "epoch": 0.8726135401055581,
      "grad_norm": 0.9629274090579781,
      "learning_rate": 2.1801039815777637e-07,
      "loss": 0.184,
      "step": 1674
    },
    {
      "epoch": 0.8731348146217501,
      "grad_norm": 0.9458912450222245,
      "learning_rate": 2.1625325726562009e-07,
      "loss": 0.1737,
      "step": 1675
    },
    {
      "epoch": 0.8736560891379422,
      "grad_norm": 0.9854272942394587,
      "learning_rate": 2.1450290607733903e-07,
      "loss": 0.1892,
      "step": 1676
    },
    {
      "epoch": 0.8741773636541343,
      "grad_norm": 0.9170439224566707,
      "learning_rate": 2.127593497968225e-07,
      "loss": 0.1625,
      "step": 1677
    },
    {
      "epoch": 0.8746986381703264,
      "grad_norm": 0.9912800697939179,
      "learning_rate": 2.1102259360775895e-07,
      "loss": 0.1869,
      "step": 1678
    },
    {
      "epoch": 0.8752199126865186,
      "grad_norm": 1.0010841890382143,
      "learning_rate": 2.0929264267361888e-07,
      "loss": 0.1876,
      "step": 1679
    },
    {
      "epoch": 0.8757411872027107,
      "grad_norm": 0.982385353388596,
      "learning_rate": 2.07569502137642e-07,
      "loss": 0.1662,
      "step": 1680
    },
    {
      "epoch": 0.8762624617189028,
      "grad_norm": 1.0012670470341332,
      "learning_rate": 2.0585317712281866e-07,
      "loss": 0.1799,
      "step": 1681
    },
    {
      "epoch": 0.8767837362350948,
      "grad_norm": 0.9876128950810698,
      "learning_rate": 2.0414367273187713e-07,
      "loss": 0.1897,
      "step": 1682
    },
    {
      "epoch": 0.8773050107512869,
      "grad_norm": 1.0244502518557284,
      "learning_rate": 2.0244099404726686e-07,
      "loss": 0.1916,
      "step": 1683
    },
    {
      "epoch": 0.877826285267479,
      "grad_norm": 0.9310013268689511,
      "learning_rate": 2.007451461311455e-07,
      "loss": 0.1784,
      "step": 1684
    },
    {
      "epoch": 0.8783475597836711,
      "grad_norm": 1.0023999616194614,
      "learning_rate": 1.990561340253608e-07,
      "loss": 0.1858,
      "step": 1685
    },
    {
      "epoch": 0.8788688342998632,
      "grad_norm": 1.0212257693639213,
      "learning_rate": 1.9737396275143816e-07,
      "loss": 0.1876,
      "step": 1686
    },
    {
      "epoch": 0.8793901088160553,
      "grad_norm": 0.9863311854678503,
      "learning_rate": 1.9569863731056544e-07,
      "loss": 0.1865,
      "step": 1687
    },
    {
      "epoch": 0.8799113833322474,
      "grad_norm": 1.0293010657590143,
      "learning_rate": 1.940301626835761e-07,
      "loss": 0.1725,
      "step": 1688
    },
    {
      "epoch": 0.8804326578484394,
      "grad_norm": 0.9900962579192577,
      "learning_rate": 1.9236854383093668e-07,
      "loss": 0.1896,
      "step": 1689
    },
    {
      "epoch": 0.8809539323646315,
      "grad_norm": 0.990946654309898,
      "learning_rate": 1.9071378569273047e-07,
      "loss": 0.1713,
      "step": 1690
    },
    {
      "epoch": 0.8814752068808236,
      "grad_norm": 0.9571615778995071,
      "learning_rate": 1.8906589318864384e-07,
      "loss": 0.1791,
      "step": 1691
    },
    {
      "epoch": 0.8819964813970157,
      "grad_norm": 1.0306381407381788,
      "learning_rate": 1.8742487121795188e-07,
      "loss": 0.1839,
      "step": 1692
    },
    {
      "epoch": 0.8825177559132078,
      "grad_norm": 0.9923315907519844,
      "learning_rate": 1.8579072465950197e-07,
      "loss": 0.1829,
      "step": 1693
    },
    {
      "epoch": 0.8830390304293999,
      "grad_norm": 1.0142772500987272,
      "learning_rate": 1.8416345837170114e-07,
      "loss": 0.1887,
      "step": 1694
    },
    {
      "epoch": 0.883560304945592,
      "grad_norm": 1.0235689781648594,
      "learning_rate": 1.8254307719250214e-07,
      "loss": 0.1876,
      "step": 1695
    },
    {
      "epoch": 0.8840815794617841,
      "grad_norm": 1.0182076394574577,
      "learning_rate": 1.8092958593938476e-07,
      "loss": 0.178,
      "step": 1696
    },
    {
      "epoch": 0.8846028539779761,
      "grad_norm": 0.9805280925563719,
      "learning_rate": 1.793229894093479e-07,
      "loss": 0.1771,
      "step": 1697
    },
    {
      "epoch": 0.8851241284941682,
      "grad_norm": 0.9615291181545174,
      "learning_rate": 1.7772329237889158e-07,
      "loss": 0.1782,
      "step": 1698
    },
    {
      "epoch": 0.8856454030103603,
      "grad_norm": 0.9507665938359553,
      "learning_rate": 1.7613049960400057e-07,
      "loss": 0.1761,
      "step": 1699
    },
    {
      "epoch": 0.8861666775265524,
      "grad_norm": 0.9922955728775525,
      "learning_rate": 1.7454461582013665e-07,
      "loss": 0.1656,
      "step": 1700
    },
    {
      "epoch": 0.8866879520427445,
      "grad_norm": 0.9994874120091833,
      "learning_rate": 1.7296564574221796e-07,
      "loss": 0.1829,
      "step": 1701
    },
    {
      "epoch": 0.8872092265589366,
      "grad_norm": 1.0136266446019506,
      "learning_rate": 1.7139359406460888e-07,
      "loss": 0.174,
      "step": 1702
    },
    {
      "epoch": 0.8877305010751287,
      "grad_norm": 0.9290349996522272,
      "learning_rate": 1.698284654611057e-07,
      "loss": 0.1709,
      "step": 1703
    },
    {
      "epoch": 0.8882517755913207,
      "grad_norm": 0.9395954285051222,
      "learning_rate": 1.6827026458492018e-07,
      "loss": 0.1808,
