{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.27058335944686485,
  "eval_steps": 500,
  "global_step": 6692,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.043385526701507e-05,
      "grad_norm": 45.212158203125,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 8.4776,
      "step": 1
    },
    {
      "epoch": 8.086771053403014e-05,
      "grad_norm": 42.83491897583008,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 7.4022,
      "step": 2
    },
    {
      "epoch": 0.00012130156580104521,
      "grad_norm": 47.69302749633789,
      "learning_rate": 3e-06,
      "loss": 8.5953,
      "step": 3
    },
    {
      "epoch": 0.00016173542106806028,
      "grad_norm": 47.8973503112793,
      "learning_rate": 4.000000000000001e-06,
      "loss": 7.7783,
      "step": 4
    },
    {
      "epoch": 0.00020216927633507535,
      "grad_norm": 43.4521598815918,
      "learning_rate": 5e-06,
      "loss": 7.9174,
      "step": 5
    },
    {
      "epoch": 0.00024260313160209043,
      "grad_norm": 46.271541595458984,
      "learning_rate": 6e-06,
      "loss": 8.6758,
      "step": 6
    },
    {
      "epoch": 0.0002830369868691055,
      "grad_norm": 45.16845703125,
      "learning_rate": 7.000000000000001e-06,
      "loss": 7.8099,
      "step": 7
    },
    {
      "epoch": 0.00032347084213612057,
      "grad_norm": 43.50260543823242,
      "learning_rate": 8.000000000000001e-06,
      "loss": 7.1801,
      "step": 8
    },
    {
      "epoch": 0.00036390469740313564,
      "grad_norm": 41.92473602294922,
      "learning_rate": 9e-06,
      "loss": 7.1574,
      "step": 9
    },
    {
      "epoch": 0.0004043385526701507,
      "grad_norm": 43.03952407836914,
      "learning_rate": 1e-05,
      "loss": 7.1354,
      "step": 10
    },
    {
      "epoch": 0.0004447724079371658,
      "grad_norm": 41.086021423339844,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 7.3757,
      "step": 11
    },
    {
      "epoch": 0.00048520626320418085,
      "grad_norm": 43.03165817260742,
      "learning_rate": 1.2e-05,
      "loss": 7.2673,
      "step": 12
    },
    {
      "epoch": 0.0005256401184711959,
      "grad_norm": 41.32279586791992,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 7.0322,
      "step": 13
    },
    {
      "epoch": 0.000566073973738211,
      "grad_norm": 40.678192138671875,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 6.9458,
      "step": 14
    },
    {
      "epoch": 0.0006065078290052261,
      "grad_norm": 42.551509857177734,
      "learning_rate": 1.5e-05,
      "loss": 7.4505,
      "step": 15
    },
    {
      "epoch": 0.0006469416842722411,
      "grad_norm": 39.79718017578125,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 6.314,
      "step": 16
    },
    {
      "epoch": 0.0006873755395392562,
      "grad_norm": 39.78065490722656,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 6.0012,
      "step": 17
    },
    {
      "epoch": 0.0007278093948062713,
      "grad_norm": 34.97587966918945,
      "learning_rate": 1.8e-05,
      "loss": 5.7898,
      "step": 18
    },
    {
      "epoch": 0.0007682432500732863,
      "grad_norm": 34.85056686401367,
      "learning_rate": 1.9e-05,
      "loss": 5.3237,
      "step": 19
    },
    {
      "epoch": 0.0008086771053403014,
      "grad_norm": 31.200321197509766,
      "learning_rate": 2e-05,
      "loss": 4.8291,
      "step": 20
    },
    {
      "epoch": 0.0008491109606073165,
      "grad_norm": 34.75196075439453,
      "learning_rate": 2.1e-05,
      "loss": 5.7917,
      "step": 21
    },
    {
      "epoch": 0.0008895448158743316,
      "grad_norm": 25.683300018310547,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 3.6869,
      "step": 22
    },
    {
      "epoch": 0.0009299786711413466,
      "grad_norm": 23.978288650512695,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 3.6214,
      "step": 23
    },
    {
      "epoch": 0.0009704125264083617,
      "grad_norm": 24.98045539855957,
      "learning_rate": 2.4e-05,
      "loss": 3.8051,
      "step": 24
    },
    {
      "epoch": 0.0010108463816753768,
      "grad_norm": 25.768600463867188,
      "learning_rate": 2.5e-05,
      "loss": 3.8138,
      "step": 25
    },
    {
      "epoch": 0.0010512802369423918,
      "grad_norm": 28.268779754638672,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 3.6824,
      "step": 26
    },
    {
      "epoch": 0.001091714092209407,
      "grad_norm": 24.55759620666504,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 2.6137,
      "step": 27
    },
    {
      "epoch": 0.001132147947476422,
      "grad_norm": 32.37775421142578,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 3.0377,
      "step": 28
    },
    {
      "epoch": 0.001172581802743437,
      "grad_norm": 24.953506469726562,
      "learning_rate": 2.9e-05,
      "loss": 2.3354,
      "step": 29
    },
    {
      "epoch": 0.0012130156580104521,
      "grad_norm": 22.368303298950195,
      "learning_rate": 3e-05,
      "loss": 1.6599,
      "step": 30
    },
    {
      "epoch": 0.0012534495132774672,
      "grad_norm": 26.778047561645508,
      "learning_rate": 3.1e-05,
      "loss": 1.6387,
      "step": 31
    },
    {
      "epoch": 0.0012938833685444823,
      "grad_norm": 13.506389617919922,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.9049,
      "step": 32
    },
    {
      "epoch": 0.0013343172238114973,
      "grad_norm": 18.525230407714844,
      "learning_rate": 3.3e-05,
      "loss": 0.8052,
      "step": 33
    },
    {
      "epoch": 0.0013747510790785124,
      "grad_norm": 25.12320899963379,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.8044,
      "step": 34
    },
    {
      "epoch": 0.0014151849343455275,
      "grad_norm": 16.060945510864258,
      "learning_rate": 3.5e-05,
      "loss": 0.6799,
      "step": 35
    },
    {
      "epoch": 0.0014556187896125426,
      "grad_norm": 18.008813858032227,
      "learning_rate": 3.6e-05,
      "loss": 0.4561,
      "step": 36
    },
    {
      "epoch": 0.0014960526448795576,
      "grad_norm": 21.687620162963867,
      "learning_rate": 3.7e-05,
      "loss": 0.613,
      "step": 37
    },
    {
      "epoch": 0.0015364865001465727,
      "grad_norm": 14.03872013092041,
      "learning_rate": 3.8e-05,
      "loss": 0.5134,
      "step": 38
    },
    {
      "epoch": 0.0015769203554135878,
      "grad_norm": 8.90583610534668,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 0.2898,
      "step": 39
    },
    {
      "epoch": 0.0016173542106806028,
      "grad_norm": 15.97493839263916,
      "learning_rate": 4e-05,
      "loss": 0.4115,
      "step": 40
    },
    {
      "epoch": 0.001657788065947618,
      "grad_norm": 6.131041526794434,
      "learning_rate": 4.1e-05,
      "loss": 0.3075,
      "step": 41
    },
    {
      "epoch": 0.001698221921214633,
      "grad_norm": 21.00753402709961,
      "learning_rate": 4.2e-05,
      "loss": 0.5586,
      "step": 42
    },
    {
      "epoch": 0.001738655776481648,
      "grad_norm": 23.8162899017334,
      "learning_rate": 4.3e-05,
      "loss": 0.6964,
      "step": 43
    },
    {
      "epoch": 0.0017790896317486631,
      "grad_norm": 22.47564125061035,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.543,
      "step": 44
    },
    {
      "epoch": 0.0018195234870156782,
      "grad_norm": 20.058208465576172,
      "learning_rate": 4.5e-05,
      "loss": 0.5529,
      "step": 45
    },
    {
      "epoch": 0.0018599573422826933,
      "grad_norm": 9.034168243408203,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.4264,
      "step": 46
    },
    {
      "epoch": 0.0019003911975497083,
      "grad_norm": 13.754554748535156,
      "learning_rate": 4.7e-05,
      "loss": 0.4332,
      "step": 47
    },
    {
      "epoch": 0.0019408250528167234,
      "grad_norm": 16.2254638671875,
      "learning_rate": 4.8e-05,
      "loss": 0.5085,
      "step": 48
    },
    {
      "epoch": 0.0019812589080837385,
      "grad_norm": 13.377281188964844,
      "learning_rate": 4.9e-05,
      "loss": 0.4047,
      "step": 49
    },
    {
      "epoch": 0.0020216927633507535,
      "grad_norm": 16.529783248901367,
      "learning_rate": 5e-05,
      "loss": 0.602,
      "step": 50
    },
    {
      "epoch": 0.0020621266186177686,
      "grad_norm": 16.30471420288086,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 0.4734,
      "step": 51
    },
    {
      "epoch": 0.0021025604738847837,
      "grad_norm": 9.81867790222168,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 0.4394,
      "step": 52
    },
    {
      "epoch": 0.0021429943291517988,
      "grad_norm": 8.821556091308594,
      "learning_rate": 5.300000000000001e-05,
      "loss": 0.3627,
      "step": 53
    },
    {
      "epoch": 0.002183428184418814,
      "grad_norm": 7.72442626953125,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 0.3547,
      "step": 54
    },
    {
      "epoch": 0.002223862039685829,
      "grad_norm": 9.638863563537598,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.3275,
      "step": 55
    },
    {
      "epoch": 0.002264295894952844,
      "grad_norm": 6.1317458152771,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 0.2867,
      "step": 56
    },
    {
      "epoch": 0.002304729750219859,
      "grad_norm": 11.842965126037598,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 0.3486,
      "step": 57
    },
    {
      "epoch": 0.002345163605486874,
      "grad_norm": 3.987241506576538,
      "learning_rate": 5.8e-05,
      "loss": 0.2699,
      "step": 58
    },
    {
      "epoch": 0.002385597460753889,
      "grad_norm": 6.591022968292236,
      "learning_rate": 5.9e-05,
      "loss": 0.3184,
      "step": 59
    },
    {
      "epoch": 0.0024260313160209043,
      "grad_norm": 7.872280120849609,
      "learning_rate": 6e-05,
      "loss": 0.3346,
      "step": 60
    },
    {
      "epoch": 0.0024664651712879193,
      "grad_norm": 2.6104869842529297,
      "learning_rate": 6.1e-05,
      "loss": 0.3243,
      "step": 61
    },
    {
      "epoch": 0.0025068990265549344,
      "grad_norm": 3.023655652999878,
      "learning_rate": 6.2e-05,
      "loss": 0.3306,
      "step": 62
    },
    {
      "epoch": 0.0025473328818219495,
      "grad_norm": 6.13469123840332,
      "learning_rate": 6.3e-05,
      "loss": 0.344,
      "step": 63
    },
    {
      "epoch": 0.0025877667370889645,
      "grad_norm": 6.2675957679748535,
      "learning_rate": 6.400000000000001e-05,
      "loss": 0.3637,
      "step": 64
    },
    {
      "epoch": 0.0026282005923559796,
      "grad_norm": 15.284539222717285,
      "learning_rate": 6.500000000000001e-05,
      "loss": 0.4253,
      "step": 65
    },
    {
      "epoch": 0.0026686344476229947,
      "grad_norm": 13.781516075134277,
      "learning_rate": 6.6e-05,
      "loss": 0.3658,
      "step": 66
    },
    {
      "epoch": 0.0027090683028900098,
      "grad_norm": 3.6815264225006104,
      "learning_rate": 6.7e-05,
      "loss": 0.3152,
      "step": 67
    },
    {
      "epoch": 0.002749502158157025,
      "grad_norm": 5.936532497406006,
      "learning_rate": 6.800000000000001e-05,
      "loss": 0.312,
      "step": 68
    },
    {
      "epoch": 0.00278993601342404,
      "grad_norm": 5.848452568054199,
      "learning_rate": 6.9e-05,
      "loss": 0.2422,
      "step": 69
    },
    {
      "epoch": 0.002830369868691055,
      "grad_norm": 19.137374877929688,
      "learning_rate": 7e-05,
      "loss": 0.4338,
      "step": 70
    },
    {
      "epoch": 0.00287080372395807,
      "grad_norm": 10.636536598205566,
      "learning_rate": 7.1e-05,
      "loss": 0.3493,
      "step": 71
    },
    {
      "epoch": 0.002911237579225085,
      "grad_norm": 4.964332580566406,
      "learning_rate": 7.2e-05,
      "loss": 0.2712,
      "step": 72
    },
    {
      "epoch": 0.0029516714344921,
      "grad_norm": 8.327373504638672,
      "learning_rate": 7.3e-05,
      "loss": 0.2796,
      "step": 73
    },
    {
      "epoch": 0.0029921052897591153,
      "grad_norm": 8.643411636352539,
      "learning_rate": 7.4e-05,
      "loss": 0.3479,
      "step": 74
    },
    {
      "epoch": 0.0030325391450261303,
      "grad_norm": 9.094339370727539,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.3282,
      "step": 75
    },
    {
      "epoch": 0.0030729730002931454,
      "grad_norm": 16.117694854736328,
      "learning_rate": 7.6e-05,
      "loss": 0.4187,
      "step": 76
    },
    {
      "epoch": 0.0031134068555601605,
      "grad_norm": 21.272748947143555,
      "learning_rate": 7.7e-05,
      "loss": 0.5197,
      "step": 77
    },
    {
      "epoch": 0.0031538407108271755,
      "grad_norm": 5.69344425201416,
      "learning_rate": 7.800000000000001e-05,
      "loss": 0.2646,
      "step": 78
    },
    {
      "epoch": 0.0031942745660941906,
      "grad_norm": 7.0776824951171875,
      "learning_rate": 7.900000000000001e-05,
      "loss": 0.282,
      "step": 79
    },
    {
      "epoch": 0.0032347084213612057,
      "grad_norm": 5.962209701538086,
      "learning_rate": 8e-05,
      "loss": 0.3024,
      "step": 80
    },
    {
      "epoch": 0.0032751422766282207,
      "grad_norm": 6.475072860717773,
      "learning_rate": 8.1e-05,
      "loss": 0.3527,
      "step": 81
    },
    {
      "epoch": 0.003315576131895236,
      "grad_norm": 10.585362434387207,
      "learning_rate": 8.2e-05,
      "loss": 0.3635,
      "step": 82
    },
    {
      "epoch": 0.003356009987162251,
      "grad_norm": 13.020295143127441,
      "learning_rate": 8.3e-05,
      "loss": 0.296,
      "step": 83
    },
    {
      "epoch": 0.003396443842429266,
      "grad_norm": 12.620179176330566,
      "learning_rate": 8.4e-05,
      "loss": 0.3679,
      "step": 84
    },
    {
      "epoch": 0.003436877697696281,
      "grad_norm": 14.295244216918945,
      "learning_rate": 8.5e-05,
      "loss": 0.3293,
      "step": 85
    },
    {
      "epoch": 0.003477311552963296,
      "grad_norm": 11.91524887084961,
      "learning_rate": 8.6e-05,
      "loss": 0.3314,
      "step": 86
    },
    {
      "epoch": 0.003517745408230311,
      "grad_norm": 4.24912166595459,
      "learning_rate": 8.7e-05,
      "loss": 0.3097,
      "step": 87
    },
    {
      "epoch": 0.0035581792634973262,
      "grad_norm": 2.8676578998565674,
      "learning_rate": 8.800000000000001e-05,
      "loss": 0.2086,
      "step": 88
    },
    {
      "epoch": 0.0035986131187643413,
      "grad_norm": 13.0736665725708,
      "learning_rate": 8.900000000000001e-05,
      "loss": 0.3789,
      "step": 89
    },
    {
      "epoch": 0.0036390469740313564,
      "grad_norm": 15.212523460388184,
      "learning_rate": 9e-05,
      "loss": 0.3977,
      "step": 90
    },
    {
      "epoch": 0.0036794808292983715,
      "grad_norm": 4.857946395874023,
      "learning_rate": 9.1e-05,
      "loss": 0.2782,
      "step": 91
    },
    {
      "epoch": 0.0037199146845653865,
      "grad_norm": 13.703444480895996,
      "learning_rate": 9.200000000000001e-05,
      "loss": 0.4441,
      "step": 92
    },
    {
      "epoch": 0.0037603485398324016,
      "grad_norm": 7.481781482696533,
      "learning_rate": 9.300000000000001e-05,
      "loss": 0.4544,
      "step": 93
    },
    {
      "epoch": 0.0038007823950994167,
      "grad_norm": 10.419188499450684,
      "learning_rate": 9.4e-05,
      "loss": 0.2466,
      "step": 94
    },
    {
      "epoch": 0.0038412162503664317,
      "grad_norm": 6.384120941162109,
      "learning_rate": 9.5e-05,
      "loss": 0.3243,
      "step": 95
    },
    {
      "epoch": 0.003881650105633447,
      "grad_norm": 5.624557971954346,
      "learning_rate": 9.6e-05,
      "loss": 0.2636,
      "step": 96
    },
    {
      "epoch": 0.003922083960900462,
      "grad_norm": 6.27712869644165,
      "learning_rate": 9.7e-05,
      "loss": 0.3291,
      "step": 97
    },
    {
      "epoch": 0.003962517816167477,
      "grad_norm": 8.306694030761719,
      "learning_rate": 9.8e-05,
      "loss": 0.3152,
      "step": 98
    },
    {
      "epoch": 0.0040029516714344925,
      "grad_norm": 10.752472877502441,
      "learning_rate": 9.900000000000001e-05,
      "loss": 0.3415,
      "step": 99
    },
    {
      "epoch": 0.004043385526701507,
      "grad_norm": 10.531058311462402,
      "learning_rate": 0.0001,
      "loss": 0.4284,
      "step": 100
    },
    {
      "epoch": 0.004083819381968523,
      "grad_norm": 16.106300354003906,
      "learning_rate": 9.99999995932986e-05,
      "loss": 0.6921,
      "step": 101
    },
    {
      "epoch": 0.004124253237235537,
      "grad_norm": 11.43233585357666,
      "learning_rate": 9.999999837319442e-05,
      "loss": 0.3084,
      "step": 102
    },
    {
      "epoch": 0.004164687092502553,
      "grad_norm": 5.925229072570801,
      "learning_rate": 9.999999633968746e-05,
      "loss": 0.2721,
      "step": 103
    },
    {
      "epoch": 0.004205120947769567,
      "grad_norm": 13.037198066711426,
      "learning_rate": 9.999999349277778e-05,
      "loss": 0.3153,
      "step": 104
    },
    {
      "epoch": 0.004245554803036583,
      "grad_norm": 11.120277404785156,
      "learning_rate": 9.999998983246538e-05,
      "loss": 0.3173,
      "step": 105
    },
    {
      "epoch": 0.0042859886583035975,
      "grad_norm": 17.156024932861328,
      "learning_rate": 9.999998535875038e-05,
      "loss": 0.582,
      "step": 106
    },
    {
      "epoch": 0.004326422513570613,
      "grad_norm": 1.7565780878067017,
      "learning_rate": 9.999998007163281e-05,
      "loss": 0.1944,
      "step": 107
    },
    {
      "epoch": 0.004366856368837628,
      "grad_norm": 4.962274074554443,
      "learning_rate": 9.999997397111278e-05,
      "loss": 0.2287,
      "step": 108
    },
    {
      "epoch": 0.004407290224104643,
      "grad_norm": 5.148132801055908,
      "learning_rate": 9.999996705719036e-05,
      "loss": 0.2062,
      "step": 109
    },
    {
      "epoch": 0.004447724079371658,
      "grad_norm": 7.428779602050781,
      "learning_rate": 9.999995932986568e-05,
      "loss": 0.2594,
      "step": 110
    },
    {
      "epoch": 0.004488157934638673,
      "grad_norm": 7.802266597747803,
      "learning_rate": 9.999995078913888e-05,
      "loss": 0.2838,
      "step": 111
    },
    {
      "epoch": 0.004528591789905688,
      "grad_norm": 9.690343856811523,
      "learning_rate": 9.999994143501008e-05,
      "loss": 0.2114,
      "step": 112
    },
    {
      "epoch": 0.0045690256451727035,
      "grad_norm": 8.961145401000977,
      "learning_rate": 9.999993126747943e-05,
      "loss": 0.1836,
      "step": 113
    },
    {
      "epoch": 0.004609459500439718,
      "grad_norm": 2.7630367279052734,
      "learning_rate": 9.999992028654711e-05,
      "loss": 0.1005,
      "step": 114
    },
    {
      "epoch": 0.004649893355706734,
      "grad_norm": 20.46099090576172,
      "learning_rate": 9.999990849221329e-05,
      "loss": 0.4513,
      "step": 115
    },
    {
      "epoch": 0.004690327210973748,
      "grad_norm": 11.76425838470459,
      "learning_rate": 9.999989588447816e-05,
      "loss": 0.348,
      "step": 116
    },
    {
      "epoch": 0.004730761066240764,
      "grad_norm": 13.530948638916016,
      "learning_rate": 9.999988246334193e-05,
      "loss": 0.2852,
      "step": 117
    },
    {
      "epoch": 0.004771194921507778,
      "grad_norm": 4.60286808013916,
      "learning_rate": 9.999986822880483e-05,
      "loss": 0.1511,
      "step": 118
    },
    {
      "epoch": 0.004811628776774794,
      "grad_norm": 8.1397705078125,
      "learning_rate": 9.999985318086706e-05,
      "loss": 0.3053,
      "step": 119
    },
    {
      "epoch": 0.0048520626320418085,
      "grad_norm": 9.727378845214844,
      "learning_rate": 9.999983731952889e-05,
      "loss": 0.3261,
      "step": 120
    },
    {
      "epoch": 0.004892496487308824,
      "grad_norm": 6.355672359466553,
      "learning_rate": 9.999982064479057e-05,
      "loss": 0.1726,
      "step": 121
    },
    {
      "epoch": 0.004932930342575839,
      "grad_norm": 8.556386947631836,
      "learning_rate": 9.999980315665237e-05,
      "loss": 0.21,
      "step": 122
    },
    {
      "epoch": 0.004973364197842854,
      "grad_norm": 9.588984489440918,
      "learning_rate": 9.999978485511459e-05,
      "loss": 0.3401,
      "step": 123
    },
    {
      "epoch": 0.005013798053109869,
      "grad_norm": 6.82341194152832,
      "learning_rate": 9.999976574017749e-05,
      "loss": 0.2865,
      "step": 124
    },
    {
      "epoch": 0.005054231908376884,
      "grad_norm": 13.025908470153809,
      "learning_rate": 9.999974581184142e-05,
      "loss": 0.3679,
      "step": 125
    },
    {
      "epoch": 0.005094665763643899,
      "grad_norm": 7.907348155975342,
      "learning_rate": 9.999972507010669e-05,
      "loss": 0.2739,
      "step": 126
    },
    {
      "epoch": 0.0051350996189109144,
      "grad_norm": 5.3686676025390625,
      "learning_rate": 9.999970351497363e-05,
      "loss": 0.1398,
      "step": 127
    },
    {
      "epoch": 0.005175533474177929,
      "grad_norm": 6.662126064300537,
      "learning_rate": 9.99996811464426e-05,
      "loss": 0.2436,
      "step": 128
    },
    {
      "epoch": 0.005215967329444945,
      "grad_norm": 7.145336627960205,
      "learning_rate": 9.999965796451397e-05,
      "loss": 0.174,
      "step": 129
    },
    {
      "epoch": 0.005256401184711959,
      "grad_norm": 7.168648719787598,
      "learning_rate": 9.99996339691881e-05,
      "loss": 0.2184,
      "step": 130
    },
    {
      "epoch": 0.005296835039978975,
      "grad_norm": 10.925111770629883,
      "learning_rate": 9.99996091604654e-05,
      "loss": 0.3104,
      "step": 131
    },
    {
      "epoch": 0.005337268895245989,
      "grad_norm": 7.4000396728515625,
      "learning_rate": 9.999958353834624e-05,
      "loss": 0.1764,
      "step": 132
    },
    {
      "epoch": 0.005377702750513005,
      "grad_norm": 14.293291091918945,
      "learning_rate": 9.999955710283109e-05,
      "loss": 0.3948,
      "step": 133
    },
    {
      "epoch": 0.0054181366057800195,
      "grad_norm": 17.54743003845215,
      "learning_rate": 9.999952985392033e-05,
      "loss": 0.4679,
      "step": 134
    },
    {
      "epoch": 0.005458570461047035,
      "grad_norm": 10.179829597473145,
      "learning_rate": 9.999950179161442e-05,
      "loss": 0.247,
      "step": 135
    },
    {
      "epoch": 0.00549900431631405,
      "grad_norm": 8.208870887756348,
      "learning_rate": 9.999947291591383e-05,
      "loss": 0.3418,
      "step": 136
    },
    {
      "epoch": 0.005539438171581065,
      "grad_norm": 7.8983917236328125,
      "learning_rate": 9.9999443226819e-05,
      "loss": 0.3445,
      "step": 137
    },
    {
      "epoch": 0.00557987202684808,
      "grad_norm": 14.267950057983398,
      "learning_rate": 9.999941272433046e-05,
      "loss": 0.3628,
      "step": 138
    },
    {
      "epoch": 0.005620305882115095,
      "grad_norm": 11.430856704711914,
      "learning_rate": 9.999938140844866e-05,
      "loss": 0.278,
      "step": 139
    },
    {
      "epoch": 0.00566073973738211,
      "grad_norm": 8.389185905456543,
      "learning_rate": 9.999934927917414e-05,
      "loss": 0.3661,
      "step": 140
    },
    {
      "epoch": 0.0057011735926491254,
      "grad_norm": 8.984382629394531,
      "learning_rate": 9.999931633650739e-05,
      "loss": 0.3152,
      "step": 141
    },
    {
      "epoch": 0.00574160744791614,
      "grad_norm": 6.51492166519165,
      "learning_rate": 9.999928258044899e-05,
      "loss": 0.2281,
      "step": 142
    },
    {
      "epoch": 0.005782041303183156,
      "grad_norm": 7.01376485824585,
      "learning_rate": 9.999924801099946e-05,
      "loss": 0.3193,
      "step": 143
    },
    {
      "epoch": 0.00582247515845017,
      "grad_norm": 7.640584468841553,
      "learning_rate": 9.999921262815936e-05,
      "loss": 0.2331,
      "step": 144
    },
    {
      "epoch": 0.005862909013717186,
      "grad_norm": 4.045457363128662,
      "learning_rate": 9.999917643192928e-05,
      "loss": 0.1184,
      "step": 145
    },
    {
      "epoch": 0.0059033428689842,
      "grad_norm": 13.00910758972168,
      "learning_rate": 9.999913942230979e-05,
      "loss": 0.3623,
      "step": 146
    },
    {
      "epoch": 0.005943776724251216,
      "grad_norm": 7.396110534667969,
      "learning_rate": 9.999910159930151e-05,
      "loss": 0.2281,
      "step": 147
    },
    {
      "epoch": 0.0059842105795182305,
      "grad_norm": 3.814600944519043,
      "learning_rate": 9.999906296290506e-05,
      "loss": 0.1162,
      "step": 148
    },
    {
      "epoch": 0.006024644434785246,
      "grad_norm": 10.155074119567871,
      "learning_rate": 9.999902351312105e-05,
      "loss": 0.2232,
      "step": 149
    },
    {
      "epoch": 0.006065078290052261,
      "grad_norm": 7.059305667877197,
      "learning_rate": 9.999898324995013e-05,
      "loss": 0.1808,
      "step": 150
    },
    {
      "epoch": 0.006105512145319276,
      "grad_norm": 12.093023300170898,
      "learning_rate": 9.999894217339296e-05,
      "loss": 0.3253,
      "step": 151
    },
    {
      "epoch": 0.006145946000586291,
      "grad_norm": 13.166694641113281,
      "learning_rate": 9.999890028345019e-05,
      "loss": 0.3509,
      "step": 152
    },
    {
      "epoch": 0.006186379855853306,
      "grad_norm": 5.3184733390808105,
      "learning_rate": 9.999885758012253e-05,
      "loss": 0.1875,
      "step": 153
    },
    {
      "epoch": 0.006226813711120321,
      "grad_norm": 7.991506099700928,
      "learning_rate": 9.999881406341065e-05,
      "loss": 0.186,
      "step": 154
    },
    {
      "epoch": 0.006267247566387336,
      "grad_norm": 10.271965026855469,
      "learning_rate": 9.999876973331528e-05,
      "loss": 0.3393,
      "step": 155
    },
    {
      "epoch": 0.006307681421654351,
      "grad_norm": 8.440420150756836,
      "learning_rate": 9.99987245898371e-05,
      "loss": 0.3662,
      "step": 156
    },
    {
      "epoch": 0.006348115276921367,
      "grad_norm": 9.392427444458008,
      "learning_rate": 9.99986786329769e-05,
      "loss": 0.2827,
      "step": 157
    },
    {
      "epoch": 0.006388549132188381,
      "grad_norm": 11.753120422363281,
      "learning_rate": 9.999863186273539e-05,
      "loss": 0.3695,
      "step": 158
    },
    {
      "epoch": 0.006428982987455397,
      "grad_norm": 7.335087776184082,
      "learning_rate": 9.999858427911335e-05,
      "loss": 0.2826,
      "step": 159
    },
    {
      "epoch": 0.006469416842722411,
      "grad_norm": 8.617646217346191,
      "learning_rate": 9.999853588211154e-05,
      "loss": 0.2268,
      "step": 160
    },
    {
      "epoch": 0.006509850697989427,
      "grad_norm": 6.348892688751221,
      "learning_rate": 9.999848667173075e-05,
      "loss": 0.2811,
      "step": 161
    },
    {
      "epoch": 0.0065502845532564415,
      "grad_norm": 9.774231910705566,
      "learning_rate": 9.999843664797178e-05,
      "loss": 0.4024,
      "step": 162
    },
    {
      "epoch": 0.006590718408523457,
      "grad_norm": 11.390604019165039,
      "learning_rate": 9.999838581083546e-05,
      "loss": 0.2387,
      "step": 163
    },
    {
      "epoch": 0.006631152263790472,
      "grad_norm": 4.1529388427734375,
      "learning_rate": 9.99983341603226e-05,
      "loss": 0.2513,
      "step": 164
    },
    {
      "epoch": 0.006671586119057487,
      "grad_norm": 9.902484893798828,
      "learning_rate": 9.999828169643404e-05,
      "loss": 0.2148,
      "step": 165
    },
    {
      "epoch": 0.006712019974324502,
      "grad_norm": 13.555265426635742,
      "learning_rate": 9.999822841917064e-05,
      "loss": 0.3568,
      "step": 166
    },
    {
      "epoch": 0.006752453829591517,
      "grad_norm": 10.916128158569336,
      "learning_rate": 9.999817432853326e-05,
      "loss": 0.2892,
      "step": 167
    },
    {
      "epoch": 0.006792887684858532,
      "grad_norm": 5.216245651245117,
      "learning_rate": 9.999811942452279e-05,
      "loss": 0.2052,
      "step": 168
    },
    {
      "epoch": 0.006833321540125547,
      "grad_norm": 11.540072441101074,
      "learning_rate": 9.999806370714011e-05,
      "loss": 0.3585,
      "step": 169
    },
    {
      "epoch": 0.006873755395392562,
      "grad_norm": 5.5519304275512695,
      "learning_rate": 9.999800717638614e-05,
      "loss": 0.2453,
      "step": 170
    },
    {
      "epoch": 0.006914189250659578,
      "grad_norm": 7.624457359313965,
      "learning_rate": 9.999794983226179e-05,
      "loss": 0.2453,
      "step": 171
    },
    {
      "epoch": 0.006954623105926592,
      "grad_norm": 3.1242964267730713,
      "learning_rate": 9.999789167476801e-05,
      "loss": 0.2162,
      "step": 172
    },
    {
      "epoch": 0.006995056961193608,
      "grad_norm": 5.320684432983398,
      "learning_rate": 9.999783270390572e-05,
      "loss": 0.2053,
      "step": 173
    },
    {
      "epoch": 0.007035490816460622,
      "grad_norm": 9.082324028015137,
      "learning_rate": 9.999777291967589e-05,
      "loss": 0.3074,
      "step": 174
    },
    {
      "epoch": 0.007075924671727638,
      "grad_norm": 9.537432670593262,
      "learning_rate": 9.999771232207951e-05,
      "loss": 0.2791,
      "step": 175
    },
    {
      "epoch": 0.0071163585269946525,
      "grad_norm": 9.438758850097656,
      "learning_rate": 9.999765091111754e-05,
      "loss": 0.2213,
      "step": 176
    },
    {
      "epoch": 0.007156792382261668,
      "grad_norm": 6.272062301635742,
      "learning_rate": 9.999758868679099e-05,
      "loss": 0.2219,
      "step": 177
    },
    {
      "epoch": 0.007197226237528683,
      "grad_norm": 3.2677524089813232,
      "learning_rate": 9.999752564910086e-05,
      "loss": 0.2241,
      "step": 178
    },
    {
      "epoch": 0.007237660092795698,
      "grad_norm": 3.407979726791382,
      "learning_rate": 9.99974617980482e-05,
      "loss": 0.1709,
      "step": 179
    },
    {
      "epoch": 0.007278093948062713,
      "grad_norm": 7.650908946990967,
      "learning_rate": 9.999739713363404e-05,
      "loss": 0.189,
      "step": 180
    },
    {
      "epoch": 0.007318527803329728,
      "grad_norm": 5.595089912414551,
      "learning_rate": 9.999733165585943e-05,
      "loss": 0.1611,
      "step": 181
    },
    {
      "epoch": 0.007358961658596743,
      "grad_norm": 5.560061931610107,
      "learning_rate": 9.999726536472542e-05,
      "loss": 0.1824,
      "step": 182
    },
    {
      "epoch": 0.007399395513863758,
      "grad_norm": 13.8944091796875,
      "learning_rate": 9.99971982602331e-05,
      "loss": 0.3737,
      "step": 183
    },
    {
      "epoch": 0.007439829369130773,
      "grad_norm": 5.863430976867676,
      "learning_rate": 9.999713034238359e-05,
      "loss": 0.2017,
      "step": 184
    },
    {
      "epoch": 0.0074802632243977886,
      "grad_norm": 4.334754467010498,
      "learning_rate": 9.999706161117795e-05,
      "loss": 0.0885,
      "step": 185
    },
    {
      "epoch": 0.007520697079664803,
      "grad_norm": 5.766237735748291,
      "learning_rate": 9.99969920666173e-05,
      "loss": 0.254,
      "step": 186
    },
    {
      "epoch": 0.007561130934931819,
      "grad_norm": 4.142415523529053,
      "learning_rate": 9.99969217087028e-05,
      "loss": 0.1412,
      "step": 187
    },
    {
      "epoch": 0.007601564790198833,
      "grad_norm": 7.84074068069458,
      "learning_rate": 9.999685053743559e-05,
      "loss": 0.1959,
      "step": 188
    },
    {
      "epoch": 0.007641998645465849,
      "grad_norm": 8.681429862976074,
      "learning_rate": 9.999677855281682e-05,
      "loss": 0.1584,
      "step": 189
    },
    {
      "epoch": 0.0076824325007328635,
      "grad_norm": 9.750258445739746,
      "learning_rate": 9.999670575484765e-05,
      "loss": 0.2074,
      "step": 190
    },
    {
      "epoch": 0.007722866355999879,
      "grad_norm": 7.412321090698242,
      "learning_rate": 9.999663214352929e-05,
      "loss": 0.1696,
      "step": 191
    },
    {
      "epoch": 0.007763300211266894,
      "grad_norm": 9.03699016571045,
      "learning_rate": 9.999655771886291e-05,
      "loss": 0.1942,
      "step": 192
    },
    {
      "epoch": 0.007803734066533909,
      "grad_norm": 7.925232887268066,
      "learning_rate": 9.999648248084974e-05,
      "loss": 0.1793,
      "step": 193
    },
    {
      "epoch": 0.007844167921800925,
      "grad_norm": 7.363532066345215,
      "learning_rate": 9.9996406429491e-05,
      "loss": 0.1718,
      "step": 194
    },
    {
      "epoch": 0.00788460177706794,
      "grad_norm": 9.17047119140625,
      "learning_rate": 9.999632956478793e-05,
      "loss": 0.404,
      "step": 195
    },
    {
      "epoch": 0.007925035632334954,
      "grad_norm": 8.83364486694336,
      "learning_rate": 9.999625188674175e-05,
      "loss": 0.2276,
      "step": 196
    },
    {
      "epoch": 0.007965469487601969,
      "grad_norm": 9.548094749450684,
      "learning_rate": 9.999617339535378e-05,
      "loss": 0.1875,
      "step": 197
    },
    {
      "epoch": 0.008005903342868985,
      "grad_norm": 6.08480167388916,
      "learning_rate": 9.999609409062525e-05,
      "loss": 0.145,
      "step": 198
    },
    {
      "epoch": 0.008046337198136,
      "grad_norm": 15.142061233520508,
      "learning_rate": 9.999601397255747e-05,
      "loss": 0.4395,
      "step": 199
    },
    {
      "epoch": 0.008086771053403014,
      "grad_norm": 15.888526916503906,
      "learning_rate": 9.999593304115174e-05,
      "loss": 0.5784,
      "step": 200
    },
    {
      "epoch": 0.008127204908670029,
      "grad_norm": 6.388537883758545,
      "learning_rate": 9.999585129640936e-05,
      "loss": 0.2958,
      "step": 201
    },
    {
      "epoch": 0.008167638763937045,
      "grad_norm": 5.720010757446289,
      "learning_rate": 9.999576873833169e-05,
      "loss": 0.1192,
      "step": 202
    },
    {
      "epoch": 0.00820807261920406,
      "grad_norm": 7.905060291290283,
      "learning_rate": 9.999568536692006e-05,
      "loss": 0.3184,
      "step": 203
    },
    {
      "epoch": 0.008248506474471074,
      "grad_norm": 3.085916519165039,
      "learning_rate": 9.999560118217583e-05,
      "loss": 0.1954,
      "step": 204
    },
    {
      "epoch": 0.008288940329738089,
      "grad_norm": 8.547829627990723,
      "learning_rate": 9.999551618410034e-05,
      "loss": 0.2605,
      "step": 205
    },
    {
      "epoch": 0.008329374185005105,
      "grad_norm": 7.506508827209473,
      "learning_rate": 9.999543037269504e-05,
      "loss": 0.3028,
      "step": 206
    },
    {
      "epoch": 0.00836980804027212,
      "grad_norm": 4.869304656982422,
      "learning_rate": 9.999534374796124e-05,
      "loss": 0.2271,
      "step": 207
    },
    {
      "epoch": 0.008410241895539135,
      "grad_norm": 7.360259056091309,
      "learning_rate": 9.999525630990041e-05,
      "loss": 0.2761,
      "step": 208
    },
    {
      "epoch": 0.00845067575080615,
      "grad_norm": 6.078726768493652,
      "learning_rate": 9.999516805851397e-05,
      "loss": 0.2623,
      "step": 209
    },
    {
      "epoch": 0.008491109606073166,
      "grad_norm": 2.553845167160034,
      "learning_rate": 9.999507899380331e-05,
      "loss": 0.1659,
      "step": 210
    },
    {
      "epoch": 0.00853154346134018,
      "grad_norm": 3.2435362339019775,
      "learning_rate": 9.999498911576993e-05,
      "loss": 0.1498,
      "step": 211
    },
    {
      "epoch": 0.008571977316607195,
      "grad_norm": 6.379277229309082,
      "learning_rate": 9.999489842441527e-05,
      "loss": 0.2309,
      "step": 212
    },
    {
      "epoch": 0.00861241117187421,
      "grad_norm": 4.347065448760986,
      "learning_rate": 9.99948069197408e-05,
      "loss": 0.2128,
      "step": 213
    },
    {
      "epoch": 0.008652845027141226,
      "grad_norm": 9.05762767791748,
      "learning_rate": 9.999471460174803e-05,
      "loss": 0.1779,
      "step": 214
    },
    {
      "epoch": 0.00869327888240824,
      "grad_norm": 4.830628871917725,
      "learning_rate": 9.999462147043843e-05,
      "loss": 0.1574,
      "step": 215
    },
    {
      "epoch": 0.008733712737675255,
      "grad_norm": 3.8125157356262207,
      "learning_rate": 9.999452752581355e-05,
      "loss": 0.1504,
      "step": 216
    },
    {
      "epoch": 0.00877414659294227,
      "grad_norm": 2.8179168701171875,
      "learning_rate": 9.999443276787489e-05,
      "loss": 0.1777,
      "step": 217
    },
    {
      "epoch": 0.008814580448209286,
      "grad_norm": 5.864883899688721,
      "learning_rate": 9.9994337196624e-05,
      "loss": 0.3495,
      "step": 218
    },
    {
      "epoch": 0.008855014303476301,
      "grad_norm": 6.230106353759766,
      "learning_rate": 9.999424081206245e-05,
      "loss": 0.1318,
      "step": 219
    },
    {
      "epoch": 0.008895448158743316,
      "grad_norm": 6.827365875244141,
      "learning_rate": 9.999414361419178e-05,
      "loss": 0.1966,
      "step": 220
    },
    {
      "epoch": 0.00893588201401033,
      "grad_norm": 12.011444091796875,
      "learning_rate": 9.99940456030136e-05,
      "loss": 0.2844,
      "step": 221
    },
    {
      "epoch": 0.008976315869277347,
      "grad_norm": 7.509864330291748,
      "learning_rate": 9.999394677852948e-05,
      "loss": 0.3334,
      "step": 222
    },
    {
      "epoch": 0.009016749724544361,
      "grad_norm": 10.87213134765625,
      "learning_rate": 9.999384714074105e-05,
      "loss": 0.2559,
      "step": 223
    },
    {
      "epoch": 0.009057183579811376,
      "grad_norm": 7.5143327713012695,
      "learning_rate": 9.99937466896499e-05,
      "loss": 0.1708,
      "step": 224
    },
    {
      "epoch": 0.00909761743507839,
      "grad_norm": 17.510313034057617,
      "learning_rate": 9.99936454252577e-05,
      "loss": 0.4556,
      "step": 225
    },
    {
      "epoch": 0.009138051290345407,
      "grad_norm": 6.468897342681885,
      "learning_rate": 9.999354334756608e-05,
      "loss": 0.2092,
      "step": 226
    },
    {
      "epoch": 0.009178485145612422,
      "grad_norm": 6.171923637390137,
      "learning_rate": 9.99934404565767e-05,
      "loss": 0.2199,
      "step": 227
    },
    {
      "epoch": 0.009218919000879436,
      "grad_norm": 8.079005241394043,
      "learning_rate": 9.999333675229123e-05,
      "loss": 0.2471,
      "step": 228
    },
    {
      "epoch": 0.00925935285614645,
      "grad_norm": 11.559858322143555,
      "learning_rate": 9.999323223471136e-05,
      "loss": 0.3881,
      "step": 229
    },
    {
      "epoch": 0.009299786711413467,
      "grad_norm": 16.177331924438477,
      "learning_rate": 9.999312690383881e-05,
      "loss": 0.4198,
      "step": 230
    },
    {
      "epoch": 0.009340220566680482,
      "grad_norm": 16.129283905029297,
      "learning_rate": 9.999302075967526e-05,
      "loss": 0.3592,
      "step": 231
    },
    {
      "epoch": 0.009380654421947496,
      "grad_norm": 13.0033597946167,
      "learning_rate": 9.999291380222246e-05,
      "loss": 0.3194,
      "step": 232
    },
    {
      "epoch": 0.009421088277214511,
      "grad_norm": 5.525213718414307,
      "learning_rate": 9.999280603148215e-05,
      "loss": 0.1632,
      "step": 233
    },
    {
      "epoch": 0.009461522132481527,
      "grad_norm": 7.511592388153076,
      "learning_rate": 9.999269744745606e-05,
      "loss": 0.2767,
      "step": 234
    },
    {
      "epoch": 0.009501955987748542,
      "grad_norm": 4.852881908416748,
      "learning_rate": 9.999258805014599e-05,
      "loss": 0.1132,
      "step": 235
    },
    {
      "epoch": 0.009542389843015557,
      "grad_norm": 5.765609264373779,
      "learning_rate": 9.999247783955369e-05,
      "loss": 0.1631,
      "step": 236
    },
    {
      "epoch": 0.009582823698282571,
      "grad_norm": 9.90166187286377,