      "step": 1704
    },
    {
      "epoch": 0.8887730501075128,
      "grad_norm": 0.9943901878207128,
      "learning_rate": 1.6671899606866966e-07,
      "loss": 0.1835,
      "step": 1705
    },
    {
      "epoch": 0.889294324623705,
      "grad_norm": 0.967900298914638,
      "learning_rate": 1.651746645243596e-07,
      "loss": 0.1679,
      "step": 1706
    },
    {
      "epoch": 0.889815599139897,
      "grad_norm": 0.9619624947520601,
      "learning_rate": 1.6363727454337202e-07,
      "loss": 0.1767,
      "step": 1707
    },
    {
      "epoch": 0.8903368736560892,
      "grad_norm": 1.0538594805661374,
      "learning_rate": 1.6210683069645212e-07,
      "loss": 0.1799,
      "step": 1708
    },
    {
      "epoch": 0.8908581481722813,
      "grad_norm": 0.9162046996093606,
      "learning_rate": 1.6058333753369265e-07,
      "loss": 0.1651,
      "step": 1709
    },
    {
      "epoch": 0.8913794226884734,
      "grad_norm": 1.0302723932161726,
      "learning_rate": 1.5906679958452215e-07,
      "loss": 0.1893,
      "step": 1710
    },
    {
      "epoch": 0.8919006972046654,
      "grad_norm": 0.9303753947237247,
      "learning_rate": 1.5755722135769036e-07,
      "loss": 0.178,
      "step": 1711
    },
    {
      "epoch": 0.8924219717208575,
      "grad_norm": 0.9865120547157974,
      "learning_rate": 1.5605460734125582e-07,
      "loss": 0.1841,
      "step": 1712
    },
    {
      "epoch": 0.8929432462370496,
      "grad_norm": 0.9679394859027707,
      "learning_rate": 1.545589620025731e-07,
      "loss": 0.1772,
      "step": 1713
    },
    {
      "epoch": 0.8934645207532417,
      "grad_norm": 0.9911898955173043,
      "learning_rate": 1.530702897882766e-07,
      "loss": 0.1743,
      "step": 1714
    },
    {
      "epoch": 0.8939857952694338,
      "grad_norm": 0.9569232485846784,
      "learning_rate": 1.5158859512427044e-07,
      "loss": 0.1797,
      "step": 1715
    },
    {
      "epoch": 0.8945070697856259,
      "grad_norm": 1.03339302500057,
      "learning_rate": 1.501138824157139e-07,
      "loss": 0.1722,
      "step": 1716
    },
    {
      "epoch": 0.895028344301818,
      "grad_norm": 0.9977760722380256,
      "learning_rate": 1.4864615604700872e-07,
      "loss": 0.1839,
      "step": 1717
    },
    {
      "epoch": 0.89554961881801,
      "grad_norm": 1.0410265865550448,
      "learning_rate": 1.471854203817852e-07,
      "loss": 0.2006,
      "step": 1718
    },
    {
      "epoch": 0.8960708933342021,
      "grad_norm": 0.9508714325641775,
      "learning_rate": 1.4573167976288994e-07,
      "loss": 0.1812,
      "step": 1719
    },
    {
      "epoch": 0.8965921678503942,
      "grad_norm": 0.9661676662319102,
      "learning_rate": 1.4428493851237403e-07,
      "loss": 0.1803,
      "step": 1720
    },
    {
      "epoch": 0.8971134423665863,
      "grad_norm": 1.0269137740020284,
      "learning_rate": 1.4284520093147818e-07,
      "loss": 0.1837,
      "step": 1721
    },
    {
      "epoch": 0.8976347168827784,
      "grad_norm": 1.0052857327761444,
      "learning_rate": 1.4141247130062069e-07,
      "loss": 0.1856,
      "step": 1722
    },
    {
      "epoch": 0.8981559913989705,
      "grad_norm": 0.9634382762365437,
      "learning_rate": 1.3998675387938476e-07,
      "loss": 0.1864,
      "step": 1723
    },
    {
      "epoch": 0.8986772659151626,
      "grad_norm": 0.8783811648251456,
      "learning_rate": 1.3856805290650644e-07,
      "loss": 0.1592,
      "step": 1724
    },
    {
      "epoch": 0.8991985404313546,
      "grad_norm": 0.9966480441661861,
      "learning_rate": 1.3715637259986147e-07,
      "loss": 0.1749,
      "step": 1725
    },
    {
      "epoch": 0.8997198149475467,
      "grad_norm": 1.0294818185014007,
      "learning_rate": 1.3575171715645235e-07,
      "loss": 0.1798,
      "step": 1726
    },
    {
      "epoch": 0.9002410894637388,
      "grad_norm": 0.9459331203996059,
      "learning_rate": 1.343540907523963e-07,
      "loss": 0.1691,
      "step": 1727
    },
    {
      "epoch": 0.9007623639799309,
      "grad_norm": 0.9905884624653489,
      "learning_rate": 1.3296349754291366e-07,
      "loss": 0.1848,
      "step": 1728
    },
    {
      "epoch": 0.901283638496123,
      "grad_norm": 0.9777004946205554,
      "learning_rate": 1.3157994166231342e-07,
      "loss": 0.1821,
      "step": 1729
    },
    {
      "epoch": 0.9018049130123151,
      "grad_norm": 0.9862503242538797,
      "learning_rate": 1.3020342722398273e-07,
      "loss": 0.179,
      "step": 1730
    },
    {
      "epoch": 0.9023261875285072,
      "grad_norm": 0.9425193881670688,
      "learning_rate": 1.2883395832037516e-07,
      "loss": 0.1847,
      "step": 1731
    },
    {
      "epoch": 0.9028474620446993,
      "grad_norm": 0.9716954562376383,
      "learning_rate": 1.2747153902299524e-07,
      "loss": 0.1866,
      "step": 1732
    },
    {
      "epoch": 0.9033687365608913,
      "grad_norm": 1.0551193782368478,
      "learning_rate": 1.2611617338239097e-07,
      "loss": 0.1985,
      "step": 1733
    },
    {
      "epoch": 0.9038900110770834,
      "grad_norm": 0.9661387222553628,
      "learning_rate": 1.247678654281384e-07,
      "loss": 0.1728,
      "step": 1734
    },
    {
      "epoch": 0.9044112855932755,
      "grad_norm": 0.9355653605767951,
      "learning_rate": 1.234266191688302e-07,