      "learning_rate": 9.999236681568097e-05,
      "loss": 0.2498,
      "step": 237
    },
    {
      "epoch": 0.009623257553549588,
      "grad_norm": 7.612752914428711,
      "learning_rate": 9.999225497852962e-05,
      "loss": 0.2766,
      "step": 238
    },
    {
      "epoch": 0.009663691408816602,
      "grad_norm": 11.317419052124023,
      "learning_rate": 9.99921423281015e-05,
      "loss": 0.2912,
      "step": 239
    },
    {
      "epoch": 0.009704125264083617,
      "grad_norm": 6.9325079917907715,
      "learning_rate": 9.999202886439838e-05,
      "loss": 0.1434,
      "step": 240
    },
    {
      "epoch": 0.009744559119350632,
      "grad_norm": 9.33085823059082,
      "learning_rate": 9.999191458742217e-05,
      "loss": 0.2937,
      "step": 241
    },
    {
      "epoch": 0.009784992974617648,
      "grad_norm": 9.718541145324707,
      "learning_rate": 9.999179949717468e-05,
      "loss": 0.2479,
      "step": 242
    },
    {
      "epoch": 0.009825426829884663,
      "grad_norm": 8.457967758178711,
      "learning_rate": 9.99916835936578e-05,
      "loss": 0.1859,
      "step": 243
    },
    {
      "epoch": 0.009865860685151677,
      "grad_norm": 5.0217604637146,
      "learning_rate": 9.999156687687342e-05,
      "loss": 0.2402,
      "step": 244
    },
    {
      "epoch": 0.009906294540418692,
      "grad_norm": 6.611813545227051,
      "learning_rate": 9.999144934682343e-05,
      "loss": 0.1342,
      "step": 245
    },
    {
      "epoch": 0.009946728395685708,
      "grad_norm": 5.274916648864746,
      "learning_rate": 9.999133100350974e-05,
      "loss": 0.1561,
      "step": 246
    },
    {
      "epoch": 0.009987162250952723,
      "grad_norm": 13.173295974731445,
      "learning_rate": 9.99912118469343e-05,
      "loss": 0.3176,
      "step": 247
    },
    {
      "epoch": 0.010027596106219738,
      "grad_norm": 6.314767360687256,
      "learning_rate": 9.999109187709903e-05,
      "loss": 0.2688,
      "step": 248
    },
    {
      "epoch": 0.010068029961486752,
      "grad_norm": 10.67939281463623,
      "learning_rate": 9.999097109400586e-05,
      "loss": 0.2501,
      "step": 249
    },
    {
      "epoch": 0.010108463816753769,
      "grad_norm": 3.957179546356201,
      "learning_rate": 9.99908494976568e-05,
      "loss": 0.1173,
      "step": 250
    },
    {
      "epoch": 0.010148897672020783,
      "grad_norm": 3.20841383934021,
      "learning_rate": 9.999072708805381e-05,
      "loss": 0.2179,
      "step": 251
    },
    {
      "epoch": 0.010189331527287798,
      "grad_norm": 4.517293930053711,
      "learning_rate": 9.999060386519886e-05,
      "loss": 0.1345,
      "step": 252
    },
    {
      "epoch": 0.010229765382554813,
      "grad_norm": 5.086652755737305,
      "learning_rate": 9.999047982909398e-05,
      "loss": 0.4833,
      "step": 253
    },
    {
      "epoch": 0.010270199237821829,
      "grad_norm": 5.176082134246826,
      "learning_rate": 9.999035497974117e-05,
      "loss": 0.2265,
      "step": 254
    },
    {
      "epoch": 0.010310633093088844,
      "grad_norm": 8.907313346862793,
      "learning_rate": 9.999022931714249e-05,
      "loss": 0.1756,
      "step": 255
    },
    {
      "epoch": 0.010351066948355858,
      "grad_norm": 5.966263771057129,
      "learning_rate": 9.999010284129994e-05,
      "loss": 0.1876,
      "step": 256
    },
    {
      "epoch": 0.010391500803622873,
      "grad_norm": 13.120492935180664,
      "learning_rate": 9.998997555221561e-05,
      "loss": 0.3462,
      "step": 257
    },
    {
      "epoch": 0.01043193465888989,
      "grad_norm": 5.3854875564575195,
      "learning_rate": 9.998984744989157e-05,
      "loss": 0.1494,
      "step": 258
    },
    {
      "epoch": 0.010472368514156904,
      "grad_norm": 3.7474594116210938,
      "learning_rate": 9.998971853432989e-05,
      "loss": 0.1319,
      "step": 259
    },
    {
      "epoch": 0.010512802369423918,
      "grad_norm": 4.1866068840026855,
      "learning_rate": 9.998958880553267e-05,
      "loss": 0.1808,
      "step": 260
    },
    {
      "epoch": 0.010553236224690933,
      "grad_norm": 5.2659149169921875,
      "learning_rate": 9.998945826350204e-05,
      "loss": 0.201,
      "step": 261
    },
    {
      "epoch": 0.01059367007995795,
      "grad_norm": 3.337062358856201,
      "learning_rate": 9.99893269082401e-05,
      "loss": 0.1585,
      "step": 262
    },
    {
      "epoch": 0.010634103935224964,
      "grad_norm": 11.464987754821777,
      "learning_rate": 9.998919473974899e-05,
      "loss": 0.2708,
      "step": 263
    },
    {
      "epoch": 0.010674537790491979,
      "grad_norm": 5.518500328063965,
      "learning_rate": 9.998906175803087e-05,
      "loss": 0.127,
      "step": 264
    },
    {
      "epoch": 0.010714971645758993,
      "grad_norm": 4.601476669311523,
      "learning_rate": 9.99889279630879e-05,
      "loss": 0.1266,
      "step": 265
    },
    {
      "epoch": 0.01075540550102601,
      "grad_norm": 4.7640838623046875,
      "learning_rate": 9.998879335492224e-05,
      "loss": 0.2607,
      "step": 266
    },
    {
      "epoch": 0.010795839356293024,
      "grad_norm": 8.744903564453125,
      "learning_rate": 9.998865793353611e-05,
      "loss": 0.1935,
      "step": 267
    },
    {
      "epoch": 0.010836273211560039,
      "grad_norm": 8.256892204284668,
      "learning_rate": 9.998852169893169e-05,
      "loss": 0.3057,
      "step": 268
    },
    {
      "epoch": 0.010876707066827054,
      "grad_norm": 4.505156517028809,
      "learning_rate": 9.998838465111121e-05,
      "loss": 0.2261,
      "step": 269
    },
    {
      "epoch": 0.01091714092209407,
      "grad_norm": 5.429905891418457,
      "learning_rate": 9.99882467900769e-05,
      "loss": 0.1722,
      "step": 270
    },
    {
      "epoch": 0.010957574777361085,
      "grad_norm": 3.315159797668457,
      "learning_rate": 9.998810811583098e-05,
      "loss": 0.3201,
      "step": 271
    },
    {
      "epoch": 0.0109980086326281,
      "grad_norm": 2.6552138328552246,
      "learning_rate": 9.998796862837574e-05,
      "loss": 0.1944,
      "step": 272
    },
    {
      "epoch": 0.011038442487895114,
      "grad_norm": 7.77968168258667,
      "learning_rate": 9.99878283277134e-05,
      "loss": 0.2229,
      "step": 273
    },
    {
      "epoch": 0.01107887634316213,
      "grad_norm": 4.262834548950195,
      "learning_rate": 9.99876872138463e-05,
      "loss": 0.2067,
      "step": 274
    },
    {
      "epoch": 0.011119310198429145,
      "grad_norm": 5.2016377449035645,
      "learning_rate": 9.998754528677671e-05,
      "loss": 0.4344,
      "step": 275
    },
    {
      "epoch": 0.01115974405369616,
      "grad_norm": 3.6373684406280518,
      "learning_rate": 9.998740254650694e-05,
      "loss": 0.1344,
      "step": 276
    },
    {
      "epoch": 0.011200177908963174,
      "grad_norm": 8.957545280456543,
      "learning_rate": 9.99872589930393e-05,
      "loss": 0.3547,
      "step": 277
    },
    {
      "epoch": 0.01124061176423019,
      "grad_norm": 11.82962417602539,
      "learning_rate": 9.998711462637614e-05,
      "loss": 0.3589,
      "step": 278
    },
    {
      "epoch": 0.011281045619497205,
      "grad_norm": 5.094110488891602,
      "learning_rate": 9.99869694465198e-05,
      "loss": 0.277,
      "step": 279
    },
    {
      "epoch": 0.01132147947476422,
      "grad_norm": 4.103588104248047,
      "learning_rate": 9.998682345347266e-05,
      "loss": 0.2051,
      "step": 280
    },
    {
      "epoch": 0.011361913330031235,
      "grad_norm": 11.468871116638184,
      "learning_rate": 9.998667664723706e-05,
      "loss": 0.2634,
      "step": 281
    },
    {
      "epoch": 0.011402347185298251,
      "grad_norm": 4.088072776794434,
      "learning_rate": 9.998652902781543e-05,
      "loss": 0.1833,
      "step": 282
    },
    {
      "epoch": 0.011442781040565266,
      "grad_norm": 6.32438850402832,
      "learning_rate": 9.998638059521015e-05,
      "loss": 0.2299,
      "step": 283
    },
    {
      "epoch": 0.01148321489583228,
      "grad_norm": 7.150467872619629,
      "learning_rate": 9.998623134942361e-05,
      "loss": 0.3572,
      "step": 284
    },
    {
      "epoch": 0.011523648751099295,
      "grad_norm": 4.513134956359863,
      "learning_rate": 9.99860812904583e-05,
      "loss": 0.1917,
      "step": 285
    },
    {
      "epoch": 0.011564082606366311,
      "grad_norm": 5.198740005493164,
      "learning_rate": 9.99859304183166e-05,
      "loss": 0.2734,
      "step": 286
    },
    {
      "epoch": 0.011604516461633326,
      "grad_norm": 7.074803829193115,
      "learning_rate": 9.998577873300099e-05,
      "loss": 0.2,
      "step": 287
    },
    {
      "epoch": 0.01164495031690034,
      "grad_norm": 8.8662109375,
      "learning_rate": 9.998562623451395e-05,
      "loss": 0.2738,
      "step": 288
    },
    {
      "epoch": 0.011685384172167355,
      "grad_norm": 3.111583709716797,
      "learning_rate": 9.998547292285794e-05,
      "loss": 0.1513,
      "step": 289
    },
    {
      "epoch": 0.011725818027434371,
      "grad_norm": 5.535727024078369,
      "learning_rate": 9.998531879803545e-05,
      "loss": 0.1386,
      "step": 290
    },
    {
      "epoch": 0.011766251882701386,
      "grad_norm": 3.3085310459136963,
      "learning_rate": 9.998516386004901e-05,
      "loss": 0.2131,
      "step": 291
    },
    {
      "epoch": 0.0118066857379684,
      "grad_norm": 6.310663223266602,
      "learning_rate": 9.998500810890115e-05,
      "loss": 0.1733,
      "step": 292
    },
    {
      "epoch": 0.011847119593235415,
      "grad_norm": 8.722447395324707,
      "learning_rate": 9.998485154459436e-05,
      "loss": 0.3804,
      "step": 293
    },
    {
      "epoch": 0.011887553448502432,
      "grad_norm": 13.239892959594727,
      "learning_rate": 9.99846941671312e-05,
      "loss": 0.3978,
      "step": 294
    },
    {
      "epoch": 0.011927987303769446,
      "grad_norm": 9.516282081604004,
      "learning_rate": 9.998453597651427e-05,
      "loss": 0.2154,
      "step": 295
    },
    {
      "epoch": 0.011968421159036461,
      "grad_norm": 6.891446590423584,
      "learning_rate": 9.998437697274609e-05,
      "loss": 0.2056,
      "step": 296
    },
    {
      "epoch": 0.012008855014303476,
      "grad_norm": 6.211913585662842,
      "learning_rate": 9.998421715582929e-05,
      "loss": 0.2311,
      "step": 297
    },
    {
      "epoch": 0.012049288869570492,
      "grad_norm": 10.808809280395508,
      "learning_rate": 9.998405652576643e-05,
      "loss": 0.1792,
      "step": 298
    },
    {
      "epoch": 0.012089722724837507,
      "grad_norm": 14.211930274963379,
      "learning_rate": 9.998389508256015e-05,
      "loss": 0.3934,
      "step": 299
    },
    {
      "epoch": 0.012130156580104521,
      "grad_norm": 10.106294631958008,
      "learning_rate": 9.998373282621308e-05,
      "loss": 0.2446,
      "step": 300
    },
    {
      "epoch": 0.012170590435371536,
      "grad_norm": 17.58440589904785,
      "learning_rate": 9.998356975672785e-05,
      "loss": 0.513,
      "step": 301
    },
    {
      "epoch": 0.012211024290638552,
      "grad_norm": 9.144328117370605,
      "learning_rate": 9.99834058741071e-05,
      "loss": 0.3244,
      "step": 302
    },
    {
      "epoch": 0.012251458145905567,
      "grad_norm": 10.438602447509766,
      "learning_rate": 9.998324117835351e-05,
      "loss": 0.2299,
      "step": 303
    },
    {
      "epoch": 0.012291892001172582,
      "grad_norm": 6.329174518585205,
      "learning_rate": 9.998307566946976e-05,
      "loss": 0.175,
      "step": 304
    },
    {
      "epoch": 0.012332325856439596,
      "grad_norm": 8.735311508178711,
      "learning_rate": 9.998290934745856e-05,
      "loss": 0.2045,
      "step": 305
    },
    {
      "epoch": 0.012372759711706613,
      "grad_norm": 7.5931854248046875,
      "learning_rate": 9.998274221232257e-05,
      "loss": 0.2463,
      "step": 306
    },
    {
      "epoch": 0.012413193566973627,
      "grad_norm": 8.052494049072266,
      "learning_rate": 9.998257426406454e-05,
      "loss": 0.1967,
      "step": 307
    },
    {
      "epoch": 0.012453627422240642,
      "grad_norm": 4.575268268585205,
      "learning_rate": 9.998240550268721e-05,
      "loss": 0.1805,
      "step": 308
    },
    {
      "epoch": 0.012494061277507656,
      "grad_norm": 4.046161651611328,
      "learning_rate": 9.99822359281933e-05,
      "loss": 0.1365,
      "step": 309
    },
    {
      "epoch": 0.012534495132774673,
      "grad_norm": 7.623053550720215,
      "learning_rate": 9.998206554058558e-05,
      "loss": 0.2339,
      "step": 310
    },
    {
      "epoch": 0.012574928988041688,
      "grad_norm": 5.255005836486816,
      "learning_rate": 9.998189433986683e-05,
      "loss": 0.146,
      "step": 311
    },
    {
      "epoch": 0.012615362843308702,
      "grad_norm": 6.908909320831299,
      "learning_rate": 9.998172232603983e-05,
      "loss": 0.2178,
      "step": 312
    },
    {
      "epoch": 0.012655796698575717,
      "grad_norm": 8.706125259399414,
      "learning_rate": 9.998154949910738e-05,
      "loss": 0.1743,
      "step": 313
    },
    {
      "epoch": 0.012696230553842733,
      "grad_norm": 8.078751564025879,
      "learning_rate": 9.998137585907228e-05,
      "loss": 0.2609,
      "step": 314
    },
    {
      "epoch": 0.012736664409109748,
      "grad_norm": 10.175312042236328,
      "learning_rate": 9.998120140593737e-05,
      "loss": 0.2437,
      "step": 315
    },
    {
      "epoch": 0.012777098264376762,
      "grad_norm": 4.451848983764648,
      "learning_rate": 9.998102613970548e-05,
      "loss": 0.1976,
      "step": 316
    },
    {
      "epoch": 0.012817532119643777,
      "grad_norm": 4.3244853019714355,
      "learning_rate": 9.998085006037946e-05,
      "loss": 0.2454,
      "step": 317
    },
    {
      "epoch": 0.012857965974910793,
      "grad_norm": 7.642861366271973,
      "learning_rate": 9.998067316796216e-05,
      "loss": 0.1998,
      "step": 318
    },
    {
      "epoch": 0.012898399830177808,
      "grad_norm": 12.059164047241211,
      "learning_rate": 9.998049546245649e-05,
      "loss": 0.2216,
      "step": 319
    },
    {
      "epoch": 0.012938833685444823,
      "grad_norm": 10.211267471313477,
      "learning_rate": 9.998031694386533e-05,
      "loss": 0.2314,
      "step": 320
    },
    {
      "epoch": 0.012979267540711837,
      "grad_norm": 10.65395450592041,
      "learning_rate": 9.998013761219157e-05,
      "loss": 0.2533,
      "step": 321
    },
    {
      "epoch": 0.013019701395978854,
      "grad_norm": 14.818538665771484,
      "learning_rate": 9.997995746743815e-05,
      "loss": 0.3403,
      "step": 322
    },
    {
      "epoch": 0.013060135251245868,
      "grad_norm": 7.847441673278809,
      "learning_rate": 9.997977650960797e-05,
      "loss": 0.295,
      "step": 323
    },
    {
      "epoch": 0.013100569106512883,
      "grad_norm": 11.467867851257324,
      "learning_rate": 9.997959473870399e-05,
      "loss": 0.2742,
      "step": 324
    },
    {
      "epoch": 0.013141002961779898,
      "grad_norm": 3.636414051055908,
      "learning_rate": 9.99794121547292e-05,
      "loss": 0.1485,
      "step": 325
    },
    {
      "epoch": 0.013181436817046914,
      "grad_norm": 10.122954368591309,
      "learning_rate": 9.99792287576865e-05,
      "loss": 0.2283,
      "step": 326
    },
    {
      "epoch": 0.013221870672313929,
      "grad_norm": 7.545888900756836,
      "learning_rate": 9.997904454757894e-05,
      "loss": 0.22,
      "step": 327
    },
    {
      "epoch": 0.013262304527580943,
      "grad_norm": 9.401891708374023,
      "learning_rate": 9.997885952440948e-05,
      "loss": 0.2872,
      "step": 328
    },
    {
      "epoch": 0.013302738382847958,
      "grad_norm": 8.097481727600098,
      "learning_rate": 9.997867368818114e-05,
      "loss": 0.1602,
      "step": 329
    },
    {
      "epoch": 0.013343172238114974,
      "grad_norm": 12.194989204406738,
      "learning_rate": 9.997848703889693e-05,
      "loss": 0.3441,
      "step": 330
    },
    {
      "epoch": 0.013383606093381989,
      "grad_norm": 5.003354072570801,
      "learning_rate": 9.997829957655991e-05,
      "loss": 0.1598,
      "step": 331
    },
    {
      "epoch": 0.013424039948649004,
      "grad_norm": 7.289619445800781,
      "learning_rate": 9.997811130117312e-05,
      "loss": 0.2624,
      "step": 332
    },
    {
      "epoch": 0.013464473803916018,
      "grad_norm": 6.675605297088623,
      "learning_rate": 9.99779222127396e-05,
      "loss": 0.2119,
      "step": 333
    },
    {
      "epoch": 0.013504907659183035,
      "grad_norm": 3.8371706008911133,
      "learning_rate": 9.997773231126249e-05,
      "loss": 0.1769,
      "step": 334
    },
    {
      "epoch": 0.01354534151445005,
      "grad_norm": 5.879626274108887,
      "learning_rate": 9.99775415967448e-05,
      "loss": 0.1408,
      "step": 335
    },
    {
      "epoch": 0.013585775369717064,
      "grad_norm": 5.4264421463012695,
      "learning_rate": 9.997735006918967e-05,
      "loss": 0.2157,
      "step": 336
    },
    {
      "epoch": 0.013626209224984078,
      "grad_norm": 10.332036018371582,
      "learning_rate": 9.997715772860024e-05,
      "loss": 0.2699,
      "step": 337
    },
    {
      "epoch": 0.013666643080251095,
      "grad_norm": 3.553473472595215,
      "learning_rate": 9.997696457497957e-05,
      "loss": 0.198,
      "step": 338
    },
    {
      "epoch": 0.01370707693551811,
      "grad_norm": 8.88234806060791,
      "learning_rate": 9.99767706083309e-05,
      "loss": 0.1843,
      "step": 339
    },
    {
      "epoch": 0.013747510790785124,
      "grad_norm": 4.623575687408447,
      "learning_rate": 9.997657582865729e-05,
      "loss": 0.1851,
      "step": 340
    },
    {
      "epoch": 0.013787944646052139,
      "grad_norm": 9.278861045837402,
      "learning_rate": 9.997638023596197e-05,
      "loss": 0.2751,
      "step": 341
    },
    {
      "epoch": 0.013828378501319155,
      "grad_norm": 7.077452659606934,
      "learning_rate": 9.99761838302481e-05,
      "loss": 0.2331,
      "step": 342
    },
    {
      "epoch": 0.01386881235658617,
      "grad_norm": 13.274353981018066,
      "learning_rate": 9.997598661151887e-05,
      "loss": 0.3568,
      "step": 343
    },
    {
      "epoch": 0.013909246211853184,
      "grad_norm": 9.196744918823242,
      "learning_rate": 9.99757885797775e-05,
      "loss": 0.2367,
      "step": 344
    },
    {
      "epoch": 0.013949680067120199,
      "grad_norm": 3.6745989322662354,
      "learning_rate": 9.997558973502721e-05,
      "loss": 0.2014,
      "step": 345
    },
    {
      "epoch": 0.013990113922387215,
      "grad_norm": 5.36557674407959,
      "learning_rate": 9.997539007727124e-05,
      "loss": 0.1432,
      "step": 346
    },
    {
      "epoch": 0.01403054777765423,
      "grad_norm": 9.858529090881348,
      "learning_rate": 9.997518960651283e-05,
      "loss": 0.25,
      "step": 347
    },
    {
      "epoch": 0.014070981632921245,
      "grad_norm": 6.524956703186035,
      "learning_rate": 9.997498832275521e-05,
      "loss": 0.256,
      "step": 348
    },
    {
      "epoch": 0.01411141548818826,
      "grad_norm": 10.811662673950195,
      "learning_rate": 9.997478622600172e-05,
      "loss": 0.1743,
      "step": 349
    },
    {
      "epoch": 0.014151849343455276,
      "grad_norm": 3.1745381355285645,
      "learning_rate": 9.99745833162556e-05,
      "loss": 0.1929,
      "step": 350
    },
    {
      "epoch": 0.01419228319872229,
      "grad_norm": 19.262676239013672,
      "learning_rate": 9.997437959352017e-05,
      "loss": 0.5264,
      "step": 351
    },
    {
      "epoch": 0.014232717053989305,
      "grad_norm": 15.925512313842773,
      "learning_rate": 9.997417505779872e-05,
      "loss": 0.4202,
      "step": 352
    },
    {
      "epoch": 0.01427315090925632,
      "grad_norm": 11.937752723693848,
      "learning_rate": 9.997396970909461e-05,
      "loss": 0.29,
      "step": 353
    },
    {
      "epoch": 0.014313584764523336,
      "grad_norm": 3.941826581954956,
      "learning_rate": 9.997376354741117e-05,
      "loss": 0.2625,
      "step": 354
    },
    {
      "epoch": 0.01435401861979035,
      "grad_norm": 4.599327564239502,
      "learning_rate": 9.997355657275174e-05,
      "loss": 0.1877,
      "step": 355
    },
    {
      "epoch": 0.014394452475057365,
      "grad_norm": 5.83558988571167,
      "learning_rate": 9.997334878511971e-05,
      "loss": 0.1983,
      "step": 356
    },
    {
      "epoch": 0.01443488633032438,
      "grad_norm": 10.315009117126465,
      "learning_rate": 9.997314018451844e-05,
      "loss": 0.2074,
      "step": 357
    },
    {
      "epoch": 0.014475320185591396,
      "grad_norm": 6.831982612609863,
      "learning_rate": 9.997293077095133e-05,
      "loss": 0.2385,
      "step": 358
    },
    {
      "epoch": 0.014515754040858411,
      "grad_norm": 7.1724958419799805,
      "learning_rate": 9.997272054442179e-05,
      "loss": 0.2749,
      "step": 359
    },
    {
      "epoch": 0.014556187896125426,
      "grad_norm": 2.9375522136688232,
      "learning_rate": 9.997250950493324e-05,
      "loss": 0.2275,
      "step": 360
    },
    {
      "epoch": 0.01459662175139244,
      "grad_norm": 7.610367774963379,
      "learning_rate": 9.997229765248912e-05,
      "loss": 0.3368,
      "step": 361
    },
    {
      "epoch": 0.014637055606659457,
      "grad_norm": 7.903773307800293,
      "learning_rate": 9.997208498709285e-05,
      "loss": 0.2111,
      "step": 362
    },
    {
      "epoch": 0.014677489461926471,
      "grad_norm": 5.814249038696289,
      "learning_rate": 9.997187150874792e-05,
      "loss": 0.1775,
      "step": 363
    },
    {
      "epoch": 0.014717923317193486,
      "grad_norm": 5.684579849243164,
      "learning_rate": 9.997165721745778e-05,
      "loss": 0.286,
      "step": 364
    },
    {
      "epoch": 0.0147583571724605,
      "grad_norm": 3.0314488410949707,
      "learning_rate": 9.997144211322595e-05,
      "loss": 0.1871,
      "step": 365
    },
    {
      "epoch": 0.014798791027727517,
      "grad_norm": 8.123518943786621,
      "learning_rate": 9.997122619605587e-05,
      "loss": 0.1906,
      "step": 366
    },
    {
      "epoch": 0.014839224882994531,
      "grad_norm": 8.576919555664062,
      "learning_rate": 9.997100946595112e-05,
      "loss": 0.2946,
      "step": 367
    },
    {
      "epoch": 0.014879658738261546,
      "grad_norm": 10.813838005065918,
      "learning_rate": 9.997079192291519e-05,
      "loss": 0.2275,
      "step": 368
    },
    {
      "epoch": 0.01492009259352856,
      "grad_norm": 11.815543174743652,
      "learning_rate": 9.99705735669516e-05,
      "loss": 0.3674,
      "step": 369
    },
    {
      "epoch": 0.014960526448795577,
      "grad_norm": 3.05165958404541,
      "learning_rate": 9.997035439806395e-05,
      "loss": 0.1633,
      "step": 370
    },
    {
      "epoch": 0.015000960304062592,
      "grad_norm": 2.033345937728882,
      "learning_rate": 9.997013441625576e-05,
      "loss": 0.1396,
      "step": 371
    },
    {
      "epoch": 0.015041394159329606,
      "grad_norm": 6.442851543426514,
      "learning_rate": 9.996991362153065e-05,
      "loss": 0.26,
      "step": 372
    },
    {
      "epoch": 0.015081828014596621,
      "grad_norm": 8.250064849853516,
      "learning_rate": 9.996969201389217e-05,
      "loss": 0.2268,
      "step": 373
    },
    {
      "epoch": 0.015122261869863637,
      "grad_norm": 12.65408706665039,
      "learning_rate": 9.996946959334396e-05,
      "loss": 0.3643,
      "step": 374
    },
    {
      "epoch": 0.015162695725130652,
      "grad_norm": 7.137821197509766,
      "learning_rate": 9.996924635988962e-05,
      "loss": 0.1717,
      "step": 375
    },
    {
      "epoch": 0.015203129580397667,
      "grad_norm": 4.720224380493164,
      "learning_rate": 9.99690223135328e-05,
      "loss": 0.1236,
      "step": 376
    },
    {
      "epoch": 0.015243563435664681,
      "grad_norm": 6.921150207519531,
      "learning_rate": 9.996879745427712e-05,
      "loss": 0.1666,
      "step": 377
    },
    {
      "epoch": 0.015283997290931698,
      "grad_norm": 12.78406810760498,
      "learning_rate": 9.996857178212625e-05,
      "loss": 0.3466,
      "step": 378
    },
    {
      "epoch": 0.015324431146198712,
      "grad_norm": 7.997376441955566,
      "learning_rate": 9.996834529708386e-05,
      "loss": 0.2124,
      "step": 379
    },
    {
      "epoch": 0.015364865001465727,
      "grad_norm": 5.641388416290283,
      "learning_rate": 9.996811799915362e-05,
      "loss": 0.3777,
      "step": 380
    },
    {
      "epoch": 0.015405298856732742,
      "grad_norm": 3.243623971939087,
      "learning_rate": 9.996788988833927e-05,
      "loss": 0.1551,
      "step": 381
    },
    {
      "epoch": 0.015445732711999758,
      "grad_norm": 11.435097694396973,
      "learning_rate": 9.996766096464447e-05,
      "loss": 0.2459,
      "step": 382
    },
    {
      "epoch": 0.015486166567266773,
      "grad_norm": 11.118339538574219,
      "learning_rate": 9.996743122807299e-05,
      "loss": 0.2865,
      "step": 383
    },
    {
      "epoch": 0.015526600422533787,
      "grad_norm": 11.727265357971191,
      "learning_rate": 9.996720067862853e-05,
      "loss": 0.2497,
      "step": 384
    },
    {
      "epoch": 0.015567034277800802,
      "grad_norm": 6.7753214836120605,
      "learning_rate": 9.996696931631486e-05,
      "loss": 0.2412,
      "step": 385
    },
    {
      "epoch": 0.015607468133067818,
      "grad_norm": 8.402801513671875,
      "learning_rate": 9.996673714113574e-05,
      "loss": 0.2423,
      "step": 386
    },
    {
      "epoch": 0.015647901988334833,
      "grad_norm": 11.142617225646973,
      "learning_rate": 9.996650415309495e-05,
      "loss": 0.3352,
      "step": 387
    },
    {
      "epoch": 0.01568833584360185,
      "grad_norm": 6.23745584487915,
      "learning_rate": 9.996627035219626e-05,
      "loss": 0.2158,
      "step": 388
    },
    {
      "epoch": 0.015728769698868862,
      "grad_norm": 5.107501983642578,
      "learning_rate": 9.99660357384435e-05,
      "loss": 0.2358,
      "step": 389
    },
    {
      "epoch": 0.01576920355413588,
      "grad_norm": 5.581637382507324,
      "learning_rate": 9.996580031184049e-05,
      "loss": 0.1375,
      "step": 390
    },
    {
      "epoch": 0.01580963740940289,
      "grad_norm": 4.091251373291016,
      "learning_rate": 9.996556407239103e-05,
      "loss": 0.1709,
      "step": 391
    },
    {
      "epoch": 0.015850071264669908,
      "grad_norm": 2.343575954437256,
      "learning_rate": 9.996532702009897e-05,
      "loss": 0.1784,
      "step": 392
    },
    {
      "epoch": 0.015890505119936924,
      "grad_norm": 4.080927848815918,
      "learning_rate": 9.996508915496821e-05,
      "loss": 0.1768,
      "step": 393
    },
    {
      "epoch": 0.015930938975203937,
      "grad_norm": 4.1473588943481445,
      "learning_rate": 9.996485047700256e-05,
      "loss": 0.1471,
      "step": 394
    },
    {
      "epoch": 0.015971372830470953,
      "grad_norm": 5.215227127075195,
      "learning_rate": 9.996461098620594e-05,
      "loss": 0.2694,
      "step": 395
    },
    {
      "epoch": 0.01601180668573797,
      "grad_norm": 6.0926055908203125,
      "learning_rate": 9.996437068258222e-05,
      "loss": 0.1453,
      "step": 396
    },
    {
      "epoch": 0.016052240541004983,
      "grad_norm": 4.204623222351074,
      "learning_rate": 9.996412956613535e-05,
      "loss": 0.1038,
      "step": 397
    },
    {
      "epoch": 0.016092674396272,
      "grad_norm": 5.5624823570251465,
      "learning_rate": 9.996388763686922e-05,
      "loss": 0.0993,
      "step": 398
    },
    {
      "epoch": 0.016133108251539012,
      "grad_norm": 8.127304077148438,
      "learning_rate": 9.996364489478775e-05,
      "loss": 0.1843,
      "step": 399
    },
    {
      "epoch": 0.01617354210680603,
      "grad_norm": 7.84940242767334,
      "learning_rate": 9.996340133989493e-05,
      "loss": 0.3037,
      "step": 400
    },
    {
      "epoch": 0.016213975962073045,
      "grad_norm": 12.833932876586914,
      "learning_rate": 9.99631569721947e-05,
      "loss": 0.3487,
      "step": 401
    },
    {
      "epoch": 0.016254409817340058,
      "grad_norm": 3.705427408218384,
      "learning_rate": 9.996291179169103e-05,
      "loss": 0.1832,
      "step": 402
    },
    {
      "epoch": 0.016294843672607074,
      "grad_norm": 8.266316413879395,
      "learning_rate": 9.996266579838793e-05,
      "loss": 0.1822,
      "step": 403
    },
    {
      "epoch": 0.01633527752787409,
      "grad_norm": 11.904566764831543,
      "learning_rate": 9.99624189922894e-05,
      "loss": 0.2984,
      "step": 404
    },
    {
      "epoch": 0.016375711383141103,
      "grad_norm": 8.782744407653809,
      "learning_rate": 9.996217137339943e-05,
      "loss": 0.2468,
      "step": 405
    },
    {
      "epoch": 0.01641614523840812,
      "grad_norm": 12.645235061645508,
      "learning_rate": 9.996192294172206e-05,
      "loss": 0.3629,
      "step": 406
    },
    {
      "epoch": 0.016456579093675133,
      "grad_norm": 7.794836044311523,
      "learning_rate": 9.996167369726132e-05,
      "loss": 0.2789,
      "step": 407
    },
    {
      "epoch": 0.01649701294894215,
      "grad_norm": 6.984973430633545,
      "learning_rate": 9.996142364002131e-05,
      "loss": 0.2155,
      "step": 408
    },
    {
      "epoch": 0.016537446804209165,
      "grad_norm": 7.008528709411621,
      "learning_rate": 9.996117277000605e-05,
      "loss": 0.1648,
      "step": 409
    },
    {
      "epoch": 0.016577880659476178,
      "grad_norm": 5.965149879455566,
      "learning_rate": 9.996092108721964e-05,
      "loss": 0.1826,
      "step": 410
    },
    {
      "epoch": 0.016618314514743195,
      "grad_norm": 5.976689338684082,
      "learning_rate": 9.996066859166617e-05,
      "loss": 0.2052,
      "step": 411
    },
    {
      "epoch": 0.01665874837001021,
      "grad_norm": 9.623112678527832,
      "learning_rate": 9.996041528334977e-05,
      "loss": 0.2493,
      "step": 412
    },
    {
      "epoch": 0.016699182225277224,
      "grad_norm": 5.069690227508545,
      "learning_rate": 9.996016116227451e-05,
      "loss": 0.1675,
      "step": 413
    },
    {
      "epoch": 0.01673961608054424,
      "grad_norm": 4.831304550170898,
      "learning_rate": 9.995990622844458e-05,
      "loss": 0.1127,
      "step": 414
    },
    {
      "epoch": 0.016780049935811253,
      "grad_norm": 4.990175724029541,
      "learning_rate": 9.995965048186409e-05,
      "loss": 0.2068,
      "step": 415
    },
    {
      "epoch": 0.01682048379107827,
      "grad_norm": 3.1760876178741455,
      "learning_rate": 9.99593939225372e-05,
      "loss": 0.1334,
      "step": 416
    },
    {
      "epoch": 0.016860917646345286,
      "grad_norm": 7.774895668029785,
      "learning_rate": 9.995913655046811e-05,
      "loss": 0.2435,
      "step": 417
    },
    {
      "epoch": 0.0169013515016123,
      "grad_norm": 7.0280022621154785,
      "learning_rate": 9.9958878365661e-05,
      "loss": 0.2241,
      "step": 418
    },
    {
      "epoch": 0.016941785356879315,
      "grad_norm": 3.5125863552093506,
      "learning_rate": 9.995861936812006e-05,
      "loss": 0.11,
      "step": 419
    },
    {
      "epoch": 0.01698221921214633,
      "grad_norm": 5.077792167663574,
      "learning_rate": 9.995835955784948e-05,
      "loss": 0.2184,
      "step": 420
    },
    {
      "epoch": 0.017022653067413344,
      "grad_norm": 13.11986255645752,
      "learning_rate": 9.995809893485355e-05,
      "loss": 0.3454,
      "step": 421
    },
    {
      "epoch": 0.01706308692268036,
      "grad_norm": 2.720851182937622,
      "learning_rate": 9.995783749913647e-05,
      "loss": 0.0694,
      "step": 422
    },
    {
      "epoch": 0.017103520777947374,
      "grad_norm": 6.168106555938721,
      "learning_rate": 9.995757525070248e-05,
      "loss": 0.1346,
      "step": 423
    },
    {
      "epoch": 0.01714395463321439,
      "grad_norm": 9.339815139770508,
      "learning_rate": 9.995731218955587e-05,
      "loss": 0.319,
      "step": 424
    },
    {
      "epoch": 0.017184388488481406,
      "grad_norm": 3.1323177814483643,
      "learning_rate": 9.995704831570092e-05,
      "loss": 0.0676,
      "step": 425
    },
    {
      "epoch": 0.01722482234374842,
      "grad_norm": 3.4409823417663574,
      "learning_rate": 9.99567836291419e-05,
      "loss": 0.183,
      "step": 426
    },
    {
      "epoch": 0.017265256199015436,
      "grad_norm": 3.833364248275757,
      "learning_rate": 9.995651812988313e-05,
      "loss": 0.151,
      "step": 427
    },
    {
      "epoch": 0.017305690054282452,
      "grad_norm": 5.801759243011475,
      "learning_rate": 9.995625181792894e-05,
      "loss": 0.2757,
      "step": 428
    },
    {
      "epoch": 0.017346123909549465,
      "grad_norm": 4.068567276000977,
      "learning_rate": 9.995598469328365e-05,
      "loss": 0.1181,
      "step": 429
    },
    {
      "epoch": 0.01738655776481648,
      "grad_norm": 7.235023021697998,
      "learning_rate": 9.995571675595162e-05,
      "loss": 0.2227,
      "step": 430
    },
    {
      "epoch": 0.017426991620083494,
      "grad_norm": 4.090055465698242,
      "learning_rate": 9.995544800593717e-05,
      "loss": 0.1512,
      "step": 431
    },
    {
      "epoch": 0.01746742547535051,
      "grad_norm": 3.7105469703674316,
      "learning_rate": 9.995517844324472e-05,
      "loss": 0.0768,
      "step": 432
    },
    {
      "epoch": 0.017507859330617527,
      "grad_norm": 14.445322036743164,
      "learning_rate": 9.995490806787862e-05,
      "loss": 0.3164,
      "step": 433
    },
    {
      "epoch": 0.01754829318588454,
      "grad_norm": 4.54771089553833,
      "learning_rate": 9.99546368798433e-05,
      "loss": 0.2829,
      "step": 434
    },
    {
      "epoch": 0.017588727041151556,
      "grad_norm": 4.201461315155029,
      "learning_rate": 9.995436487914313e-05,
      "loss": 0.2141,
      "step": 435
    },
    {
      "epoch": 0.017629160896418573,
      "grad_norm": 3.3759236335754395,
      "learning_rate": 9.995409206578258e-05,
      "loss": 0.1236,
      "step": 436
    },
    {
      "epoch": 0.017669594751685586,
      "grad_norm": 6.27281379699707,
      "learning_rate": 9.995381843976606e-05,
      "loss": 0.2143,
      "step": 437
    },
    {
      "epoch": 0.017710028606952602,
      "grad_norm": 3.710667848587036,
      "learning_rate": 9.995354400109802e-05,
      "loss": 0.1455,
      "step": 438
    },
    {
      "epoch": 0.017750462462219615,
      "grad_norm": 5.249466419219971,
      "learning_rate": 9.995326874978295e-05,
      "loss": 0.1711,
      "step": 439
    },
    {
      "epoch": 0.01779089631748663,
      "grad_norm": 4.103797912597656,
      "learning_rate": 9.99529926858253e-05,
      "loss": 0.1697,
      "step": 440
    },
    {
      "epoch": 0.017831330172753648,
      "grad_norm": 11.167952537536621,
      "learning_rate": 9.995271580922958e-05,
      "loss": 0.3297,
      "step": 441
    },
    {
      "epoch": 0.01787176402802066,
      "grad_norm": 5.391033172607422,
      "learning_rate": 9.995243812000028e-05,
      "loss": 0.2127,
      "step": 442
    },
    {
      "epoch": 0.017912197883287677,
      "grad_norm": 5.8259196281433105,
      "learning_rate": 9.995215961814193e-05,
      "loss": 0.1725,
      "step": 443
    },
    {
      "epoch": 0.017952631738554693,
      "grad_norm": 4.095852375030518,
      "learning_rate": 9.995188030365903e-05,
      "loss": 0.2509,
      "step": 444
    },
    {
      "epoch": 0.017993065593821706,
      "grad_norm": 5.3281331062316895,
      "learning_rate": 9.995160017655618e-05,
      "loss": 0.1941,
      "step": 445
    },
    {
      "epoch": 0.018033499449088723,
      "grad_norm": 5.681808948516846,
      "learning_rate": 9.99513192368379e-05,
      "loss": 0.108,
      "step": 446
    },
    {
      "epoch": 0.018073933304355735,
      "grad_norm": 5.700495719909668,
      "learning_rate": 9.995103748450876e-05,
      "loss": 0.2278,
      "step": 447
    },
    {
      "epoch": 0.018114367159622752,
      "grad_norm": 4.327421188354492,
      "learning_rate": 9.995075491957334e-05,
      "loss": 0.2088,
      "step": 448
    },
    {
      "epoch": 0.018154801014889768,
      "grad_norm": 4.755593776702881,
      "learning_rate": 9.995047154203626e-05,
      "loss": 0.1653,
      "step": 449
    },
    {
      "epoch": 0.01819523487015678,
      "grad_norm": 7.158329486846924,
      "learning_rate": 9.99501873519021e-05,
      "loss": 0.2319,
      "step": 450
    },
    {
      "epoch": 0.018235668725423797,
      "grad_norm": 15.826255798339844,
      "learning_rate": 9.994990234917551e-05,
      "loss": 0.2939,
      "step": 451
    },
    {
      "epoch": 0.018276102580690814,
      "grad_norm": 13.161656379699707,
      "learning_rate": 9.994961653386113e-05,
      "loss": 0.2945,
      "step": 452
    },
    {
      "epoch": 0.018316536435957827,
      "grad_norm": 14.521732330322266,
      "learning_rate": 9.994932990596358e-05,
      "loss": 0.4143,
      "step": 453
    },
    {
      "epoch": 0.018356970291224843,
      "grad_norm": 9.566047668457031,
      "learning_rate": 9.994904246548754e-05,
      "loss": 0.3087,
      "step": 454
    },
    {
      "epoch": 0.018397404146491856,
      "grad_norm": 9.943325996398926,
      "learning_rate": 9.994875421243767e-05,
      "loss": 0.2999,
      "step": 455
    },
    {
      "epoch": 0.018437838001758872,
      "grad_norm": 3.9468533992767334,
      "learning_rate": 9.99484651468187e-05,
      "loss": 0.1816,
      "step": 456
    },
    {
      "epoch": 0.01847827185702589,
      "grad_norm": 9.679075241088867,
      "learning_rate": 9.99481752686353e-05,
      "loss": 0.2048,
      "step": 457
    },
    {
      "epoch": 0.0185187057122929,
      "grad_norm": 4.950482368469238,
      "learning_rate": 9.994788457789217e-05,
      "loss": 0.144,
      "step": 458
    },
    {
      "epoch": 0.018559139567559918,
      "grad_norm": 12.975287437438965,
      "learning_rate": 9.994759307459408e-05,
      "loss": 0.2848,
      "step": 459
    },
    {
      "epoch": 0.018599573422826934,
      "grad_norm": 5.840356349945068,
      "learning_rate": 9.994730075874575e-05,
      "loss": 0.1907,
      "step": 460
    },
    {
      "epoch": 0.018640007278093947,
      "grad_norm": 14.390122413635254,
      "learning_rate": 9.994700763035194e-05,
      "loss": 0.3579,
      "step": 461
    },
    {
      "epoch": 0.018680441133360964,
      "grad_norm": 5.588078498840332,
      "learning_rate": 9.994671368941741e-05,
      "loss": 0.1293,
      "step": 462
    },
    {
      "epoch": 0.018720874988627977,
      "grad_norm": 5.6230854988098145,
      "learning_rate": 9.994641893594696e-05,
      "loss": 0.2584,
      "step": 463
    },
    {
      "epoch": 0.018761308843894993,
      "grad_norm": 6.206297874450684,
      "learning_rate": 9.994612336994537e-05,
      "loss": 0.1587,
      "step": 464
    },
    {
      "epoch": 0.01880174269916201,
      "grad_norm": 5.694906711578369,
      "learning_rate": 9.994582699141745e-05,
      "loss": 0.1666,
      "step": 465
    },
    {
      "epoch": 0.018842176554429022,
      "grad_norm": 6.807285308837891,
      "learning_rate": 9.994552980036804e-05,
      "loss": 0.1615,
      "step": 466
    },
    {
      "epoch": 0.01888261040969604,
      "grad_norm": 11.35270881652832,
      "learning_rate": 9.994523179680193e-05,
      "loss": 0.2496,
      "step": 467
    },
    {
      "epoch": 0.018923044264963055,
      "grad_norm": 5.333317279815674,
      "learning_rate": 9.994493298072401e-05,
      "loss": 0.2415,