      "loss": 0.1624,
      "step": 1735
    },
    {
      "epoch": 0.9049325601094677,
      "grad_norm": 1.0231771733704602,
      "learning_rate": 1.2209243859206577e-07,
      "loss": 0.1828,
      "step": 1736
    },
    {
      "epoch": 0.9054538346256598,
      "grad_norm": 0.9316307065559123,
      "learning_rate": 1.2076532766443578e-07,
      "loss": 0.1718,
      "step": 1737
    },
    {
      "epoch": 0.9059751091418519,
      "grad_norm": 0.9809448672654858,
      "learning_rate": 1.1944529033151436e-07,
      "loss": 0.1793,
      "step": 1738
    },
    {
      "epoch": 0.906496383658044,
      "grad_norm": 0.929916278700842,
      "learning_rate": 1.1813233051784417e-07,
      "loss": 0.1548,
      "step": 1739
    },
    {
      "epoch": 0.907017658174236,
      "grad_norm": 0.9390839502307686,
      "learning_rate": 1.1682645212692634e-07,
      "loss": 0.1735,
      "step": 1740
    },
    {
      "epoch": 0.9075389326904281,
      "grad_norm": 0.9234037263108668,
      "learning_rate": 1.1552765904120966e-07,
      "loss": 0.1723,
      "step": 1741
    },
    {
      "epoch": 0.9080602072066202,
      "grad_norm": 0.9300393036632397,
      "learning_rate": 1.1423595512207591e-07,
      "loss": 0.1707,
      "step": 1742
    },
    {
      "epoch": 0.9085814817228123,
      "grad_norm": 0.9172461012788677,
      "learning_rate": 1.1295134420983179e-07,
      "loss": 0.1727,
      "step": 1743
    },
    {
      "epoch": 0.9091027562390044,
      "grad_norm": 0.9728819849512029,
      "learning_rate": 1.1167383012369609e-07,
      "loss": 0.1719,
      "step": 1744
    },
    {
      "epoch": 0.9096240307551965,
      "grad_norm": 0.9265239942171128,
      "learning_rate": 1.1040341666178733e-07,
      "loss": 0.1732,
      "step": 1745
    },
    {
      "epoch": 0.9101453052713886,
      "grad_norm": 0.9663796287818257,
      "learning_rate": 1.0914010760111476e-07,
      "loss": 0.1814,
      "step": 1746
    },
    {
      "epoch": 0.9106665797875806,
      "grad_norm": 0.9621454320512921,
      "learning_rate": 1.078839066975651e-07,
      "loss": 0.1755,
      "step": 1747
    },
    {
      "epoch": 0.9111878543037727,
      "grad_norm": 0.9923823819866243,
      "learning_rate": 1.0663481768589196e-07,
      "loss": 0.1845,
      "step": 1748
    },
    {
      "epoch": 0.9117091288199648,
      "grad_norm": 0.9525261178445472,
      "learning_rate": 1.0539284427970559e-07,
      "loss": 0.1709,
      "step": 1749
    },
    {
      "epoch": 0.9122304033361569,
      "grad_norm": 0.9096762584124705,
      "learning_rate": 1.0415799017146094e-07,
      "loss": 0.1746,
      "step": 1750
    },
    {
      "epoch": 0.912751677852349,
      "grad_norm": 0.9350011193890286,
      "learning_rate": 1.029302590324463e-07,
      "loss": 0.173,
      "step": 1751
    },
    {
      "epoch": 0.9132729523685411,
      "grad_norm": 0.9554852724445152,
      "learning_rate": 1.0170965451277382e-07,
      "loss": 0.1766,
      "step": 1752
    },
    {
      "epoch": 0.9137942268847332,
      "grad_norm": 0.9392313516485675,
      "learning_rate": 1.0049618024136703e-07,
      "loss": 0.1831,
      "step": 1753
    },
    {
      "epoch": 0.9143155014009252,
      "grad_norm": 0.9347402829794793,
      "learning_rate": 9.928983982595175e-08,
      "loss": 0.1649,
      "step": 1754
    },
    {
      "epoch": 0.9148367759171173,
      "grad_norm": 1.0157651703635753,
      "learning_rate": 9.809063685304376e-08,
      "loss": 0.192,
      "step": 1755
    },
    {
      "epoch": 0.9153580504333094,
      "grad_norm": 0.9877484103720534,
      "learning_rate": 9.689857488793891e-08,
      "loss": 0.1837,
      "step": 1756
    },
    {
      "epoch": 0.9158793249495015,
      "grad_norm": 0.972763311198933,
      "learning_rate": 9.571365747470251e-08,
      "loss": 0.1837,
      "step": 1757
    },
    {
      "epoch": 0.9164005994656936,
      "grad_norm": 0.9461520910270284,
      "learning_rate": 9.45358881361591e-08,
      "loss": 0.1731,
      "step": 1758
    },
    {
      "epoch": 0.9169218739818857,
      "grad_norm": 0.9798701003507027,
      "learning_rate": 9.336527037388132e-08,
      "loss": 0.1756,
      "step": 1759
    },
    {
      "epoch": 0.9174431484980778,
      "grad_norm": 0.947510382571863,
      "learning_rate": 9.22018076681791e-08,
      "loss": 0.1826,
      "step": 1760
    },
    {
      "epoch": 0.9179644230142698,
      "grad_norm": 0.9414920161240843,
      "learning_rate": 9.10455034780916e-08,
      "loss": 0.1721,
      "step": 1761
    },
    {
      "epoch": 0.9184856975304619,
      "grad_norm": 1.0507502280842818,
      "learning_rate": 8.989636124137363e-08,
      "loss": 0.1827,
      "step": 1762
    },
    {
      "epoch": 0.919006972046654,
      "grad_norm": 1.012501147988642,
      "learning_rate": 8.875438437448813e-08,
      "loss": 0.1785,
      "step": 1763
    },
    {
      "epoch": 0.9195282465628462,
      "grad_norm": 0.9943000289098797,
      "learning_rate": 8.761957627259509e-08,
      "loss": 0.1718,
      "step": 1764
    },
    {
      "epoch": 0.9200495210790383,
      "grad_norm": 0.9614370675987417,
      "learning_rate": 8.649194030953989e-08,
      "loss": 0.1744,
      "step": 1765
    },
    {
      "epoch": 0.9205707955952304,
      "grad_norm": 0.9802822351352877,
      "learning_rate": 8.537147983784716e-08,