      "step": 468
    },
    {
      "epoch": 0.018963478120230068,
      "grad_norm": 2.6711456775665283,
      "learning_rate": 9.994463335213914e-05,
      "loss": 0.0961,
      "step": 469
    },
    {
      "epoch": 0.019003911975497084,
      "grad_norm": 9.240446090698242,
      "learning_rate": 9.994433291105216e-05,
      "loss": 0.2672,
      "step": 470
    },
    {
      "epoch": 0.019044345830764097,
      "grad_norm": 6.926660060882568,
      "learning_rate": 9.994403165746799e-05,
      "loss": 0.146,
      "step": 471
    },
    {
      "epoch": 0.019084779686031113,
      "grad_norm": 9.662714004516602,
      "learning_rate": 9.994372959139154e-05,
      "loss": 0.2153,
      "step": 472
    },
    {
      "epoch": 0.01912521354129813,
      "grad_norm": 9.086503028869629,
      "learning_rate": 9.994342671282769e-05,
      "loss": 0.3234,
      "step": 473
    },
    {
      "epoch": 0.019165647396565143,
      "grad_norm": 3.6324682235717773,
      "learning_rate": 9.994312302178138e-05,
      "loss": 0.1072,
      "step": 474
    },
    {
      "epoch": 0.01920608125183216,
      "grad_norm": 6.956814765930176,
      "learning_rate": 9.994281851825753e-05,
      "loss": 0.4186,
      "step": 475
    },
    {
      "epoch": 0.019246515107099176,
      "grad_norm": 5.071710586547852,
      "learning_rate": 9.994251320226115e-05,
      "loss": 0.1056,
      "step": 476
    },
    {
      "epoch": 0.01928694896236619,
      "grad_norm": 4.245335578918457,
      "learning_rate": 9.994220707379716e-05,
      "loss": 0.273,
      "step": 477
    },
    {
      "epoch": 0.019327382817633205,
      "grad_norm": 5.091325283050537,
      "learning_rate": 9.994190013287054e-05,
      "loss": 0.1432,
      "step": 478
    },
    {
      "epoch": 0.019367816672900218,
      "grad_norm": 6.0667009353637695,
      "learning_rate": 9.994159237948631e-05,
      "loss": 0.1769,
      "step": 479
    },
    {
      "epoch": 0.019408250528167234,
      "grad_norm": 4.6784281730651855,
      "learning_rate": 9.994128381364947e-05,
      "loss": 0.1911,
      "step": 480
    },
    {
      "epoch": 0.01944868438343425,
      "grad_norm": 5.2691650390625,
      "learning_rate": 9.994097443536502e-05,
      "loss": 0.1886,
      "step": 481
    },
    {
      "epoch": 0.019489118238701263,
      "grad_norm": 5.6240925788879395,
      "learning_rate": 9.994066424463802e-05,
      "loss": 0.2848,
      "step": 482
    },
    {
      "epoch": 0.01952955209396828,
      "grad_norm": 13.263115882873535,
      "learning_rate": 9.994035324147349e-05,
      "loss": 0.4158,
      "step": 483
    },
    {
      "epoch": 0.019569985949235296,
      "grad_norm": 11.123750686645508,
      "learning_rate": 9.994004142587649e-05,
      "loss": 0.226,
      "step": 484
    },
    {
      "epoch": 0.01961041980450231,
      "grad_norm": 9.562678337097168,
      "learning_rate": 9.993972879785211e-05,
      "loss": 0.2728,
      "step": 485
    },
    {
      "epoch": 0.019650853659769325,
      "grad_norm": 3.476072072982788,
      "learning_rate": 9.993941535740546e-05,
      "loss": 0.3587,
      "step": 486
    },
    {
      "epoch": 0.019691287515036338,
      "grad_norm": 4.485725402832031,
      "learning_rate": 9.993910110454156e-05,
      "loss": 0.1258,
      "step": 487
    },
    {
      "epoch": 0.019731721370303355,
      "grad_norm": 7.6921916007995605,
      "learning_rate": 9.993878603926561e-05,
      "loss": 0.2003,
      "step": 488
    },
    {
      "epoch": 0.01977215522557037,
      "grad_norm": 9.044139862060547,
      "learning_rate": 9.993847016158269e-05,
      "loss": 0.1727,
      "step": 489
    },
    {
      "epoch": 0.019812589080837384,
      "grad_norm": 4.3051605224609375,
      "learning_rate": 9.993815347149792e-05,
      "loss": 0.1498,
      "step": 490
    },
    {
      "epoch": 0.0198530229361044,
      "grad_norm": 8.388853073120117,
      "learning_rate": 9.993783596901652e-05,
      "loss": 0.2131,
      "step": 491
    },
    {
      "epoch": 0.019893456791371417,
      "grad_norm": 6.8371405601501465,
      "learning_rate": 9.993751765414358e-05,
      "loss": 0.1996,
      "step": 492
    },
    {
      "epoch": 0.01993389064663843,
      "grad_norm": 9.110433578491211,
      "learning_rate": 9.99371985268843e-05,
      "loss": 0.2001,
      "step": 493
    },
    {
      "epoch": 0.019974324501905446,
      "grad_norm": 3.504476308822632,
      "learning_rate": 9.993687858724392e-05,
      "loss": 0.0949,
      "step": 494
    },
    {
      "epoch": 0.02001475835717246,
      "grad_norm": 6.312266826629639,
      "learning_rate": 9.993655783522758e-05,
      "loss": 0.345,
      "step": 495
    },
    {
      "epoch": 0.020055192212439475,
      "grad_norm": 2.2558364868164062,
      "learning_rate": 9.993623627084054e-05,
      "loss": 0.0706,
      "step": 496
    },
    {
      "epoch": 0.02009562606770649,
      "grad_norm": 6.923905372619629,
      "learning_rate": 9.993591389408801e-05,
      "loss": 0.3129,
      "step": 497
    },
    {
      "epoch": 0.020136059922973504,
      "grad_norm": 3.831315040588379,
      "learning_rate": 9.993559070497523e-05,
      "loss": 0.1103,
      "step": 498
    },
    {
      "epoch": 0.02017649377824052,
      "grad_norm": 5.774306774139404,
      "learning_rate": 9.993526670350748e-05,
      "loss": 0.2065,
      "step": 499
    },
    {
      "epoch": 0.020216927633507537,
      "grad_norm": 11.199854850769043,
      "learning_rate": 9.993494188969002e-05,
      "loss": 0.2734,
      "step": 500
    },
    {
      "epoch": 0.02025736148877455,
      "grad_norm": 15.771286010742188,
      "learning_rate": 9.993461626352813e-05,
      "loss": 0.3646,
      "step": 501
    },
    {
      "epoch": 0.020297795344041566,
      "grad_norm": 11.929840087890625,
      "learning_rate": 9.993428982502712e-05,
      "loss": 0.2521,
      "step": 502
    },
    {
      "epoch": 0.02033822919930858,
      "grad_norm": 5.950290203094482,
      "learning_rate": 9.993396257419228e-05,
      "loss": 0.279,
      "step": 503
    },
    {
      "epoch": 0.020378663054575596,
      "grad_norm": 10.887775421142578,
      "learning_rate": 9.993363451102894e-05,
      "loss": 0.3195,
      "step": 504
    },
    {
      "epoch": 0.020419096909842612,
      "grad_norm": 4.067485332489014,
      "learning_rate": 9.993330563554246e-05,
      "loss": 0.2364,
      "step": 505
    },
    {
      "epoch": 0.020459530765109625,
      "grad_norm": 2.6544528007507324,
      "learning_rate": 9.993297594773816e-05,
      "loss": 0.1916,
      "step": 506
    },
    {
      "epoch": 0.02049996462037664,
      "grad_norm": 3.099802017211914,
      "learning_rate": 9.993264544762142e-05,
      "loss": 0.1606,
      "step": 507
    },
    {
      "epoch": 0.020540398475643658,
      "grad_norm": 9.183789253234863,
      "learning_rate": 9.993231413519761e-05,
      "loss": 0.2167,
      "step": 508
    },
    {
      "epoch": 0.02058083233091067,
      "grad_norm": 15.220685005187988,
      "learning_rate": 9.993198201047211e-05,
      "loss": 0.3037,
      "step": 509
    },
    {
      "epoch": 0.020621266186177687,
      "grad_norm": 7.174010276794434,
      "learning_rate": 9.993164907345036e-05,
      "loss": 0.18,
      "step": 510
    },
    {
      "epoch": 0.0206617000414447,
      "grad_norm": 9.323453903198242,
      "learning_rate": 9.993131532413773e-05,
      "loss": 0.2346,
      "step": 511
    },
    {
      "epoch": 0.020702133896711716,
      "grad_norm": 9.454221725463867,
      "learning_rate": 9.993098076253968e-05,
      "loss": 0.2933,
      "step": 512
    },
    {
      "epoch": 0.020742567751978733,
      "grad_norm": 5.286993503570557,
      "learning_rate": 9.993064538866164e-05,
      "loss": 0.2368,
      "step": 513
    },
    {
      "epoch": 0.020783001607245746,
      "grad_norm": 6.411348819732666,
      "learning_rate": 9.993030920250907e-05,
      "loss": 0.2048,
      "step": 514
    },
    {
      "epoch": 0.020823435462512762,
      "grad_norm": 4.886948585510254,
      "learning_rate": 9.992997220408744e-05,
      "loss": 0.0842,
      "step": 515
    },
    {
      "epoch": 0.02086386931777978,
      "grad_norm": 5.179337978363037,
      "learning_rate": 9.992963439340224e-05,
      "loss": 0.2604,
      "step": 516
    },
    {
      "epoch": 0.02090430317304679,
      "grad_norm": 3.0143651962280273,
      "learning_rate": 9.992929577045894e-05,
      "loss": 0.1686,
      "step": 517
    },
    {
      "epoch": 0.020944737028313808,
      "grad_norm": 5.324845790863037,
      "learning_rate": 9.992895633526307e-05,
      "loss": 0.1658,
      "step": 518
    },
    {
      "epoch": 0.02098517088358082,
      "grad_norm": 4.457522392272949,
      "learning_rate": 9.992861608782015e-05,
      "loss": 0.1264,
      "step": 519
    },
    {
      "epoch": 0.021025604738847837,
      "grad_norm": 8.354854583740234,
      "learning_rate": 9.99282750281357e-05,
      "loss": 0.1949,
      "step": 520
    },
    {
      "epoch": 0.021066038594114853,
      "grad_norm": 4.758010387420654,
      "learning_rate": 9.99279331562153e-05,
      "loss": 0.1307,
      "step": 521
    },
    {
      "epoch": 0.021106472449381866,
      "grad_norm": 7.02042293548584,
      "learning_rate": 9.992759047206448e-05,
      "loss": 0.2242,
      "step": 522
    },
    {
      "epoch": 0.021146906304648883,
      "grad_norm": 5.318552017211914,
      "learning_rate": 9.992724697568882e-05,
      "loss": 0.2815,
      "step": 523
    },
    {
      "epoch": 0.0211873401599159,
      "grad_norm": 6.275263786315918,
      "learning_rate": 9.992690266709392e-05,
      "loss": 0.1755,
      "step": 524
    },
    {
      "epoch": 0.021227774015182912,
      "grad_norm": 3.590404510498047,
      "learning_rate": 9.992655754628539e-05,
      "loss": 0.1313,
      "step": 525
    },
    {
      "epoch": 0.021268207870449928,
      "grad_norm": 4.934257507324219,
      "learning_rate": 9.992621161326881e-05,
      "loss": 0.1565,
      "step": 526
    },
    {
      "epoch": 0.02130864172571694,
      "grad_norm": 5.31673002243042,
      "learning_rate": 9.992586486804985e-05,
      "loss": 0.1816,
      "step": 527
    },
    {
      "epoch": 0.021349075580983957,
      "grad_norm": 11.531829833984375,
      "learning_rate": 9.992551731063411e-05,
      "loss": 0.2601,
      "step": 528
    },
    {
      "epoch": 0.021389509436250974,
      "grad_norm": 5.013904571533203,
      "learning_rate": 9.992516894102727e-05,
      "loss": 0.1521,
      "step": 529
    },
    {
      "epoch": 0.021429943291517987,
      "grad_norm": 4.303083419799805,
      "learning_rate": 9.9924819759235e-05,
      "loss": 0.1413,
      "step": 530
    },
    {
      "epoch": 0.021470377146785003,
      "grad_norm": 8.451056480407715,
      "learning_rate": 9.992446976526294e-05,
      "loss": 0.2917,
      "step": 531
    },
    {
      "epoch": 0.02151081100205202,
      "grad_norm": 3.1860673427581787,
      "learning_rate": 9.992411895911684e-05,
      "loss": 0.0583,
      "step": 532
    },
    {
      "epoch": 0.021551244857319032,
      "grad_norm": 8.839323997497559,
      "learning_rate": 9.992376734080237e-05,
      "loss": 0.2227,
      "step": 533
    },
    {
      "epoch": 0.02159167871258605,
      "grad_norm": 3.5903005599975586,
      "learning_rate": 9.992341491032528e-05,
      "loss": 0.2177,
      "step": 534
    },
    {
      "epoch": 0.02163211256785306,
      "grad_norm": 4.861203670501709,
      "learning_rate": 9.992306166769127e-05,
      "loss": 0.1614,
      "step": 535
    },
    {
      "epoch": 0.021672546423120078,
      "grad_norm": 9.597570419311523,
      "learning_rate": 9.992270761290609e-05,
      "loss": 0.2092,
      "step": 536
    },
    {
      "epoch": 0.021712980278387094,
      "grad_norm": 7.367400169372559,
      "learning_rate": 9.992235274597555e-05,
      "loss": 0.261,
      "step": 537
    },
    {
      "epoch": 0.021753414133654107,
      "grad_norm": 8.208807945251465,
      "learning_rate": 9.992199706690535e-05,
      "loss": 0.1807,
      "step": 538
    },
    {
      "epoch": 0.021793847988921124,
      "grad_norm": 9.680448532104492,
      "learning_rate": 9.992164057570132e-05,
      "loss": 0.134,
      "step": 539
    },
    {
      "epoch": 0.02183428184418814,
      "grad_norm": 9.173952102661133,
      "learning_rate": 9.992128327236925e-05,
      "loss": 0.1794,
      "step": 540
    },
    {
      "epoch": 0.021874715699455153,
      "grad_norm": 5.462802410125732,
      "learning_rate": 9.992092515691496e-05,
      "loss": 0.2219,
      "step": 541
    },
    {
      "epoch": 0.02191514955472217,
      "grad_norm": 10.719511985778809,
      "learning_rate": 9.992056622934425e-05,
      "loss": 0.2187,
      "step": 542
    },
    {
      "epoch": 0.021955583409989182,
      "grad_norm": 7.020593166351318,
      "learning_rate": 9.992020648966299e-05,
      "loss": 0.2417,
      "step": 543
    },
    {
      "epoch": 0.0219960172652562,
      "grad_norm": 7.914300918579102,
      "learning_rate": 9.991984593787702e-05,
      "loss": 0.2056,
      "step": 544
    },
    {
      "epoch": 0.022036451120523215,
      "grad_norm": 4.527122974395752,
      "learning_rate": 9.991948457399221e-05,
      "loss": 0.2227,
      "step": 545
    },
    {
      "epoch": 0.022076884975790228,
      "grad_norm": 3.5496299266815186,
      "learning_rate": 9.991912239801442e-05,
      "loss": 0.0694,
      "step": 546
    },
    {
      "epoch": 0.022117318831057244,
      "grad_norm": 9.399176597595215,
      "learning_rate": 9.991875940994957e-05,
      "loss": 0.2068,
      "step": 547
    },
    {
      "epoch": 0.02215775268632426,
      "grad_norm": 4.161809921264648,
      "learning_rate": 9.991839560980354e-05,
      "loss": 0.1092,
      "step": 548
    },
    {
      "epoch": 0.022198186541591274,
      "grad_norm": 4.159708023071289,
      "learning_rate": 9.991803099758226e-05,
      "loss": 0.1153,
      "step": 549
    },
    {
      "epoch": 0.02223862039685829,
      "grad_norm": 13.013190269470215,
      "learning_rate": 9.991766557329167e-05,
      "loss": 0.3937,
      "step": 550
    },
    {
      "epoch": 0.022279054252125303,
      "grad_norm": 7.847819805145264,
      "learning_rate": 9.99172993369377e-05,
      "loss": 0.3498,
      "step": 551
    },
    {
      "epoch": 0.02231948810739232,
      "grad_norm": 4.256953239440918,
      "learning_rate": 9.99169322885263e-05,
      "loss": 0.1811,
      "step": 552
    },
    {
      "epoch": 0.022359921962659336,
      "grad_norm": 9.252525329589844,
      "learning_rate": 9.991656442806347e-05,
      "loss": 0.4303,
      "step": 553
    },
    {
      "epoch": 0.02240035581792635,
      "grad_norm": 5.369123935699463,
      "learning_rate": 9.991619575555518e-05,
      "loss": 0.2266,
      "step": 554
    },
    {
      "epoch": 0.022440789673193365,
      "grad_norm": 5.058295726776123,
      "learning_rate": 9.991582627100742e-05,
      "loss": 0.1266,
      "step": 555
    },
    {
      "epoch": 0.02248122352846038,
      "grad_norm": 1.8897539377212524,
      "learning_rate": 9.991545597442622e-05,
      "loss": 0.0642,
      "step": 556
    },
    {
      "epoch": 0.022521657383727394,
      "grad_norm": 4.504596710205078,
      "learning_rate": 9.991508486581757e-05,
      "loss": 0.2547,
      "step": 557
    },
    {
      "epoch": 0.02256209123899441,
      "grad_norm": 3.1885995864868164,
      "learning_rate": 9.991471294518755e-05,
      "loss": 0.1891,
      "step": 558
    },
    {
      "epoch": 0.022602525094261423,
      "grad_norm": 3.646301746368408,
      "learning_rate": 9.991434021254219e-05,
      "loss": 0.1742,
      "step": 559
    },
    {
      "epoch": 0.02264295894952844,
      "grad_norm": 9.444262504577637,
      "learning_rate": 9.991396666788753e-05,
      "loss": 0.2703,
      "step": 560
    },
    {
      "epoch": 0.022683392804795456,
      "grad_norm": 3.328944206237793,
      "learning_rate": 9.991359231122968e-05,
      "loss": 0.1655,
      "step": 561
    },
    {
      "epoch": 0.02272382666006247,
      "grad_norm": 3.9876956939697266,
      "learning_rate": 9.991321714257472e-05,
      "loss": 0.1785,
      "step": 562
    },
    {
      "epoch": 0.022764260515329485,
      "grad_norm": 6.405686855316162,
      "learning_rate": 9.991284116192875e-05,
      "loss": 0.3021,
      "step": 563
    },
    {
      "epoch": 0.022804694370596502,
      "grad_norm": 9.157509803771973,
      "learning_rate": 9.991246436929791e-05,
      "loss": 0.3052,
      "step": 564
    },
    {
      "epoch": 0.022845128225863515,
      "grad_norm": 4.270457744598389,
      "learning_rate": 9.991208676468829e-05,
      "loss": 0.1162,
      "step": 565
    },
    {
      "epoch": 0.02288556208113053,
      "grad_norm": 4.0909271240234375,
      "learning_rate": 9.991170834810604e-05,
      "loss": 0.12,
      "step": 566
    },
    {
      "epoch": 0.022925995936397544,
      "grad_norm": 4.51482629776001,
      "learning_rate": 9.991132911955734e-05,
      "loss": 0.1831,
      "step": 567
    },
    {
      "epoch": 0.02296642979166456,
      "grad_norm": 6.682913780212402,
      "learning_rate": 9.991094907904837e-05,
      "loss": 0.1584,
      "step": 568
    },
    {
      "epoch": 0.023006863646931577,
      "grad_norm": 6.1220173835754395,
      "learning_rate": 9.991056822658526e-05,
      "loss": 0.1802,
      "step": 569
    },
    {
      "epoch": 0.02304729750219859,
      "grad_norm": 8.298602104187012,
      "learning_rate": 9.991018656217424e-05,
      "loss": 0.3793,
      "step": 570
    },
    {
      "epoch": 0.023087731357465606,
      "grad_norm": 5.743035316467285,
      "learning_rate": 9.990980408582152e-05,
      "loss": 0.247,
      "step": 571
    },
    {
      "epoch": 0.023128165212732622,
      "grad_norm": 5.235231876373291,
      "learning_rate": 9.990942079753331e-05,
      "loss": 0.2558,
      "step": 572
    },
    {
      "epoch": 0.023168599067999635,
      "grad_norm": 4.871448516845703,
      "learning_rate": 9.990903669731586e-05,
      "loss": 0.2043,
      "step": 573
    },
    {
      "epoch": 0.02320903292326665,
      "grad_norm": 9.752013206481934,
      "learning_rate": 9.990865178517541e-05,
      "loss": 0.1533,
      "step": 574
    },
    {
      "epoch": 0.023249466778533668,
      "grad_norm": 4.829668045043945,
      "learning_rate": 9.990826606111823e-05,
      "loss": 0.1735,
      "step": 575
    },
    {
      "epoch": 0.02328990063380068,
      "grad_norm": 5.898210525512695,
      "learning_rate": 9.990787952515058e-05,
      "loss": 0.2135,
      "step": 576
    },
    {
      "epoch": 0.023330334489067697,
      "grad_norm": 5.213702201843262,
      "learning_rate": 9.990749217727875e-05,
      "loss": 0.1842,
      "step": 577
    },
    {
      "epoch": 0.02337076834433471,
      "grad_norm": 6.2908453941345215,
      "learning_rate": 9.990710401750905e-05,
      "loss": 0.1595,
      "step": 578
    },
    {
      "epoch": 0.023411202199601727,
      "grad_norm": 6.161911487579346,
      "learning_rate": 9.990671504584778e-05,
      "loss": 0.1074,
      "step": 579
    },
    {
      "epoch": 0.023451636054868743,
      "grad_norm": 3.668738842010498,
      "learning_rate": 9.990632526230131e-05,
      "loss": 0.1659,
      "step": 580
    },
    {
      "epoch": 0.023492069910135756,
      "grad_norm": 4.5764665603637695,
      "learning_rate": 9.990593466687592e-05,
      "loss": 0.1327,
      "step": 581
    },
    {
      "epoch": 0.023532503765402772,
      "grad_norm": 3.269049882888794,
      "learning_rate": 9.9905543259578e-05,
      "loss": 0.1402,
      "step": 582
    },
    {
      "epoch": 0.02357293762066979,
      "grad_norm": 5.684671401977539,
      "learning_rate": 9.990515104041393e-05,
      "loss": 0.0809,
      "step": 583
    },
    {
      "epoch": 0.0236133714759368,
      "grad_norm": 3.8124773502349854,
      "learning_rate": 9.990475800939005e-05,