      "loss": 0.1751,
      "step": 1766
    },
    {
      "epoch": 0.9210920701114225,
      "grad_norm": 0.987961849964845,
      "learning_rate": 8.425819818870667e-08,
      "loss": 0.1757,
      "step": 1767
    },
    {
      "epoch": 0.9216133446276146,
      "grad_norm": 0.9858588747564501,
      "learning_rate": 8.3152098671965e-08,
      "loss": 0.1854,
      "step": 1768
    },
    {
      "epoch": 0.9221346191438066,
      "grad_norm": 0.9862525017141891,
      "learning_rate": 8.205318457611772e-08,
      "loss": 0.1859,
      "step": 1769
    },
    {
      "epoch": 0.9226558936599987,
      "grad_norm": 1.0116874533509572,
      "learning_rate": 8.096145916829529e-08,
      "loss": 0.1756,
      "step": 1770
    },
    {
      "epoch": 0.9231771681761908,
      "grad_norm": 0.966563812364647,
      "learning_rate": 7.98769256942572e-08,
      "loss": 0.1679,
      "step": 1771
    },
    {
      "epoch": 0.9236984426923829,
      "grad_norm": 0.9882146281688113,
      "learning_rate": 7.879958737838201e-08,
      "loss": 0.1877,
      "step": 1772
    },
    {
      "epoch": 0.924219717208575,
      "grad_norm": 0.9573779183902307,
      "learning_rate": 7.77294474236534e-08,
      "loss": 0.1773,
      "step": 1773
    },
    {
      "epoch": 0.9247409917247671,
      "grad_norm": 1.0125029487857253,
      "learning_rate": 7.66665090116575e-08,
      "loss": 0.1889,
      "step": 1774
    },
    {
      "epoch": 0.9252622662409592,
      "grad_norm": 0.9814908494118568,
      "learning_rate": 7.561077530256755e-08,
      "loss": 0.1837,
      "step": 1775
    },
    {
      "epoch": 0.9257835407571512,
      "grad_norm": 1.0057087926289285,
      "learning_rate": 7.456224943513779e-08,
      "loss": 0.1952,
      "step": 1776
    },
    {
      "epoch": 0.9263048152733433,
      "grad_norm": 0.9491641116301497,
      "learning_rate": 7.352093452669324e-08,
      "loss": 0.1787,
      "step": 1777
    },
    {
      "epoch": 0.9268260897895354,
      "grad_norm": 0.9733883352714918,
      "learning_rate": 7.248683367311937e-08,
      "loss": 0.1841,
      "step": 1778
    },
    {
      "epoch": 0.9273473643057275,
      "grad_norm": 1.0650855434205495,
      "learning_rate": 7.145994994885497e-08,
      "loss": 0.1821,
      "step": 1779
    },
    {
      "epoch": 0.9278686388219196,
      "grad_norm": 0.9395412545960169,
      "learning_rate": 7.044028640688122e-08,
      "loss": 0.1768,
      "step": 1780
    },
    {
      "epoch": 0.9283899133381117,
      "grad_norm": 0.939290034198391,
      "learning_rate": 6.942784607871373e-08,
      "loss": 0.1759,
      "step": 1781
    },
    {
      "epoch": 0.9289111878543038,
      "grad_norm": 0.975299666327165,
      "learning_rate": 6.842263197439303e-08,
      "loss": 0.1776,
      "step": 1782
    },
    {
      "epoch": 0.9294324623704958,
      "grad_norm": 0.9578243465235713,
      "learning_rate": 6.742464708247548e-08,
      "loss": 0.1889,
      "step": 1783
    },
    {
      "epoch": 0.9299537368866879,
      "grad_norm": 0.976760719372984,
      "learning_rate": 6.643389437002462e-08,
      "loss": 0.1671,
      "step": 1784
    },
    {
      "epoch": 0.93047501140288,
      "grad_norm": 0.9681019698776531,
      "learning_rate": 6.545037678260257e-08,
      "loss": 0.1823,
      "step": 1785
    },
    {
      "epoch": 0.9309962859190721,
      "grad_norm": 0.9418208808797467,
      "learning_rate": 6.447409724426063e-08,
      "loss": 0.1688,
      "step": 1786
    },
    {
      "epoch": 0.9315175604352642,
      "grad_norm": 0.9542619947417413,
      "learning_rate": 6.350505865753142e-08,
      "loss": 0.1722,
      "step": 1787
    },
    {
      "epoch": 0.9320388349514563,
      "grad_norm": 0.9039000863194366,
      "learning_rate": 6.254326390341958e-08,
      "loss": 0.1577,
      "step": 1788
    },
    {
      "epoch": 0.9325601094676484,
      "grad_norm": 1.0115354290432894,
      "learning_rate": 6.158871584139303e-08,
      "loss": 0.1828,
      "step": 1789
    },
    {
      "epoch": 0.9330813839838404,
      "grad_norm": 0.9691949292307307,
      "learning_rate": 6.06414173093764e-08,
      "loss": 0.1841,
      "step": 1790
    },
    {
      "epoch": 0.9336026585000325,
      "grad_norm": 0.9779034851126065,
      "learning_rate": 5.970137112373903e-08,
      "loss": 0.1789,
      "step": 1791
    },
    {
      "epoch": 0.9341239330162247,
      "grad_norm": 1.0407528483833595,
      "learning_rate": 5.876858007929004e-08,
      "loss": 0.1953,
      "step": 1792
    },
    {
      "epoch": 0.9346452075324168,
      "grad_norm": 0.9502877127875533,
      "learning_rate": 5.784304694926773e-08,
      "loss": 0.1655,
      "step": 1793
    },
    {
      "epoch": 0.9351664820486089,
      "grad_norm": 0.9913769974126586,
      "learning_rate": 5.692477448533351e-08,
      "loss": 0.1764,
      "step": 1794
    },
    {
      "epoch": 0.935687756564801,
      "grad_norm": 0.9549107760173099,
      "learning_rate": 5.601376541756076e-08,
      "loss": 0.1748,
      "step": 1795
    },
    {
      "epoch": 0.9362090310809931,
      "grad_norm": 0.929469099400491,
      "learning_rate": 5.511002245442987e-08,
      "loss": 0.1738,
      "step": 1796
    },
    {
      "epoch": 0.9367303055971851,
      "grad_norm": 0.9376408086550527,
      "learning_rate": 5.4213548282817664e-08,
      "loss": 0.1787,
      "step": 1797
    },
    {
      "epoch": 0.9372515801133772,
      "grad_norm": 0.9752712888369567,
      "learning_rate": 5.3324345567990485e-08,
      "loss": 0.1811,
      "step": 1798
    },
    {
      "epoch": 0.9377728546295693,
      "grad_norm": 0.9339109536282504,
      "learning_rate": 5.244241695359753e-08,
      "loss": 0.177,
      "step": 1799
    },
    {
      "epoch": 0.9382941291457614,
      "grad_norm": 0.9362640052206154,
      "learning_rate": 5.156776506166028e-08,
      "loss": 0.1651,
      "step": 1800
    },
    {
      "epoch": 0.9388154036619535,
      "grad_norm": 1.023708920025889,
      "learning_rate": 5.070039249256642e-08,
      "loss": 0.2017,
      "step": 1801
    },
    {
      "epoch": 0.9393366781781456,
      "grad_norm": 1.0157857758661593,
      "learning_rate": 4.984030182506233e-08,
      "loss": 0.1755,
      "step": 1802
    },
    {
      "epoch": 0.9398579526943377,
      "grad_norm": 0.9717374919195858,
      "learning_rate": 4.89874956162445e-08,
      "loss": 0.183,
      "step": 1803
    },
    {
      "epoch": 0.9403792272105298,
      "grad_norm": 0.9900306970894935,
      "learning_rate": 4.814197640155199e-08,
      "loss": 0.1795,
      "step": 1804
    },
    {
      "epoch": 0.9409005017267218,
      "grad_norm": 0.967636437974983,
      "learning_rate": 4.73037466947604e-08,
      "loss": 0.1775,
      "step": 1805
    },
    {
      "epoch": 0.9414217762429139,
      "grad_norm": 0.9676033944229104,
      "learning_rate": 4.64728089879718e-08,
      "loss": 0.1862,
      "step": 1806
    },
    {
      "epoch": 0.941943050759106,
      "grad_norm": 0.9423911448551001,
      "learning_rate": 4.564916575160977e-08,
      "loss": 0.1729,
      "step": 1807
    },
    {
      "epoch": 0.9424643252752981,
      "grad_norm": 0.9835107361519987,
      "learning_rate": 4.4832819434410535e-08,
      "loss": 0.1844,
      "step": 1808
    },
    {
      "epoch": 0.9429855997914902,
      "grad_norm": 0.9705696207920599,
      "learning_rate": 4.402377246341627e-08,
      "loss": 0.17,
      "step": 1809
    },
    {
      "epoch": 0.9435068743076823,
      "grad_norm": 1.011121337318435,
      "learning_rate": 4.322202724396818e-08,
      "loss": 0.1765,
      "step": 1810
    },
    {
      "epoch": 0.9440281488238744,
      "grad_norm": 0.9970207191170647,
      "learning_rate": 4.2427586159697886e-08,
      "loss": 0.1811,
      "step": 1811
    },
    {
      "epoch": 0.9445494233400664,
      "grad_norm": 0.9538788894330599,
      "learning_rate": 4.164045157252272e-08,
      "loss": 0.1796,
      "step": 1812
    },
    {
      "epoch": 0.9450706978562585,
      "grad_norm": 0.9552600819442721,
      "learning_rate": 4.086062582263656e-08,
      "loss": 0.1669,
      "step": 1813
    },
    {
      "epoch": 0.9455919723724506,
      "grad_norm": 1.0041872565936156,
      "learning_rate": 4.008811122850426e-08,
      "loss": 0.1843,
      "step": 1814
    },
    {
      "epoch": 0.9461132468886427,
      "grad_norm": 1.0088670031477684,
      "learning_rate": 3.9322910086853625e-08,
      "loss": 0.1894,
      "step": 1815
    },
    {
      "epoch": 0.9466345214048348,
      "grad_norm": 0.9739660114381572,
      "learning_rate": 3.856502467266987e-08,
      "loss": 0.1737,
      "step": 1816
    },
    {
      "epoch": 0.9471557959210269,
      "grad_norm": 0.9807421580490452,
      "learning_rate": 3.7814457239187255e-08,
      "loss": 0.1792,
      "step": 1817
    },
    {
      "epoch": 0.947677070437219,
      "grad_norm": 1.0412592625400767,
      "learning_rate": 3.707121001788438e-08,
      "loss": 0.1973,
      "step": 1818
    },
    {
      "epoch": 0.948198344953411,
      "grad_norm": 0.9606151347345824,
      "learning_rate": 3.633528521847507e-08,
      "loss": 0.1765,
      "step": 1819
    },
    {
      "epoch": 0.9487196194696031,
      "grad_norm": 0.9910087858061905,
      "learning_rate": 3.5606685028904686e-08,
      "loss": 0.1896,
      "step": 1820
    },
    {
      "epoch": 0.9492408939857953,
      "grad_norm": 1.0248547952029403,
      "learning_rate": 3.4885411615341034e-08,
      "loss": 0.1827,
      "step": 1821
    },
    {
      "epoch": 0.9497621685019874,
      "grad_norm": 0.9804471832119095,
      "learning_rate": 3.4171467122169344e-08,
      "loss": 0.1791,
      "step": 1822
    },
    {
      "epoch": 0.9502834430181795,
      "grad_norm": 0.9426961210980224,
      "learning_rate": 3.346485367198588e-08,
      "loss": 0.1821,
      "step": 1823
    },
    {
      "epoch": 0.9508047175343716,
      "grad_norm": 0.9666961026314688,
      "learning_rate": 3.2765573365591e-08,
      "loss": 0.1771,
      "step": 1824
    },
    {
      "epoch": 0.9513259920505637,
      "grad_norm": 0.9843902692560007,
      "learning_rate": 3.207362828198307e-08,
      "loss": 0.1812,
      "step": 1825
    },
    {
      "epoch": 0.9518472665667557,
      "grad_norm": 0.954584741810203,
      "learning_rate": 3.138902047835263e-08,
      "loss": 0.1781,
      "step": 1826
    },
    {
      "epoch": 0.9523685410829478,
      "grad_norm": 0.9758177238201036,
      "learning_rate": 3.071175199007653e-08,
      "loss": 0.1763,
      "step": 1827
    },
    {
      "epoch": 0.9528898155991399,
      "grad_norm": 0.9565731632841014,
      "learning_rate": 3.00418248307105e-08,