      "loss": 0.1168,
      "step": 584
    },
    {
      "epoch": 0.023653805331203818,
      "grad_norm": 4.991212368011475,
      "learning_rate": 9.990436416651279e-05,
      "loss": 0.1202,
      "step": 585
    },
    {
      "epoch": 0.02369423918647083,
      "grad_norm": 3.4905292987823486,
      "learning_rate": 9.990396951178854e-05,
      "loss": 0.1412,
      "step": 586
    },
    {
      "epoch": 0.023734673041737847,
      "grad_norm": 4.138618469238281,
      "learning_rate": 9.990357404522372e-05,
      "loss": 0.1006,
      "step": 587
    },
    {
      "epoch": 0.023775106897004863,
      "grad_norm": 10.040672302246094,
      "learning_rate": 9.990317776682479e-05,
      "loss": 0.2141,
      "step": 588
    },
    {
      "epoch": 0.023815540752271876,
      "grad_norm": 3.4008100032806396,
      "learning_rate": 9.990278067659815e-05,
      "loss": 0.0741,
      "step": 589
    },
    {
      "epoch": 0.023855974607538893,
      "grad_norm": 2.954993486404419,
      "learning_rate": 9.990238277455029e-05,
      "loss": 0.0886,
      "step": 590
    },
    {
      "epoch": 0.02389640846280591,
      "grad_norm": 8.132010459899902,
      "learning_rate": 9.990198406068766e-05,
      "loss": 0.1878,
      "step": 591
    },
    {
      "epoch": 0.023936842318072922,
      "grad_norm": 5.87846565246582,
      "learning_rate": 9.990158453501679e-05,
      "loss": 0.2051,
      "step": 592
    },
    {
      "epoch": 0.02397727617333994,
      "grad_norm": 6.974608421325684,
      "learning_rate": 9.990118419754414e-05,
      "loss": 0.3675,
      "step": 593
    },
    {
      "epoch": 0.02401771002860695,
      "grad_norm": 5.92131233215332,
      "learning_rate": 9.990078304827623e-05,
      "loss": 0.2629,
      "step": 594
    },
    {
      "epoch": 0.024058143883873968,
      "grad_norm": 4.652924537658691,
      "learning_rate": 9.99003810872196e-05,
      "loss": 0.2138,
      "step": 595
    },
    {
      "epoch": 0.024098577739140984,
      "grad_norm": 7.456907749176025,
      "learning_rate": 9.989997831438079e-05,
      "loss": 0.2253,
      "step": 596
    },
    {
      "epoch": 0.024139011594407997,
      "grad_norm": 4.625363349914551,
      "learning_rate": 9.989957472976633e-05,
      "loss": 0.223,
      "step": 597
    },
    {
      "epoch": 0.024179445449675013,
      "grad_norm": 7.904758453369141,
      "learning_rate": 9.98991703333828e-05,
      "loss": 0.1499,
      "step": 598
    },
    {
      "epoch": 0.02421987930494203,
      "grad_norm": 2.615682601928711,
      "learning_rate": 9.989876512523678e-05,
      "loss": 0.1338,
      "step": 599
    },
    {
      "epoch": 0.024260313160209043,
      "grad_norm": 15.01430606842041,
      "learning_rate": 9.989835910533485e-05,
      "loss": 0.4107,
      "step": 600
    },
    {
      "epoch": 0.02430074701547606,
      "grad_norm": 13.508016586303711,
      "learning_rate": 9.989795227368364e-05,
      "loss": 0.3165,
      "step": 601
    },
    {
      "epoch": 0.024341180870743072,
      "grad_norm": 12.953534126281738,
      "learning_rate": 9.989754463028974e-05,
      "loss": 0.3711,
      "step": 602
    },
    {
      "epoch": 0.024381614726010088,
      "grad_norm": 14.830718994140625,
      "learning_rate": 9.98971361751598e-05,
      "loss": 0.4022,
      "step": 603
    },
    {
      "epoch": 0.024422048581277105,
      "grad_norm": 13.413792610168457,
      "learning_rate": 9.989672690830046e-05,
      "loss": 0.334,
      "step": 604
    },
    {
      "epoch": 0.024462482436544117,
      "grad_norm": 5.003862380981445,
      "learning_rate": 9.989631682971837e-05,
      "loss": 0.283,
      "step": 605
    },
    {
      "epoch": 0.024502916291811134,
      "grad_norm": 1.462073802947998,
      "learning_rate": 9.989590593942021e-05,
      "loss": 0.1347,
      "step": 606
    },
    {
      "epoch": 0.02454335014707815,
      "grad_norm": 11.80557632446289,
      "learning_rate": 9.989549423741267e-05,
      "loss": 0.3063,
      "step": 607
    },
    {
      "epoch": 0.024583784002345163,
      "grad_norm": 7.493838787078857,
      "learning_rate": 9.989508172370244e-05,
      "loss": 0.1952,
      "step": 608
    },
    {
      "epoch": 0.02462421785761218,
      "grad_norm": 6.926411151885986,
      "learning_rate": 9.989466839829623e-05,
      "loss": 0.291,
      "step": 609
    },
    {
      "epoch": 0.024664651712879192,
      "grad_norm": 3.218076229095459,
      "learning_rate": 9.989425426120073e-05,
      "loss": 0.1934,
      "step": 610
    },
    {
      "epoch": 0.02470508556814621,
      "grad_norm": 5.472060680389404,
      "learning_rate": 9.989383931242276e-05,
      "loss": 0.2018,
      "step": 611
    },
    {
      "epoch": 0.024745519423413225,
      "grad_norm": 2.9727845191955566,
      "learning_rate": 9.9893423551969e-05,
      "loss": 0.1822,
      "step": 612
    },
    {
      "epoch": 0.024785953278680238,
      "grad_norm": 2.999659299850464,
      "learning_rate": 9.989300697984624e-05,
      "loss": 0.2308,
      "step": 613
    },
    {
      "epoch": 0.024826387133947254,
      "grad_norm": 6.793681621551514,
      "learning_rate": 9.989258959606125e-05,
      "loss": 0.2046,
      "step": 614
    },
    {
      "epoch": 0.02486682098921427,
      "grad_norm": 2.2995734214782715,
      "learning_rate": 9.989217140062083e-05,
      "loss": 0.1184,
      "step": 615
    },
    {
      "epoch": 0.024907254844481284,
      "grad_norm": 2.442793846130371,
      "learning_rate": 9.989175239353176e-05,
      "loss": 0.1806,
      "step": 616
    },
    {
      "epoch": 0.0249476886997483,
      "grad_norm": 7.631139278411865,
      "learning_rate": 9.98913325748009e-05,
      "loss": 0.2633,
      "step": 617
    },
    {
      "epoch": 0.024988122555015313,
      "grad_norm": 3.9635837078094482,
      "learning_rate": 9.989091194443503e-05,
      "loss": 0.1934,
      "step": 618
    },
    {
      "epoch": 0.02502855641028233,
      "grad_norm": 3.4910688400268555,
      "learning_rate": 9.989049050244101e-05,
      "loss": 0.2095,
      "step": 619
    },
    {
      "epoch": 0.025068990265549346,
      "grad_norm": 10.227604866027832,
      "learning_rate": 9.989006824882572e-05,
      "loss": 0.3045,
      "step": 620
    },
    {
      "epoch": 0.02510942412081636,
      "grad_norm": 2.927266836166382,
      "learning_rate": 9.988964518359601e-05,
      "loss": 0.177,
      "step": 621
    },
    {
      "epoch": 0.025149857976083375,
      "grad_norm": 9.954039573669434,
      "learning_rate": 9.988922130675874e-05,
      "loss": 0.2808,
      "step": 622
    },
    {
      "epoch": 0.02519029183135039,
      "grad_norm": 4.357359409332275,
      "learning_rate": 9.988879661832086e-05,
      "loss": 0.1336,
      "step": 623
    },
    {
      "epoch": 0.025230725686617404,
      "grad_norm": 4.323237895965576,
      "learning_rate": 9.988837111828923e-05,
      "loss": 0.1469,
      "step": 624
    },
    {
      "epoch": 0.02527115954188442,
      "grad_norm": 2.7954049110412598,
      "learning_rate": 9.988794480667078e-05,
      "loss": 0.1523,
      "step": 625
    },
    {
      "epoch": 0.025311593397151434,
      "grad_norm": 10.383594512939453,
      "learning_rate": 9.988751768347248e-05,
      "loss": 0.2357,
      "step": 626
    },
    {
      "epoch": 0.02535202725241845,
      "grad_norm": 11.658588409423828,
      "learning_rate": 9.988708974870123e-05,
      "loss": 0.3412,
      "step": 627
    },
    {
      "epoch": 0.025392461107685466,
      "grad_norm": 3.1625382900238037,
      "learning_rate": 9.988666100236404e-05,
      "loss": 0.124,
      "step": 628
    },
    {
      "epoch": 0.02543289496295248,
      "grad_norm": 5.50991153717041,
      "learning_rate": 9.988623144446782e-05,
      "loss": 0.1705,
      "step": 629
    },
    {
      "epoch": 0.025473328818219496,
      "grad_norm": 5.058486461639404,
      "learning_rate": 9.988580107501964e-05,
      "loss": 0.23,
      "step": 630
    },
    {
      "epoch": 0.025513762673486512,
      "grad_norm": 6.850649356842041,
      "learning_rate": 9.988536989402644e-05,
      "loss": 0.275,
      "step": 631
    },
    {
      "epoch": 0.025554196528753525,
      "grad_norm": 5.482839107513428,
      "learning_rate": 9.988493790149525e-05,
      "loss": 0.3801,
      "step": 632
    },
    {
      "epoch": 0.02559463038402054,
      "grad_norm": 8.151007652282715,
      "learning_rate": 9.98845050974331e-05,
      "loss": 0.2501,
      "step": 633
    },
    {
      "epoch": 0.025635064239287554,
      "grad_norm": 7.366528511047363,
      "learning_rate": 9.988407148184705e-05,
      "loss": 0.1655,
      "step": 634
    },
    {
      "epoch": 0.02567549809455457,
      "grad_norm": 7.802221298217773,
      "learning_rate": 9.98836370547441e-05,
      "loss": 0.2149,
      "step": 635
    },
    {
      "epoch": 0.025715931949821587,
      "grad_norm": 10.164329528808594,
      "learning_rate": 9.988320181613137e-05,
      "loss": 0.2048,
      "step": 636
    },
    {
      "epoch": 0.0257563658050886,
      "grad_norm": 10.122926712036133,
      "learning_rate": 9.988276576601594e-05,
      "loss": 0.2335,
      "step": 637
    },
    {
      "epoch": 0.025796799660355616,
      "grad_norm": 7.697607517242432,
      "learning_rate": 9.988232890440488e-05,
      "loss": 0.1431,
      "step": 638
    },
    {
      "epoch": 0.025837233515622633,
      "grad_norm": 3.179948329925537,
      "learning_rate": 9.98818912313053e-05,
      "loss": 0.1368,
      "step": 639
    },
    {
      "epoch": 0.025877667370889645,
      "grad_norm": 8.614984512329102,
      "learning_rate": 9.988145274672432e-05,
      "loss": 0.175,
      "step": 640
    },
    {
      "epoch": 0.025918101226156662,
      "grad_norm": 5.296281814575195,
      "learning_rate": 9.988101345066909e-05,
      "loss": 0.1433,
      "step": 641
    },
    {
      "epoch": 0.025958535081423675,
      "grad_norm": 4.220842361450195,
      "learning_rate": 9.988057334314673e-05,
      "loss": 0.141,
      "step": 642
    },
    {
      "epoch": 0.02599896893669069,
      "grad_norm": 9.185120582580566,
      "learning_rate": 9.988013242416443e-05,
      "loss": 0.1675,
      "step": 643
    },
    {
      "epoch": 0.026039402791957707,
      "grad_norm": 7.908389091491699,
      "learning_rate": 9.987969069372934e-05,
      "loss": 0.1884,
      "step": 644
    },
    {
      "epoch": 0.02607983664722472,
      "grad_norm": 7.235873699188232,
      "learning_rate": 9.987924815184865e-05,
      "loss": 0.1253,
      "step": 645
    },
    {
      "epoch": 0.026120270502491737,
      "grad_norm": 7.630735397338867,
      "learning_rate": 9.987880479852956e-05,
      "loss": 0.299,
      "step": 646
    },
    {
      "epoch": 0.026160704357758753,
      "grad_norm": 4.494728088378906,
      "learning_rate": 9.987836063377929e-05,
      "loss": 0.0955,
      "step": 647
    },
    {
      "epoch": 0.026201138213025766,
      "grad_norm": 7.4130144119262695,
      "learning_rate": 9.987791565760506e-05,
      "loss": 0.212,
      "step": 648
    },
    {
      "epoch": 0.026241572068292782,
      "grad_norm": 8.46578598022461,
      "learning_rate": 9.98774698700141e-05,
      "loss": 0.2048,
      "step": 649
    },
    {
      "epoch": 0.026282005923559795,
      "grad_norm": 7.845007419586182,
      "learning_rate": 9.987702327101369e-05,
      "loss": 0.2837,
      "step": 650
    },
    {
      "epoch": 0.02632243977882681,
      "grad_norm": 7.150285720825195,
      "learning_rate": 9.987657586061106e-05,
      "loss": 0.1171,
      "step": 651
    },
    {
      "epoch": 0.026362873634093828,
      "grad_norm": 9.526632308959961,
      "learning_rate": 9.98761276388135e-05,
      "loss": 0.4257,
      "step": 652
    },
    {
      "epoch": 0.02640330748936084,
      "grad_norm": 15.521952629089355,
      "learning_rate": 9.987567860562833e-05,
      "loss": 0.4389,
      "step": 653
    },
    {
      "epoch": 0.026443741344627857,
      "grad_norm": 9.229066848754883,
      "learning_rate": 9.987522876106283e-05,
      "loss": 0.3371,
      "step": 654
    },
    {
      "epoch": 0.026484175199894874,
      "grad_norm": 9.44521713256836,
      "learning_rate": 9.98747781051243e-05,
      "loss": 0.4914,
      "step": 655
    },
    {
      "epoch": 0.026524609055161887,
      "grad_norm": 8.148359298706055,
      "learning_rate": 9.98743266378201e-05,
      "loss": 0.2649,
      "step": 656
    },
    {
      "epoch": 0.026565042910428903,
      "grad_norm": 2.5083189010620117,
      "learning_rate": 9.987387435915757e-05,
      "loss": 0.1646,
      "step": 657
    },
    {
      "epoch": 0.026605476765695916,
      "grad_norm": 6.02241325378418,
      "learning_rate": 9.987342126914406e-05,
      "loss": 0.1544,
      "step": 658
    },
    {
      "epoch": 0.026645910620962932,
      "grad_norm": 6.7186737060546875,
      "learning_rate": 9.987296736778696e-05,
      "loss": 0.329,
      "step": 659
    },
    {
      "epoch": 0.02668634447622995,
      "grad_norm": 7.492613792419434,
      "learning_rate": 9.987251265509361e-05,
      "loss": 0.2352,
      "step": 660
    },
    {
      "epoch": 0.02672677833149696,
      "grad_norm": 9.458938598632812,
      "learning_rate": 9.987205713107145e-05,
      "loss": 0.31,
      "step": 661
    },
    {
      "epoch": 0.026767212186763978,
      "grad_norm": 8.664854049682617,
      "learning_rate": 9.987160079572788e-05,
      "loss": 0.1648,
      "step": 662
    },
    {
      "epoch": 0.026807646042030994,
      "grad_norm": 10.723663330078125,
      "learning_rate": 9.98711436490703e-05,
      "loss": 0.2615,
      "step": 663
    },
    {
      "epoch": 0.026848079897298007,
      "grad_norm": 9.836989402770996,
      "learning_rate": 9.98706856911062e-05,
      "loss": 0.3566,
      "step": 664
    },
    {
      "epoch": 0.026888513752565023,
      "grad_norm": 5.883004188537598,
      "learning_rate": 9.987022692184297e-05,
      "loss": 0.2396,
      "step": 665
    },
    {
      "epoch": 0.026928947607832036,
      "grad_norm": 8.773782730102539,
      "learning_rate": 9.986976734128811e-05,
      "loss": 0.2597,
      "step": 666
    },
    {
      "epoch": 0.026969381463099053,
      "grad_norm": 5.26301908493042,
      "learning_rate": 9.98693069494491e-05,
      "loss": 0.1271,
      "step": 667
    },
    {
      "epoch": 0.02700981531836607,
      "grad_norm": 5.541378498077393,
      "learning_rate": 9.98688457463334e-05,
      "loss": 0.2659,
      "step": 668
    },
    {
      "epoch": 0.027050249173633082,
      "grad_norm": 4.1378655433654785,
      "learning_rate": 9.98683837319485e-05,
      "loss": 0.1846,
      "step": 669
    },
    {
      "epoch": 0.0270906830289001,
      "grad_norm": 7.509443759918213,
      "learning_rate": 9.986792090630199e-05,
      "loss": 0.2538,
      "step": 670
    },
    {
      "epoch": 0.027131116884167115,
      "grad_norm": 9.867937088012695,
      "learning_rate": 9.986745726940134e-05,
      "loss": 0.2974,
      "step": 671
    },
    {
      "epoch": 0.027171550739434128,
      "grad_norm": 3.8687992095947266,
      "learning_rate": 9.986699282125412e-05,
      "loss": 0.1809,
      "step": 672
    },
    {
      "epoch": 0.027211984594701144,
      "grad_norm": 8.495218276977539,
      "learning_rate": 9.986652756186784e-05,
      "loss": 0.2481,
      "step": 673
    },
    {
      "epoch": 0.027252418449968157,
      "grad_norm": 4.442294597625732,
      "learning_rate": 9.986606149125013e-05,
      "loss": 0.1718,
      "step": 674
    },
    {
      "epoch": 0.027292852305235173,
      "grad_norm": 2.2122108936309814,
      "learning_rate": 9.986559460940854e-05,
      "loss": 0.1587,
      "step": 675
    },
    {
      "epoch": 0.02733328616050219,
      "grad_norm": 9.302861213684082,
      "learning_rate": 9.986512691635066e-05,
      "loss": 0.2219,
      "step": 676
    },
    {
      "epoch": 0.027373720015769203,
      "grad_norm": 10.460810661315918,
      "learning_rate": 9.986465841208412e-05,
      "loss": 0.3179,
      "step": 677
    },
    {
      "epoch": 0.02741415387103622,
      "grad_norm": 3.9048337936401367,
      "learning_rate": 9.986418909661652e-05,
      "loss": 0.2441,
      "step": 678
    },
    {
      "epoch": 0.027454587726303235,
      "grad_norm": 3.0725464820861816,
      "learning_rate": 9.98637189699555e-05,
      "loss": 0.1441,
      "step": 679
    },
    {
      "epoch": 0.027495021581570248,
      "grad_norm": 2.9670376777648926,
      "learning_rate": 9.986324803210872e-05,
      "loss": 0.2333,
      "step": 680
    },
    {
      "epoch": 0.027535455436837265,
      "grad_norm": 3.0021870136260986,
      "learning_rate": 9.986277628308382e-05,
      "loss": 0.1382,
      "step": 681
    },
    {
      "epoch": 0.027575889292104278,
      "grad_norm": 5.315797328948975,
      "learning_rate": 9.98623037228885e-05,
      "loss": 0.168,
      "step": 682
    },
    {
      "epoch": 0.027616323147371294,
      "grad_norm": 7.026374340057373,
      "learning_rate": 9.986183035153044e-05,
      "loss": 0.1804,
      "step": 683
    },
    {
      "epoch": 0.02765675700263831,
      "grad_norm": 5.097654819488525,
      "learning_rate": 9.986135616901733e-05,
      "loss": 0.2533,
      "step": 684
    },
    {
      "epoch": 0.027697190857905323,
      "grad_norm": 4.727205276489258,
      "learning_rate": 9.986088117535688e-05,
      "loss": 0.1448,
      "step": 685
    },
    {
      "epoch": 0.02773762471317234,
      "grad_norm": 3.388516426086426,
      "learning_rate": 9.986040537055685e-05,
      "loss": 0.2177,
      "step": 686
    },
    {
      "epoch": 0.027778058568439356,
      "grad_norm": 5.883607864379883,
      "learning_rate": 9.985992875462493e-05,
      "loss": 0.2412,
      "step": 687
    },
    {
      "epoch": 0.02781849242370637,
      "grad_norm": 3.4712960720062256,
      "learning_rate": 9.985945132756893e-05,
      "loss": 0.2327,
      "step": 688
    },
    {
      "epoch": 0.027858926278973385,
      "grad_norm": 3.2009572982788086,
      "learning_rate": 9.985897308939656e-05,
      "loss": 0.0819,
      "step": 689
    },
    {
      "epoch": 0.027899360134240398,
      "grad_norm": 7.193490982055664,
      "learning_rate": 9.985849404011565e-05,
      "loss": 0.2269,
      "step": 690
    },
    {
      "epoch": 0.027939793989507414,
      "grad_norm": 8.777181625366211,
      "learning_rate": 9.985801417973394e-05,
      "loss": 0.2405,
      "step": 691
    },
    {
      "epoch": 0.02798022784477443,
      "grad_norm": 2.643279552459717,
      "learning_rate": 9.98575335082593e-05,
      "loss": 0.2221,
      "step": 692
    },
    {
      "epoch": 0.028020661700041444,
      "grad_norm": 5.934566974639893,
      "learning_rate": 9.98570520256995e-05,
      "loss": 0.1866,
      "step": 693
    },
    {
      "epoch": 0.02806109555530846,
      "grad_norm": 9.89050579071045,
      "learning_rate": 9.98565697320624e-05,
      "loss": 0.2569,
      "step": 694
    },
    {
      "epoch": 0.028101529410575476,
      "grad_norm": 2.067317008972168,
      "learning_rate": 9.985608662735584e-05,
      "loss": 0.0593,
      "step": 695
    },
    {
      "epoch": 0.02814196326584249,
      "grad_norm": 7.369929790496826,
      "learning_rate": 9.985560271158767e-05,
      "loss": 0.1809,
      "step": 696
    },
    {
      "epoch": 0.028182397121109506,
      "grad_norm": 2.843261957168579,
      "learning_rate": 9.985511798476577e-05,
      "loss": 0.1711,
      "step": 697
    },
    {
      "epoch": 0.02822283097637652,
      "grad_norm": 5.012716293334961,
      "learning_rate": 9.985463244689802e-05,
      "loss": 0.1224,
      "step": 698
    },
    {
      "epoch": 0.028263264831643535,
      "grad_norm": 4.1079230308532715,
      "learning_rate": 9.98541460979923e-05,
      "loss": 0.1478,
      "step": 699
    },
    {
      "epoch": 0.02830369868691055,
      "grad_norm": 4.032195091247559,