      "loss": 0.1762,
      "step": 1828
    },
    {
      "epoch": 0.953411090115332,
      "grad_norm": 0.982455562401073,
      "learning_rate": 2.9379240991984635e-08,
      "loss": 0.1769,
      "step": 1829
    },
    {
      "epoch": 0.9539323646315241,
      "grad_norm": 0.9777036199887083,
      "learning_rate": 2.8724002443797083e-08,
      "loss": 0.1694,
      "step": 1830
    },
    {
      "epoch": 0.9544536391477162,
      "grad_norm": 0.9873801352066159,
      "learning_rate": 2.807611113420816e-08,
      "loss": 0.1901,
      "step": 1831
    },
    {
      "epoch": 0.9549749136639083,
      "grad_norm": 0.9383563774784727,
      "learning_rate": 2.743556898943345e-08,
      "loss": 0.1732,
      "step": 1832
    },
    {
      "epoch": 0.9554961881801003,
      "grad_norm": 0.9941922540262297,
      "learning_rate": 2.680237791384044e-08,
      "loss": 0.1687,
      "step": 1833
    },
    {
      "epoch": 0.9560174626962924,
      "grad_norm": 0.9811859829646321,
      "learning_rate": 2.6176539789939947e-08,
      "loss": 0.1787,
      "step": 1834
    },
    {
      "epoch": 0.9565387372124845,
      "grad_norm": 0.9723634244862422,
      "learning_rate": 2.5558056478383887e-08,
      "loss": 0.1721,
      "step": 1835
    },
    {
      "epoch": 0.9570600117286766,
      "grad_norm": 0.9529231554509275,
      "learning_rate": 2.4946929817956376e-08,
      "loss": 0.1726,
      "step": 1836
    },
    {
      "epoch": 0.9575812862448687,
      "grad_norm": 0.9847804287996057,
      "learning_rate": 2.4343161625570433e-08,
      "loss": 0.1799,
      "step": 1837
    },
    {
      "epoch": 0.9581025607610608,
      "grad_norm": 0.9941204892407931,
      "learning_rate": 2.3746753696262113e-08,
      "loss": 0.1767,
      "step": 1838
    },
    {
      "epoch": 0.9586238352772529,
      "grad_norm": 0.9577619056788892,
      "learning_rate": 2.3157707803184438e-08,
      "loss": 0.1735,
      "step": 1839
    },
    {
      "epoch": 0.959145109793445,
      "grad_norm": 1.0057338275939725,
      "learning_rate": 2.2576025697603198e-08,
      "loss": 0.179,
      "step": 1840
    },
    {
      "epoch": 0.959666384309637,
      "grad_norm": 0.9547311082594001,
      "learning_rate": 2.2001709108891155e-08,
      "loss": 0.1721,
      "step": 1841
    },
    {
      "epoch": 0.9601876588258291,
      "grad_norm": 0.9520610446159212,
      "learning_rate": 2.143475974452275e-08,
      "loss": 0.1816,
      "step": 1842
    },
    {
      "epoch": 0.9607089333420212,
      "grad_norm": 0.9653141350324072,
      "learning_rate": 2.0875179290069934e-08,
      "loss": 0.1759,
      "step": 1843
    },
    {
      "epoch": 0.9612302078582133,
      "grad_norm": 0.9856672742904604,
      "learning_rate": 2.032296940919526e-08,
      "loss": 0.1792,
      "step": 1844
    },
    {
      "epoch": 0.9617514823744054,
      "grad_norm": 1.0179941740243892,
      "learning_rate": 1.9778131743649064e-08,
      "loss": 0.1823,
      "step": 1845
    },
    {
      "epoch": 0.9622727568905975,
      "grad_norm": 0.9494362255500836,
      "learning_rate": 1.9240667913264233e-08,
      "loss": 0.1639,
      "step": 1846
    },
    {
      "epoch": 0.9627940314067897,
      "grad_norm": 0.9549890993068286,
      "learning_rate": 1.8710579515948957e-08,
      "loss": 0.1877,
      "step": 1847
    },
    {
      "epoch": 0.9633153059229816,
      "grad_norm": 0.9899460282043995,
      "learning_rate": 1.8187868127685914e-08,
      "loss": 0.1812,
      "step": 1848
    },
    {
      "epoch": 0.9638365804391738,
      "grad_norm": 0.9220303796166097,
      "learning_rate": 1.767253530252422e-08,
      "loss": 0.165,
      "step": 1849
    },
    {
      "epoch": 0.9643578549553659,
      "grad_norm": 0.9530499549516781,
      "learning_rate": 1.716458257257636e-08,
      "loss": 0.1846,
      "step": 1850
    },
    {
      "epoch": 0.964879129471558,
      "grad_norm": 0.9523111335894338,
      "learning_rate": 1.666401144801405e-08,
      "loss": 0.1845,
      "step": 1851
    },
    {
      "epoch": 0.9654004039877501,
      "grad_norm": 0.9562990194496472,
      "learning_rate": 1.6170823417062386e-08,
      "loss": 0.1809,
      "step": 1852
    },
    {
      "epoch": 0.9659216785039422,
      "grad_norm": 0.9631481610629716,
      "learning_rate": 1.568501994599625e-08,
      "loss": 0.1865,
      "step": 1853
    },
    {
      "epoch": 0.9664429530201343,
      "grad_norm": 1.0239927293856008,
      "learning_rate": 1.5206602479135857e-08,
      "loss": 0.1912,
      "step": 1854
    },
    {
      "epoch": 0.9669642275363263,
      "grad_norm": 0.9467044982817666,
      "learning_rate": 1.4735572438842605e-08,
      "loss": 0.1788,
      "step": 1855
    },
    {
      "epoch": 0.9674855020525184,
      "grad_norm": 0.9330536711923619,
      "learning_rate": 1.4271931225514624e-08,
      "loss": 0.1653,
      "step": 1856
    },
    {
      "epoch": 0.9680067765687105,
      "grad_norm": 0.9853895262009704,
      "learning_rate": 1.381568021758234e-08,
      "loss": 0.1838,
      "step": 1857
    },
    {
      "epoch": 0.9685280510849026,
      "grad_norm": 0.9829637868018,
      "learning_rate": 1.336682077150514e-08,
      "loss": 0.1801,
      "step": 1858
    },
    {
      "epoch": 0.9690493256010947,
      "grad_norm": 1.019613063122009,
      "learning_rate": 1.2925354221766661e-08,
      "loss": 0.1944,
      "step": 1859
    },
    {
      "epoch": 0.9695706001172868,
      "grad_norm": 0.9769879106201981,
      "learning_rate": 1.2491281880871175e-08,
      "loss": 0.1866,
      "step": 1860
    },
    {
      "epoch": 0.9700918746334789,
      "grad_norm": 0.9529620322406079,
      "learning_rate": 1.206460503933915e-08,
      "loss": 0.1726,
      "step": 1861
    },
    {
      "epoch": 0.9706131491496709,
      "grad_norm": 0.9871500196046391,
      "learning_rate": 1.1645324965704473e-08,
      "loss": 0.1764,
      "step": 1862
    },
    {
      "epoch": 0.971134423665863,
      "grad_norm": 0.9920422198495212,
      "learning_rate": 1.1233442906509462e-08,
      "loss": 0.1856,
      "step": 1863
    },
    {
      "epoch": 0.9716556981820551,
      "grad_norm": 0.9926789149113099,
      "learning_rate": 1.0828960086302075e-08,
      "loss": 0.1735,
      "step": 1864
    },
    {
      "epoch": 0.9721769726982472,
      "grad_norm": 1.0341066215790733,
      "learning_rate": 1.0431877707632043e-08,
      "loss": 0.1865,
      "step": 1865
    },
    {
      "epoch": 0.9726982472144393,
      "grad_norm": 0.9854581872034558,
      "learning_rate": 1.0042196951046968e-08,
      "loss": 0.1834,
      "step": 1866
    },
    {
      "epoch": 0.9732195217306314,
      "grad_norm": 1.1297948530399125,
      "learning_rate": 9.659918975088444e-09,
      "loss": 0.1955,
      "step": 1867
    },
    {
      "epoch": 0.9737407962468235,
      "grad_norm": 1.0247512856595942,
      "learning_rate": 9.285044916290675e-09,
      "loss": 0.1891,
      "step": 1868
    },
    {
      "epoch": 0.9742620707630155,
      "grad_norm": 1.004117650985569,
      "learning_rate": 8.91757588917408e-09,
      "loss": 0.1802,
      "step": 1869
    },
    {
      "epoch": 0.9747833452792076,
      "grad_norm": 0.9811005564124128,
      "learning_rate": 8.557512986244464e-09,
      "loss": 0.1824,
      "step": 1870
    },
    {
      "epoch": 0.9753046197953997,
      "grad_norm": 1.0235065678803086,
      "learning_rate": 8.20485727798831e-09,
      "loss": 0.1766,
      "step": 1871
    },
    {
      "epoch": 0.9758258943115918,
      "grad_norm": 0.9684314864439314,
      "learning_rate": 7.85960981287026e-09,
      "loss": 0.173,
      "step": 1872
    },
    {
      "epoch": 0.9763471688277839,
      "grad_norm": 1.0244425663418408,
      "learning_rate": 7.521771617330365e-09,
      "loss": 0.1688,
      "step": 1873
    },
    {
      "epoch": 0.976868443343976,
      "grad_norm": 1.0859697598329392,
      "learning_rate": 7.191343695779618e-09,
      "loss": 0.1881,
      "step": 1874
    },
    {
      "epoch": 0.9773897178601682,
      "grad_norm": 1.0027582554431416,
      "learning_rate": 6.86832703059831e-09,
      "loss": 0.1907,
      "step": 1875
    },
    {
      "epoch": 0.9779109923763603,
      "grad_norm": 0.966843277209175,
      "learning_rate": 6.552722582132687e-09,
      "loss": 0.1774,
      "step": 1876
    },
    {
      "epoch": 0.9784322668925522,
      "grad_norm": 0.9805017694840367,
      "learning_rate": 6.24453128869218e-09,
      "loss": 0.1806,
      "step": 1877
    },
    {
      "epoch": 0.9789535414087444,
      "grad_norm": 0.9485250193876247,
      "learning_rate": 5.94375406654607e-09,
      "loss": 0.1768,
      "step": 1878
    },
    {
      "epoch": 0.9794748159249365,
      "grad_norm": 0.9889744757779998,
      "learning_rate": 5.650391809922107e-09,
      "loss": 0.1861,
      "step": 1879
    },
    {
      "epoch": 0.9799960904411286,
      "grad_norm": 0.9203513698825401,
      "learning_rate": 5.364445391001228e-09,
      "loss": 0.1728,
      "step": 1880
    },
    {
      "epoch": 0.9805173649573207,
      "grad_norm": 0.9577200856228432,
      "learning_rate": 5.085915659918672e-09,
      "loss": 0.1743,
      "step": 1881
    },
    {
      "epoch": 0.9810386394735128,
      "grad_norm": 1.0558998893036378,
      "learning_rate": 4.814803444758431e-09,
      "loss": 0.1763,
      "step": 1882
    },
    {
      "epoch": 0.9815599139897049,
      "grad_norm": 1.030845679885158,
      "learning_rate": 4.551109551551581e-09,
      "loss": 0.1911,
      "step": 1883
    },
    {
      "epoch": 0.9820811885058969,
      "grad_norm": 1.0383646331449794,
      "learning_rate": 4.294834764274614e-09,
      "loss": 0.175,
      "step": 1884
    },
    {
      "epoch": 0.982602463022089,
      "grad_norm": 0.9543632616013837,
      "learning_rate": 4.04597984484667e-09,
      "loss": 0.1782,
      "step": 1885
    },
    {
      "epoch": 0.9831237375382811,
      "grad_norm": 0.9717416279826157,
      "learning_rate": 3.804545533126758e-09,
      "loss": 0.183,
      "step": 1886
    },
    {
      "epoch": 0.9836450120544732,
      "grad_norm": 0.9729548931314076,
      "learning_rate": 3.5705325469123644e-09,
      "loss": 0.1711,
      "step": 1887
    },
    {
      "epoch": 0.9841662865706653,
      "grad_norm": 0.9930751786965978,
      "learning_rate": 3.3439415819369604e-09,
      "loss": 0.1839,
      "step": 1888
    },
    {
      "epoch": 0.9846875610868574,
      "grad_norm": 0.9311441519535106,
      "learning_rate": 3.124773311867779e-09,
      "loss": 0.1733,
      "step": 1889
    },
    {
      "epoch": 0.9852088356030495,
      "grad_norm": 0.9076890202701972,