      "learning_rate": 9.985365893805658e-05,
      "loss": 0.1966,
      "step": 700
    },
    {
      "epoch": 0.028344132542177564,
      "grad_norm": 14.765571594238281,
      "learning_rate": 9.985317096709874e-05,
      "loss": 0.2721,
      "step": 701
    },
    {
      "epoch": 0.02838456639744458,
      "grad_norm": 14.63611125946045,
      "learning_rate": 9.985268218512671e-05,
      "loss": 0.3835,
      "step": 702
    },
    {
      "epoch": 0.028425000252711597,
      "grad_norm": 8.562143325805664,
      "learning_rate": 9.985219259214848e-05,
      "loss": 0.2595,
      "step": 703
    },
    {
      "epoch": 0.02846543410797861,
      "grad_norm": 7.470378875732422,
      "learning_rate": 9.985170218817196e-05,
      "loss": 0.2903,
      "step": 704
    },
    {
      "epoch": 0.028505867963245626,
      "grad_norm": 9.900849342346191,
      "learning_rate": 9.98512109732052e-05,
      "loss": 0.2921,
      "step": 705
    },
    {
      "epoch": 0.02854630181851264,
      "grad_norm": 9.876456260681152,
      "learning_rate": 9.985071894725614e-05,
      "loss": 0.3102,
      "step": 706
    },
    {
      "epoch": 0.028586735673779656,
      "grad_norm": 5.808203220367432,
      "learning_rate": 9.985022611033278e-05,
      "loss": 0.1562,
      "step": 707
    },
    {
      "epoch": 0.028627169529046672,
      "grad_norm": 3.8940303325653076,
      "learning_rate": 9.984973246244318e-05,
      "loss": 0.1007,
      "step": 708
    },
    {
      "epoch": 0.028667603384313685,
      "grad_norm": 11.69302749633789,
      "learning_rate": 9.984923800359533e-05,
      "loss": 0.3488,
      "step": 709
    },
    {
      "epoch": 0.0287080372395807,
      "grad_norm": 8.050931930541992,
      "learning_rate": 9.98487427337973e-05,
      "loss": 0.335,
      "step": 710
    },
    {
      "epoch": 0.028748471094847718,
      "grad_norm": 5.279017448425293,
      "learning_rate": 9.984824665305713e-05,
      "loss": 0.1758,
      "step": 711
    },
    {
      "epoch": 0.02878890495011473,
      "grad_norm": 6.152420520782471,
      "learning_rate": 9.98477497613829e-05,
      "loss": 0.2743,
      "step": 712
    },
    {
      "epoch": 0.028829338805381747,
      "grad_norm": 4.773203372955322,
      "learning_rate": 9.98472520587827e-05,
      "loss": 0.1864,
      "step": 713
    },
    {
      "epoch": 0.02886977266064876,
      "grad_norm": 4.273560523986816,
      "learning_rate": 9.984675354526458e-05,
      "loss": 0.23,
      "step": 714
    },
    {
      "epoch": 0.028910206515915776,
      "grad_norm": 3.7935972213745117,
      "learning_rate": 9.984625422083672e-05,
      "loss": 0.1946,
      "step": 715
    },
    {
      "epoch": 0.028950640371182793,
      "grad_norm": 6.885613918304443,
      "learning_rate": 9.984575408550719e-05,
      "loss": 0.2122,
      "step": 716
    },
    {
      "epoch": 0.028991074226449805,
      "grad_norm": 3.47890043258667,
      "learning_rate": 9.984525313928416e-05,
      "loss": 0.1133,
      "step": 717
    },
    {
      "epoch": 0.029031508081716822,
      "grad_norm": 4.814701080322266,
      "learning_rate": 9.984475138217575e-05,
      "loss": 0.179,
      "step": 718
    },
    {
      "epoch": 0.029071941936983838,
      "grad_norm": 3.4611752033233643,
      "learning_rate": 9.984424881419015e-05,
      "loss": 0.1593,
      "step": 719
    },
    {
      "epoch": 0.02911237579225085,
      "grad_norm": 7.355879306793213,
      "learning_rate": 9.984374543533553e-05,
      "loss": 0.1866,
      "step": 720
    },
    {
      "epoch": 0.029152809647517867,
      "grad_norm": 7.713857650756836,
      "learning_rate": 9.984324124562005e-05,
      "loss": 0.1545,
      "step": 721
    },
    {
      "epoch": 0.02919324350278488,
      "grad_norm": 5.917805194854736,
      "learning_rate": 9.984273624505195e-05,
      "loss": 0.167,
      "step": 722
    },
    {
      "epoch": 0.029233677358051897,
      "grad_norm": 3.051311731338501,
      "learning_rate": 9.984223043363943e-05,
      "loss": 0.1007,
      "step": 723
    },
    {
      "epoch": 0.029274111213318913,
      "grad_norm": 5.5038652420043945,
      "learning_rate": 9.984172381139072e-05,
      "loss": 0.1571,
      "step": 724
    },
    {
      "epoch": 0.029314545068585926,
      "grad_norm": 14.65781307220459,
      "learning_rate": 9.984121637831405e-05,
      "loss": 0.2997,
      "step": 725
    },
    {
      "epoch": 0.029354978923852942,
      "grad_norm": 14.468057632446289,
      "learning_rate": 9.984070813441769e-05,
      "loss": 0.3262,
      "step": 726
    },
    {
      "epoch": 0.02939541277911996,
      "grad_norm": 10.01978588104248,
      "learning_rate": 9.984019907970989e-05,
      "loss": 0.3252,
      "step": 727
    },
    {
      "epoch": 0.02943584663438697,
      "grad_norm": 9.07807445526123,
      "learning_rate": 9.983968921419897e-05,
      "loss": 0.2489,
      "step": 728
    },
    {
      "epoch": 0.029476280489653988,
      "grad_norm": 11.9734525680542,
      "learning_rate": 9.98391785378932e-05,
      "loss": 0.3385,
      "step": 729
    },
    {
      "epoch": 0.029516714344921,
      "grad_norm": 5.344473838806152,
      "learning_rate": 9.983866705080087e-05,
      "loss": 0.2174,
      "step": 730
    },
    {
      "epoch": 0.029557148200188017,
      "grad_norm": 5.054398059844971,
      "learning_rate": 9.983815475293032e-05,
      "loss": 0.1554,
      "step": 731
    },
    {
      "epoch": 0.029597582055455034,
      "grad_norm": 8.386213302612305,
      "learning_rate": 9.983764164428988e-05,
      "loss": 0.1593,
      "step": 732
    },
    {
      "epoch": 0.029638015910722047,
      "grad_norm": 5.48167085647583,
      "learning_rate": 9.983712772488792e-05,
      "loss": 0.2128,
      "step": 733
    },
    {
      "epoch": 0.029678449765989063,
      "grad_norm": 5.021555423736572,
      "learning_rate": 9.983661299473274e-05,
      "loss": 0.1427,
      "step": 734
    },
    {
      "epoch": 0.02971888362125608,
      "grad_norm": 7.41727352142334,
      "learning_rate": 9.983609745383279e-05,
      "loss": 0.2523,
      "step": 735
    },
    {
      "epoch": 0.029759317476523092,
      "grad_norm": 5.963095188140869,
      "learning_rate": 9.983558110219641e-05,
      "loss": 0.1643,
      "step": 736
    },
    {
      "epoch": 0.02979975133179011,
      "grad_norm": 4.523470401763916,
      "learning_rate": 9.983506393983201e-05,
      "loss": 0.089,
      "step": 737
    },
    {
      "epoch": 0.02984018518705712,
      "grad_norm": 7.40480375289917,
      "learning_rate": 9.9834545966748e-05,
      "loss": 0.3762,
      "step": 738
    },
    {
      "epoch": 0.029880619042324138,
      "grad_norm": 2.1949849128723145,
      "learning_rate": 9.983402718295281e-05,
      "loss": 0.1425,
      "step": 739
    },
    {
      "epoch": 0.029921052897591154,
      "grad_norm": 6.313375473022461,
      "learning_rate": 9.983350758845487e-05,
      "loss": 0.2581,
      "step": 740
    },
    {
      "epoch": 0.029961486752858167,
      "grad_norm": 3.194705009460449,
      "learning_rate": 9.983298718326265e-05,
      "loss": 0.1905,
      "step": 741
    },
    {
      "epoch": 0.030001920608125184,
      "grad_norm": 11.069114685058594,
      "learning_rate": 9.983246596738462e-05,
      "loss": 0.2945,
      "step": 742
    },
    {
      "epoch": 0.0300423544633922,
      "grad_norm": 7.233796119689941,
      "learning_rate": 9.983194394082924e-05,
      "loss": 0.1795,
      "step": 743
    },
    {
      "epoch": 0.030082788318659213,
      "grad_norm": 5.389409065246582,
      "learning_rate": 9.983142110360502e-05,
      "loss": 0.2728,
      "step": 744
    },
    {
      "epoch": 0.03012322217392623,
      "grad_norm": 4.685056209564209,
      "learning_rate": 9.983089745572043e-05,
      "loss": 0.2514,
      "step": 745
    },
    {
      "epoch": 0.030163656029193242,
      "grad_norm": 6.272307395935059,
      "learning_rate": 9.983037299718403e-05,
      "loss": 0.2123,
      "step": 746
    },
    {
      "epoch": 0.03020408988446026,
      "grad_norm": 5.039308071136475,
      "learning_rate": 9.982984772800434e-05,
      "loss": 0.2046,
      "step": 747
    },
    {
      "epoch": 0.030244523739727275,
      "grad_norm": 6.528295516967773,
      "learning_rate": 9.98293216481899e-05,
      "loss": 0.2321,
      "step": 748
    },
    {
      "epoch": 0.030284957594994288,
      "grad_norm": 4.320619106292725,
      "learning_rate": 9.982879475774926e-05,
      "loss": 0.1807,
      "step": 749
    },
    {
      "epoch": 0.030325391450261304,
      "grad_norm": 1.8070778846740723,
      "learning_rate": 9.9828267056691e-05,
      "loss": 0.1588,
      "step": 750
    },
    {
      "epoch": 0.03036582530552832,
      "grad_norm": 7.710047721862793,
      "learning_rate": 9.982773854502371e-05,
      "loss": 0.1633,
      "step": 751
    },
    {
      "epoch": 0.030406259160795333,
      "grad_norm": 9.199658393859863,
      "learning_rate": 9.982720922275599e-05,
      "loss": 0.1813,
      "step": 752
    },
    {
      "epoch": 0.03044669301606235,
      "grad_norm": 4.615907669067383,
      "learning_rate": 9.982667908989643e-05,
      "loss": 0.2575,
      "step": 753
    },
    {
      "epoch": 0.030487126871329363,
      "grad_norm": 5.2957868576049805,
      "learning_rate": 9.982614814645368e-05,
      "loss": 0.2068,
      "step": 754
    },
    {
      "epoch": 0.03052756072659638,
      "grad_norm": 10.10409927368164,
      "learning_rate": 9.982561639243637e-05,
      "loss": 0.2624,
      "step": 755
    },
    {
      "epoch": 0.030567994581863395,
      "grad_norm": 4.214273929595947,
      "learning_rate": 9.982508382785313e-05,
      "loss": 0.1369,
      "step": 756
    },
    {
      "epoch": 0.03060842843713041,
      "grad_norm": 6.632080554962158,
      "learning_rate": 9.982455045271267e-05,
      "loss": 0.2308,
      "step": 757
    },
    {
      "epoch": 0.030648862292397425,
      "grad_norm": 3.2878658771514893,
      "learning_rate": 9.98240162670236e-05,
      "loss": 0.092,
      "step": 758
    },
    {
      "epoch": 0.03068929614766444,
      "grad_norm": 2.833817481994629,
      "learning_rate": 9.982348127079467e-05,
      "loss": 0.2506,
      "step": 759
    },
    {
      "epoch": 0.030729730002931454,
      "grad_norm": 2.717789649963379,
      "learning_rate": 9.982294546403457e-05,
      "loss": 0.0749,
      "step": 760
    },
    {
      "epoch": 0.03077016385819847,
      "grad_norm": 8.115143775939941,
      "learning_rate": 9.982240884675198e-05,
      "loss": 0.2931,
      "step": 761
    },
    {
      "epoch": 0.030810597713465483,
      "grad_norm": 8.07234001159668,
      "learning_rate": 9.982187141895568e-05,
      "loss": 0.2885,
      "step": 762
    },
    {
      "epoch": 0.0308510315687325,
      "grad_norm": 9.224542617797852,
      "learning_rate": 9.982133318065439e-05,
      "loss": 0.3375,
      "step": 763
    },
    {
      "epoch": 0.030891465423999516,
      "grad_norm": 10.921764373779297,
      "learning_rate": 9.982079413185685e-05,
      "loss": 0.3216,
      "step": 764
    },
    {
      "epoch": 0.03093189927926653,
      "grad_norm": 8.849933624267578,
      "learning_rate": 9.982025427257185e-05,
      "loss": 0.2519,
      "step": 765
    },
    {
      "epoch": 0.030972333134533545,
      "grad_norm": 3.9318888187408447,
      "learning_rate": 9.981971360280818e-05,
      "loss": 0.1705,
      "step": 766
    },
    {
      "epoch": 0.03101276698980056,
      "grad_norm": 3.7941842079162598,
      "learning_rate": 9.981917212257462e-05,
      "loss": 0.1694,
      "step": 767
    },
    {
      "epoch": 0.031053200845067574,
      "grad_norm": 10.645840644836426,
      "learning_rate": 9.981862983187997e-05,
      "loss": 0.3064,
      "step": 768
    },
    {
      "epoch": 0.03109363470033459,
      "grad_norm": 5.097426414489746,
      "learning_rate": 9.981808673073307e-05,
      "loss": 0.1297,
      "step": 769
    },
    {
      "epoch": 0.031134068555601604,
      "grad_norm": 3.904343366622925,
      "learning_rate": 9.981754281914275e-05,
      "loss": 0.2394,
      "step": 770
    },
    {
      "epoch": 0.03117450241086862,
      "grad_norm": 9.943769454956055,
      "learning_rate": 9.981699809711786e-05,
      "loss": 0.2139,
      "step": 771
    },
    {
      "epoch": 0.031214936266135637,
      "grad_norm": 4.494277477264404,
      "learning_rate": 9.981645256466727e-05,
      "loss": 0.1422,
      "step": 772
    },
    {
      "epoch": 0.03125537012140265,
      "grad_norm": 6.064607620239258,
      "learning_rate": 9.981590622179982e-05,
      "loss": 0.231,
      "step": 773
    },
    {
      "epoch": 0.031295803976669666,
      "grad_norm": 5.5445876121521,
      "learning_rate": 9.981535906852444e-05,
      "loss": 0.1781,
      "step": 774
    },
    {
      "epoch": 0.03133623783193668,
      "grad_norm": 13.826998710632324,
      "learning_rate": 9.981481110485001e-05,
      "loss": 0.3502,
      "step": 775
    },
    {
      "epoch": 0.0313766716872037,
      "grad_norm": 8.732311248779297,
      "learning_rate": 9.981426233078546e-05,
      "loss": 0.2149,
      "step": 776
    },
    {
      "epoch": 0.03141710554247071,
      "grad_norm": 5.431211948394775,
      "learning_rate": 9.981371274633968e-05,
      "loss": 0.1487,
      "step": 777
    },
    {
      "epoch": 0.031457539397737724,
      "grad_norm": 4.0659966468811035,
      "learning_rate": 9.981316235152165e-05,
      "loss": 0.0611,
      "step": 778
    },
    {
      "epoch": 0.03149797325300474,
      "grad_norm": 6.683130741119385,
      "learning_rate": 9.981261114634031e-05,
      "loss": 0.2351,
      "step": 779
    },
    {
      "epoch": 0.03153840710827176,
      "grad_norm": 6.116979122161865,
      "learning_rate": 9.981205913080462e-05,
      "loss": 0.2752,
      "step": 780
    },
    {
      "epoch": 0.03157884096353877,
      "grad_norm": 6.24696683883667,
      "learning_rate": 9.981150630492357e-05,
      "loss": 0.1675,
      "step": 781
    },
    {
      "epoch": 0.03161927481880578,
      "grad_norm": 3.6955323219299316,
      "learning_rate": 9.981095266870616e-05,
      "loss": 0.1043,
      "step": 782
    },
    {
      "epoch": 0.0316597086740728,
      "grad_norm": 9.421211242675781,
      "learning_rate": 9.981039822216138e-05,
      "loss": 0.3138,
      "step": 783
    },
    {
      "epoch": 0.031700142529339816,
      "grad_norm": 7.978635787963867,
      "learning_rate": 9.980984296529825e-05,
      "loss": 0.5177,
      "step": 784
    },
    {
      "epoch": 0.03174057638460683,
      "grad_norm": 10.833230972290039,
      "learning_rate": 9.980928689812581e-05,
      "loss": 0.2596,
      "step": 785
    },
    {
      "epoch": 0.03178101023987385,
      "grad_norm": 5.850692272186279,
      "learning_rate": 9.980873002065312e-05,
      "loss": 0.2647,
      "step": 786
    },
    {
      "epoch": 0.031821444095140865,
      "grad_norm": 8.674549102783203,
      "learning_rate": 9.980817233288921e-05,
      "loss": 0.1375,
      "step": 787
    },
    {
      "epoch": 0.031861877950407874,
      "grad_norm": 6.093148708343506,
      "learning_rate": 9.980761383484319e-05,
      "loss": 0.31,
      "step": 788
    },
    {
      "epoch": 0.03190231180567489,
      "grad_norm": 4.2311482429504395,
      "learning_rate": 9.98070545265241e-05,
      "loss": 0.1607,
      "step": 789
    },
    {
      "epoch": 0.03194274566094191,
      "grad_norm": 8.54882526397705,
      "learning_rate": 9.980649440794107e-05,
      "loss": 0.2581,
      "step": 790
    },
    {
      "epoch": 0.03198317951620892,
      "grad_norm": 7.2482218742370605,
      "learning_rate": 9.98059334791032e-05,
      "loss": 0.1883,
      "step": 791
    },
    {
      "epoch": 0.03202361337147594,
      "grad_norm": 2.746206760406494,
      "learning_rate": 9.980537174001962e-05,
      "loss": 0.1866,
      "step": 792
    },
    {
      "epoch": 0.03206404722674295,
      "grad_norm": 12.297630310058594,
      "learning_rate": 9.980480919069948e-05,
      "loss": 0.3103,
      "step": 793
    },
    {
      "epoch": 0.032104481082009965,
      "grad_norm": 3.2502999305725098,
      "learning_rate": 9.980424583115191e-05,
      "loss": 0.151,
      "step": 794
    },
    {
      "epoch": 0.03214491493727698,
      "grad_norm": 6.139448165893555,
      "learning_rate": 9.980368166138609e-05,
      "loss": 0.2244,
      "step": 795
    },
    {
      "epoch": 0.032185348792544,
      "grad_norm": 3.870300769805908,
      "learning_rate": 9.980311668141119e-05,
      "loss": 0.2243,
      "step": 796
    },
    {
      "epoch": 0.032225782647811015,
      "grad_norm": 4.218263149261475,
      "learning_rate": 9.98025508912364e-05,
      "loss": 0.0906,
      "step": 797
    },
    {
      "epoch": 0.032266216503078024,
      "grad_norm": 6.0168232917785645,
      "learning_rate": 9.980198429087092e-05,
      "loss": 0.2044,
      "step": 798
    },
    {
      "epoch": 0.03230665035834504,
      "grad_norm": 2.4285359382629395,
      "learning_rate": 9.980141688032399e-05,
      "loss": 0.1382,
      "step": 799
    },
    {
      "epoch": 0.03234708421361206,
      "grad_norm": 6.8135833740234375,
      "learning_rate": 9.980084865960482e-05,
      "loss": 0.2804,
      "step": 800
    },
    {
      "epoch": 0.03238751806887907,
      "grad_norm": 14.248740196228027,
      "learning_rate": 9.980027962872266e-05,
      "loss": 0.3435,
      "step": 801
    },
    {
      "epoch": 0.03242795192414609,
      "grad_norm": 10.927685737609863,
      "learning_rate": 9.979970978768677e-05,
      "loss": 0.1859,
      "step": 802
    },
    {
      "epoch": 0.032468385779413106,
      "grad_norm": 15.27355670928955,
      "learning_rate": 9.97991391365064e-05,
      "loss": 0.3279,
      "step": 803
    },
    {
      "epoch": 0.032508819634680115,
      "grad_norm": 10.57928466796875,
      "learning_rate": 9.979856767519087e-05,
      "loss": 0.151,
      "step": 804
    },
    {
      "epoch": 0.03254925348994713,
      "grad_norm": 5.602120399475098,
      "learning_rate": 9.979799540374945e-05,
      "loss": 0.1484,
      "step": 805
    },
    {
      "epoch": 0.03258968734521415,
      "grad_norm": 3.0610568523406982,
      "learning_rate": 9.979742232219145e-05,
      "loss": 0.1665,
      "step": 806
    },
    {
      "epoch": 0.032630121200481164,
      "grad_norm": 5.5097126960754395,
      "learning_rate": 9.979684843052621e-05,
      "loss": 0.19,
      "step": 807
    },
    {
      "epoch": 0.03267055505574818,
      "grad_norm": 3.7976698875427246,
      "learning_rate": 9.979627372876306e-05,
      "loss": 0.1978,
      "step": 808
    },
    {
      "epoch": 0.03271098891101519,
      "grad_norm": 4.449186325073242,
      "learning_rate": 9.979569821691133e-05,
      "loss": 0.1683,
      "step": 809
    },
    {
      "epoch": 0.03275142276628221,
      "grad_norm": 6.5003509521484375,
      "learning_rate": 9.97951218949804e-05,
      "loss": 0.1226,
      "step": 810
    },
    {
      "epoch": 0.03279185662154922,
      "grad_norm": 4.912683963775635,
      "learning_rate": 9.979454476297964e-05,
      "loss": 0.2334,
      "step": 811
    },
    {
      "epoch": 0.03283229047681624,
      "grad_norm": 10.277607917785645,
      "learning_rate": 9.979396682091846e-05,
      "loss": 0.2913,
      "step": 812
    },
    {
      "epoch": 0.032872724332083256,
      "grad_norm": 12.071046829223633,
      "learning_rate": 9.979338806880623e-05,
      "loss": 0.2862,
      "step": 813
    },
    {
      "epoch": 0.032913158187350265,
      "grad_norm": 4.660379409790039,
      "learning_rate": 9.979280850665238e-05,
      "loss": 0.1843,
      "step": 814
    },
    {
      "epoch": 0.03295359204261728,
      "grad_norm": 11.3518705368042,
      "learning_rate": 9.979222813446634e-05,
      "loss": 0.3813,
      "step": 815
    },
    {
      "epoch": 0.0329940258978843,
      "grad_norm": 3.658555507659912,