      "learning_rate": 2.9130283883044285e-09,
      "loss": 0.1617,
      "step": 1890
    },
    {
      "epoch": 0.9857301101192415,
      "grad_norm": 0.95940450023715,
      "learning_rate": 2.70870744077556e-09,
      "loss": 0.1814,
      "step": 1891
    },
    {
      "epoch": 0.9862513846354336,
      "grad_norm": 0.9828810600577166,
      "learning_rate": 2.5118110767388682e-09,
      "loss": 0.1815,
      "step": 1892
    },
    {
      "epoch": 0.9867726591516257,
      "grad_norm": 0.9701859016924141,
      "learning_rate": 2.322339881577762e-09,
      "loss": 0.1867,
      "step": 1893
    },
    {
      "epoch": 0.9872939336678178,
      "grad_norm": 0.9454338957561875,
      "learning_rate": 2.140294418600808e-09,
      "loss": 0.1683,
      "step": 1894
    },
    {
      "epoch": 0.9878152081840099,
      "grad_norm": 0.9976385720375264,
      "learning_rate": 1.965675229038677e-09,
      "loss": 0.1847,
      "step": 1895
    },
    {
      "epoch": 0.988336482700202,
      "grad_norm": 0.9459984251155135,
      "learning_rate": 1.7984828320444236e-09,
      "loss": 0.1759,
      "step": 1896
    },
    {
      "epoch": 0.9888577572163941,
      "grad_norm": 0.9943708549205489,
      "learning_rate": 1.6387177246893205e-09,
      "loss": 0.1794,
      "step": 1897
    },
    {
      "epoch": 0.9893790317325861,
      "grad_norm": 0.955276488511367,
      "learning_rate": 1.486380381964525e-09,
      "loss": 0.1804,
      "step": 1898
    },
    {
      "epoch": 0.9899003062487782,
      "grad_norm": 1.0166893589029946,
      "learning_rate": 1.3414712567769161e-09,
      "loss": 0.1828,
      "step": 1899
    },
    {
      "epoch": 0.9904215807649703,
      "grad_norm": 0.9808551871654355,
      "learning_rate": 1.2039907799490935e-09,
      "loss": 0.1801,
      "step": 1900
    },
    {
      "epoch": 0.9909428552811624,
      "grad_norm": 0.9382284653572882,
      "learning_rate": 1.0739393602185454e-09,
      "loss": 0.1755,
      "step": 1901
    },
    {
      "epoch": 0.9914641297973545,
      "grad_norm": 0.9763362834740261,
      "learning_rate": 9.513173842348732e-10,
      "loss": 0.1699,
      "step": 1902
    },
    {
      "epoch": 0.9919854043135466,
      "grad_norm": 1.0171291150299013,
      "learning_rate": 8.361252165597911e-10,
      "loss": 0.2021,
      "step": 1903
    },
    {
      "epoch": 0.9925066788297388,
      "grad_norm": 0.9744374859882523,
      "learning_rate": 7.283631996662933e-10,
      "loss": 0.1709,
      "step": 1904
    },
    {
      "epoch": 0.9930279533459307,
      "grad_norm": 0.9653344994919608,
      "learning_rate": 6.28031653936434e-10,
      "loss": 0.1874,
      "step": 1905
    },
    {
      "epoch": 0.9935492278621229,
      "grad_norm": 0.9621859639605892,
      "learning_rate": 5.351308776613273e-10,
      "loss": 0.1696,
      "step": 1906
    },
    {
      "epoch": 0.994070502378315,
      "grad_norm": 0.9895447054675127,
      "learning_rate": 4.49661147040592e-10,
      "loss": 0.1886,
      "step": 1907
    },
    {
      "epoch": 0.9945917768945071,
      "grad_norm": 0.956049965794608,
      "learning_rate": 3.7162271617985357e-10,
      "loss": 0.1804,
      "step": 1908
    },
    {
      "epoch": 0.9951130514106992,
      "grad_norm": 0.9678203013495367,
      "learning_rate": 3.0101581709185424e-10,
      "loss": 0.1739,
      "step": 1909
    },
    {
      "epoch": 0.9956343259268913,
      "grad_norm": 0.9580407642873671,
      "learning_rate": 2.3784065969451043e-10,
      "loss": 0.1751,
      "step": 1910
    },
    {
      "epoch": 0.9961556004430834,
      "grad_norm": 0.9948205703446217,
      "learning_rate": 1.8209743181146766e-10,
      "loss": 0.1773,
      "step": 1911
    },
    {
      "epoch": 0.9966768749592755,
      "grad_norm": 1.0286522179292514,
      "learning_rate": 1.3378629917015772e-10,
      "loss": 0.1898,
      "step": 1912
    },
    {
      "epoch": 0.9971981494754675,
      "grad_norm": 1.0318534355598075,
      "learning_rate": 9.290740540207621e-11,
      "loss": 0.2015,
      "step": 1913
    },
    {
      "epoch": 0.9977194239916596,
      "grad_norm": 1.0006269657703835,
      "learning_rate": 5.94608720427825e-11,
      "loss": 0.1774,
      "step": 1914
    },
    {
      "epoch": 0.9982406985078517,
      "grad_norm": 0.9509156275592338,
      "learning_rate": 3.344679853023447e-11,
      "loss": 0.1849,
      "step": 1915
    },
    {
      "epoch": 0.9987619730240438,
      "grad_norm": 0.9960321151226665,
      "learning_rate": 1.4865262205898678e-11,
      "loss": 0.1845,
      "step": 1916
    },
    {
      "epoch": 0.9992832475402359,
      "grad_norm": 0.9897232776542689,
      "learning_rate": 3.716318313640166e-12,
      "loss": 0.1807,
      "step": 1917
    },
    {
      "epoch": 0.999804522056428,
      "grad_norm": 0.9846363278734621,
      "learning_rate": 0.0,
      "loss": 0.1743,
      "step": 1918
    },
    {
      "epoch": 0.999804522056428,
      "step": 1918,
      "total_flos": 331183228846080.0,
      "train_loss": 0.20316935279353193,
      "train_runtime": 20098.2432,
      "train_samples_per_second": 6.108,
      "train_steps_per_second": 0.095
    }
  ],
  "logging_steps": 1,
  "max_steps": 1918,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 331183228846080.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}