      "learning_rate": 9.979164695225753e-05,
      "loss": 0.2799,
      "step": 816
    },
    {
      "epoch": 0.033034459753151314,
      "grad_norm": 4.347692489624023,
      "learning_rate": 9.979106496003545e-05,
      "loss": 0.1628,
      "step": 817
    },
    {
      "epoch": 0.03307489360841833,
      "grad_norm": 6.3739800453186035,
      "learning_rate": 9.979048215780953e-05,
      "loss": 0.1831,
      "step": 818
    },
    {
      "epoch": 0.03311532746368535,
      "grad_norm": 4.444432735443115,
      "learning_rate": 9.978989854558927e-05,
      "loss": 0.1072,
      "step": 819
    },
    {
      "epoch": 0.033155761318952356,
      "grad_norm": 9.649750709533691,
      "learning_rate": 9.978931412338415e-05,
      "loss": 0.3173,
      "step": 820
    },
    {
      "epoch": 0.03319619517421937,
      "grad_norm": 7.612451076507568,
      "learning_rate": 9.978872889120367e-05,
      "loss": 0.2706,
      "step": 821
    },
    {
      "epoch": 0.03323662902948639,
      "grad_norm": 11.028070449829102,
      "learning_rate": 9.978814284905739e-05,
      "loss": 0.2533,
      "step": 822
    },
    {
      "epoch": 0.033277062884753406,
      "grad_norm": 3.2694690227508545,
      "learning_rate": 9.97875559969548e-05,
      "loss": 0.1932,
      "step": 823
    },
    {
      "epoch": 0.03331749674002042,
      "grad_norm": 7.95664119720459,
      "learning_rate": 9.978696833490547e-05,
      "loss": 0.2822,
      "step": 824
    },
    {
      "epoch": 0.03335793059528743,
      "grad_norm": 5.8272624015808105,
      "learning_rate": 9.978637986291896e-05,
      "loss": 0.2509,
      "step": 825
    },
    {
      "epoch": 0.03339836445055445,
      "grad_norm": 6.905600547790527,
      "learning_rate": 9.978579058100483e-05,
      "loss": 0.2413,
      "step": 826
    },
    {
      "epoch": 0.033438798305821464,
      "grad_norm": 3.813549518585205,
      "learning_rate": 9.978520048917269e-05,
      "loss": 0.1334,
      "step": 827
    },
    {
      "epoch": 0.03347923216108848,
      "grad_norm": 7.520639419555664,
      "learning_rate": 9.978460958743212e-05,
      "loss": 0.2249,
      "step": 828
    },
    {
      "epoch": 0.0335196660163555,
      "grad_norm": 4.187660217285156,
      "learning_rate": 9.978401787579273e-05,
      "loss": 0.1148,
      "step": 829
    },
    {
      "epoch": 0.033560099871622506,
      "grad_norm": 5.920393943786621,
      "learning_rate": 9.978342535426416e-05,
      "loss": 0.2144,
      "step": 830
    },
    {
      "epoch": 0.03360053372688952,
      "grad_norm": 7.997655868530273,
      "learning_rate": 9.978283202285605e-05,
      "loss": 0.2753,
      "step": 831
    },
    {
      "epoch": 0.03364096758215654,
      "grad_norm": 5.6766533851623535,
      "learning_rate": 9.978223788157803e-05,
      "loss": 0.3013,
      "step": 832
    },
    {
      "epoch": 0.033681401437423555,
      "grad_norm": 4.9550676345825195,
      "learning_rate": 9.978164293043979e-05,
      "loss": 0.1762,
      "step": 833
    },
    {
      "epoch": 0.03372183529269057,
      "grad_norm": 4.913577079772949,
      "learning_rate": 9.978104716945099e-05,
      "loss": 0.2557,
      "step": 834
    },
    {
      "epoch": 0.03376226914795759,
      "grad_norm": 3.7246878147125244,
      "learning_rate": 9.978045059862133e-05,
      "loss": 0.1948,
      "step": 835
    },
    {
      "epoch": 0.0338027030032246,
      "grad_norm": 4.334644794464111,
      "learning_rate": 9.977985321796053e-05,
      "loss": 0.1677,
      "step": 836
    },
    {
      "epoch": 0.033843136858491614,
      "grad_norm": 7.720291614532471,
      "learning_rate": 9.97792550274783e-05,
      "loss": 0.1762,
      "step": 837
    },
    {
      "epoch": 0.03388357071375863,
      "grad_norm": 7.63993501663208,
      "learning_rate": 9.977865602718435e-05,
      "loss": 0.2857,
      "step": 838
    },
    {
      "epoch": 0.03392400456902565,
      "grad_norm": 2.1309170722961426,
      "learning_rate": 9.977805621708844e-05,
      "loss": 0.1518,
      "step": 839
    },
    {
      "epoch": 0.03396443842429266,
      "grad_norm": 6.689241409301758,
      "learning_rate": 9.977745559720034e-05,
      "loss": 0.3492,
      "step": 840
    },
    {
      "epoch": 0.03400487227955967,
      "grad_norm": 3.2680318355560303,
      "learning_rate": 9.97768541675298e-05,
      "loss": 0.218,
      "step": 841
    },
    {
      "epoch": 0.03404530613482669,
      "grad_norm": 8.970392227172852,
      "learning_rate": 9.977625192808662e-05,
      "loss": 0.1784,
      "step": 842
    },
    {
      "epoch": 0.034085739990093705,
      "grad_norm": 5.828888893127441,
      "learning_rate": 9.977564887888059e-05,
      "loss": 0.2155,
      "step": 843
    },
    {
      "epoch": 0.03412617384536072,
      "grad_norm": 6.663506507873535,
      "learning_rate": 9.97750450199215e-05,
      "loss": 0.1923,
      "step": 844
    },
    {
      "epoch": 0.03416660770062774,
      "grad_norm": 3.342913866043091,
      "learning_rate": 9.977444035121922e-05,
      "loss": 0.0847,
      "step": 845
    },
    {
      "epoch": 0.03420704155589475,
      "grad_norm": 6.70332145690918,
      "learning_rate": 9.977383487278356e-05,
      "loss": 0.288,
      "step": 846
    },
    {
      "epoch": 0.034247475411161764,
      "grad_norm": 6.838769912719727,
      "learning_rate": 9.977322858462437e-05,
      "loss": 0.1948,
      "step": 847
    },
    {
      "epoch": 0.03428790926642878,
      "grad_norm": 4.318086624145508,
      "learning_rate": 9.97726214867515e-05,
      "loss": 0.2274,
      "step": 848
    },
    {
      "epoch": 0.0343283431216958,
      "grad_norm": 5.76066780090332,
      "learning_rate": 9.977201357917485e-05,
      "loss": 0.1973,
      "step": 849
    },
    {
      "epoch": 0.03436877697696281,
      "grad_norm": 6.076119422912598,
      "learning_rate": 9.97714048619043e-05,
      "loss": 0.1307,
      "step": 850
    },
    {
      "epoch": 0.03440921083222983,
      "grad_norm": 15.187824249267578,
      "learning_rate": 9.977079533494976e-05,
      "loss": 0.4078,
      "step": 851
    },
    {
      "epoch": 0.03444964468749684,
      "grad_norm": 9.927777290344238,
      "learning_rate": 9.977018499832113e-05,
      "loss": 0.2832,
      "step": 852
    },
    {
      "epoch": 0.034490078542763855,
      "grad_norm": 14.828780174255371,
      "learning_rate": 9.976957385202834e-05,
      "loss": 0.3256,
      "step": 853
    },
    {
      "epoch": 0.03453051239803087,
      "grad_norm": 7.831463813781738,
      "learning_rate": 9.976896189608137e-05,
      "loss": 0.169,
      "step": 854
    },
    {
      "epoch": 0.03457094625329789,
      "grad_norm": 4.30358362197876,
      "learning_rate": 9.976834913049011e-05,
      "loss": 0.1868,
      "step": 855
    },
    {
      "epoch": 0.034611380108564904,
      "grad_norm": 7.565115928649902,
      "learning_rate": 9.976773555526459e-05,
      "loss": 0.1694,
      "step": 856
    },
    {
      "epoch": 0.034651813963831914,
      "grad_norm": 4.121387958526611,
      "learning_rate": 9.976712117041476e-05,
      "loss": 0.177,
      "step": 857
    },
    {
      "epoch": 0.03469224781909893,
      "grad_norm": 6.83624267578125,
      "learning_rate": 9.976650597595061e-05,
      "loss": 0.3447,
      "step": 858
    },
    {
      "epoch": 0.034732681674365946,
      "grad_norm": 5.594140529632568,
      "learning_rate": 9.976588997188216e-05,
      "loss": 0.3042,
      "step": 859
    },
    {
      "epoch": 0.03477311552963296,
      "grad_norm": 19.720571517944336,
      "learning_rate": 9.976527315821942e-05,
      "loss": 0.5235,
      "step": 860
    },
    {
      "epoch": 0.03481354938489998,
      "grad_norm": 11.797995567321777,
      "learning_rate": 9.976465553497246e-05,
      "loss": 0.3712,
      "step": 861
    },
    {
      "epoch": 0.03485398324016699,
      "grad_norm": 6.199066638946533,
      "learning_rate": 9.976403710215128e-05,
      "loss": 0.1635,
      "step": 862
    },
    {
      "epoch": 0.034894417095434005,
      "grad_norm": 5.9693756103515625,
      "learning_rate": 9.976341785976597e-05,
      "loss": 0.1679,
      "step": 863
    },
    {
      "epoch": 0.03493485095070102,
      "grad_norm": 2.7871015071868896,
      "learning_rate": 9.97627978078266e-05,
      "loss": 0.189,
      "step": 864
    },
    {
      "epoch": 0.03497528480596804,
      "grad_norm": 7.259538173675537,
      "learning_rate": 9.976217694634325e-05,
      "loss": 0.2057,
      "step": 865
    },
    {
      "epoch": 0.035015718661235054,
      "grad_norm": 3.852043390274048,
      "learning_rate": 9.976155527532603e-05,
      "loss": 0.208,
      "step": 866
    },
    {
      "epoch": 0.03505615251650207,
      "grad_norm": 3.5917744636535645,
      "learning_rate": 9.976093279478502e-05,
      "loss": 0.1247,
      "step": 867
    },
    {
      "epoch": 0.03509658637176908,
      "grad_norm": 4.931051254272461,
      "learning_rate": 9.97603095047304e-05,
      "loss": 0.2643,
      "step": 868
    },
    {
      "epoch": 0.035137020227036096,
      "grad_norm": 10.863868713378906,
      "learning_rate": 9.975968540517228e-05,
      "loss": 0.212,
      "step": 869
    },
    {
      "epoch": 0.03517745408230311,
      "grad_norm": 10.337514877319336,
      "learning_rate": 9.97590604961208e-05,
      "loss": 0.2274,
      "step": 870
    },
    {
      "epoch": 0.03521788793757013,
      "grad_norm": 8.256402015686035,
      "learning_rate": 9.975843477758616e-05,
      "loss": 0.1994,
      "step": 871
    },
    {
      "epoch": 0.035258321792837145,
      "grad_norm": 6.377293109893799,
      "learning_rate": 9.975780824957851e-05,
      "loss": 0.1315,
      "step": 872
    },
    {
      "epoch": 0.035298755648104155,
      "grad_norm": 8.764927864074707,
      "learning_rate": 9.975718091210806e-05,
      "loss": 0.1863,
      "step": 873
    },
    {
      "epoch": 0.03533918950337117,
      "grad_norm": 2.7882156372070312,
      "learning_rate": 9.975655276518501e-05,
      "loss": 0.1758,
      "step": 874
    },
    {
      "epoch": 0.03537962335863819,
      "grad_norm": 2.7560551166534424,
      "learning_rate": 9.975592380881956e-05,
      "loss": 0.1917,
      "step": 875
    },
    {
      "epoch": 0.035420057213905204,
      "grad_norm": 12.475833892822266,
      "learning_rate": 9.975529404302197e-05,
      "loss": 0.3889,
      "step": 876
    },
    {
      "epoch": 0.03546049106917222,
      "grad_norm": 4.974170207977295,
      "learning_rate": 9.975466346780249e-05,
      "loss": 0.1876,
      "step": 877
    },
    {
      "epoch": 0.03550092492443923,
      "grad_norm": 10.734651565551758,
      "learning_rate": 9.975403208317133e-05,
      "loss": 0.314,
      "step": 878
    },
    {
      "epoch": 0.035541358779706246,
      "grad_norm": 4.398242950439453,
      "learning_rate": 9.975339988913883e-05,
      "loss": 0.1413,
      "step": 879
    },
    {
      "epoch": 0.03558179263497326,
      "grad_norm": 5.32480001449585,
      "learning_rate": 9.975276688571523e-05,
      "loss": 0.1327,
      "step": 880
    },
    {
      "epoch": 0.03562222649024028,
      "grad_norm": 5.457766056060791,
      "learning_rate": 9.975213307291082e-05,
      "loss": 0.1672,
      "step": 881
    },
    {
      "epoch": 0.035662660345507295,
      "grad_norm": 5.842166900634766,
      "learning_rate": 9.975149845073594e-05,
      "loss": 0.2905,
      "step": 882
    },
    {
      "epoch": 0.03570309420077431,
      "grad_norm": 5.4621806144714355,
      "learning_rate": 9.975086301920089e-05,
      "loss": 0.1564,
      "step": 883
    },
    {
      "epoch": 0.03574352805604132,
      "grad_norm": 6.922879219055176,
      "learning_rate": 9.975022677831603e-05,
      "loss": 0.3445,
      "step": 884
    },
    {
      "epoch": 0.03578396191130834,
      "grad_norm": 4.066812038421631,
      "learning_rate": 9.97495897280917e-05,
      "loss": 0.2774,
      "step": 885
    },
    {
      "epoch": 0.035824395766575354,
      "grad_norm": 3.8221025466918945,
      "learning_rate": 9.974895186853825e-05,
      "loss": 0.2706,
      "step": 886
    },
    {
      "epoch": 0.03586482962184237,
      "grad_norm": 6.305502891540527,
      "learning_rate": 9.97483131996661e-05,
      "loss": 0.1923,
      "step": 887
    },
    {
      "epoch": 0.035905263477109386,
      "grad_norm": 6.533466815948486,
      "learning_rate": 9.974767372148557e-05,
      "loss": 0.2343,
      "step": 888
    },
    {
      "epoch": 0.035945697332376396,
      "grad_norm": 2.0116050243377686,
      "learning_rate": 9.974703343400713e-05,
      "loss": 0.219,
      "step": 889
    },
    {
      "epoch": 0.03598613118764341,
      "grad_norm": 4.449499607086182,
      "learning_rate": 9.974639233724115e-05,
      "loss": 0.2303,
      "step": 890
    },
    {
      "epoch": 0.03602656504291043,
      "grad_norm": 3.246387481689453,
      "learning_rate": 9.974575043119809e-05,
      "loss": 0.1756,
      "step": 891
    },
    {
      "epoch": 0.036066998898177445,
      "grad_norm": 8.11133098602295,
      "learning_rate": 9.974510771588839e-05,
      "loss": 0.28,
      "step": 892
    },
    {
      "epoch": 0.03610743275344446,
      "grad_norm": 4.603014945983887,
      "learning_rate": 9.974446419132248e-05,
      "loss": 0.1582,
      "step": 893
    },
    {
      "epoch": 0.03614786660871147,
      "grad_norm": 4.219942569732666,
      "learning_rate": 9.974381985751084e-05,
      "loss": 0.0992,
      "step": 894
    },
    {
      "epoch": 0.03618830046397849,
      "grad_norm": 3.36643385887146,
      "learning_rate": 9.974317471446398e-05,
      "loss": 0.1385,
      "step": 895
    },
    {
      "epoch": 0.036228734319245504,
      "grad_norm": 5.307717800140381,
      "learning_rate": 9.974252876219235e-05,
      "loss": 0.2306,
      "step": 896
    },
    {
      "epoch": 0.03626916817451252,
      "grad_norm": 2.925379753112793,
      "learning_rate": 9.974188200070649e-05,
      "loss": 0.169,
      "step": 897
    },
    {
      "epoch": 0.036309602029779536,
      "grad_norm": 4.985349178314209,
      "learning_rate": 9.974123443001691e-05,
      "loss": 0.1698,
      "step": 898
    },
    {
      "epoch": 0.03635003588504655,
      "grad_norm": 6.789764881134033,
      "learning_rate": 9.974058605013414e-05,
      "loss": 0.1568,
      "step": 899
    },
    {
      "epoch": 0.03639046974031356,
      "grad_norm": 9.326662063598633,
      "learning_rate": 9.973993686106875e-05,
      "loss": 0.6595,
      "step": 900
    },
    {
      "epoch": 0.03643090359558058,
      "grad_norm": 3.322753667831421,
      "learning_rate": 9.973928686283129e-05,
      "loss": 0.1959,
      "step": 901
    },
    {
      "epoch": 0.036471337450847595,
      "grad_norm": 4.179361820220947,
      "learning_rate": 9.973863605543232e-05,
      "loss": 0.2801,
      "step": 902
    },
    {
      "epoch": 0.03651177130611461,
      "grad_norm": 4.589921474456787,
      "learning_rate": 9.973798443888243e-05,
      "loss": 0.1343,
      "step": 903
    },
    {
      "epoch": 0.03655220516138163,
      "grad_norm": 3.5719082355499268,
      "learning_rate": 9.973733201319225e-05,
      "loss": 0.1904,
      "step": 904
    },
    {
      "epoch": 0.03659263901664864,
      "grad_norm": 4.962691783905029,
      "learning_rate": 9.973667877837236e-05,
      "loss": 0.209,
      "step": 905
    },
    {
      "epoch": 0.03663307287191565,
      "grad_norm": 5.860391616821289,
      "learning_rate": 9.97360247344334e-05,
      "loss": 0.3143,
      "step": 906
    },
    {
      "epoch": 0.03667350672718267,
      "grad_norm": 5.504148960113525,
      "learning_rate": 9.9735369881386e-05,
      "loss": 0.1416,
      "step": 907
    },
    {
      "epoch": 0.036713940582449686,
      "grad_norm": 3.0422158241271973,
      "learning_rate": 9.973471421924083e-05,
      "loss": 0.1719,
      "step": 908
    },
    {
      "epoch": 0.0367543744377167,
      "grad_norm": 4.765471935272217,
      "learning_rate": 9.973405774800854e-05,
      "loss": 0.2518,
      "step": 909
    },
    {
      "epoch": 0.03679480829298371,
      "grad_norm": 2.1533994674682617,
      "learning_rate": 9.973340046769983e-05,
      "loss": 0.0726,
      "step": 910
    },
    {
      "epoch": 0.03683524214825073,
      "grad_norm": 3.029025077819824,
      "learning_rate": 9.973274237832537e-05,
      "loss": 0.1928,
      "step": 911
    },
    {
      "epoch": 0.036875676003517745,
      "grad_norm": 1.596284031867981,
      "learning_rate": 9.973208347989589e-05,
      "loss": 0.0419,
      "step": 912
    },
    {
      "epoch": 0.03691610985878476,
      "grad_norm": 5.428725719451904,
      "learning_rate": 9.973142377242209e-05,
      "loss": 0.1424,
      "step": 913
    },
    {
      "epoch": 0.03695654371405178,
      "grad_norm": 7.879098892211914,
      "learning_rate": 9.973076325591471e-05,
      "loss": 0.2318,
      "step": 914
    },
    {
      "epoch": 0.036996977569318794,
      "grad_norm": 4.392131805419922,
      "learning_rate": 9.973010193038448e-05,
      "loss": 0.0756,
      "step": 915
    },
    {
      "epoch": 0.0370374114245858,
      "grad_norm": 4.797752857208252,
      "learning_rate": 9.972943979584217e-05,
      "loss": 0.1925,
      "step": 916
    },
    {
      "epoch": 0.03707784527985282,
      "grad_norm": 6.208285331726074,
      "learning_rate": 9.972877685229857e-05,
      "loss": 0.095,
      "step": 917
    },
    {
      "epoch": 0.037118279135119836,
      "grad_norm": 4.34920597076416,
      "learning_rate": 9.972811309976444e-05,
      "loss": 0.1972,
      "step": 918
    },
    {
      "epoch": 0.03715871299038685,
      "grad_norm": 8.596395492553711,
      "learning_rate": 9.972744853825058e-05,
      "loss": 0.2043,
      "step": 919
    },
    {
      "epoch": 0.03719914684565387,
      "grad_norm": 6.011871814727783,
      "learning_rate": 9.972678316776782e-05,
      "loss": 0.1508,
      "step": 920
    },
    {
      "epoch": 0.03723958070092088,
      "grad_norm": 7.894033908843994,
      "learning_rate": 9.972611698832697e-05,
      "loss": 0.1389,
      "step": 921
    },
    {
      "epoch": 0.037280014556187895,
      "grad_norm": 4.6798295974731445,
      "learning_rate": 9.972544999993885e-05,
      "loss": 0.3508,
      "step": 922
    },
    {
      "epoch": 0.03732044841145491,
      "grad_norm": 9.059563636779785,
      "learning_rate": 9.972478220261435e-05,
      "loss": 0.3061,
      "step": 923
    },
    {
      "epoch": 0.03736088226672193,
      "grad_norm": 4.755893230438232,
      "learning_rate": 9.97241135963643e-05,
      "loss": 0.1632,
      "step": 924
    },
    {
      "epoch": 0.037401316121988944,
      "grad_norm": 7.823504447937012,
      "learning_rate": 9.972344418119959e-05,
      "loss": 0.2758,
      "step": 925
    },
    {
      "epoch": 0.03744174997725595,
      "grad_norm": 5.830699920654297,
      "learning_rate": 9.972277395713112e-05,
      "loss": 0.2382,
      "step": 926
    },
    {
      "epoch": 0.03748218383252297,
      "grad_norm": 7.588654041290283,
      "learning_rate": 9.972210292416978e-05,
      "loss": 0.1598,
      "step": 927
    },
    {
      "epoch": 0.037522617687789986,
      "grad_norm": 6.8968706130981445,
      "learning_rate": 9.972143108232648e-05,
      "loss": 0.2321,
      "step": 928
    },
    {
      "epoch": 0.037563051543057,
      "grad_norm": 7.499555587768555,
      "learning_rate": 9.972075843161216e-05,
      "loss": 0.2312,
      "step": 929
    },
    {
      "epoch": 0.03760348539832402,
      "grad_norm": 6.824891090393066,
      "learning_rate": 9.972008497203778e-05,
      "loss": 0.1922,
      "step": 930
    },
    {
      "epoch": 0.037643919253591035,
      "grad_norm": 4.587335586547852,
      "learning_rate": 9.971941070361426e-05,
      "loss": 0.1098,
      "step": 931
    },
    {
      "epoch": 0.037684353108858044,
      "grad_norm": 4.977753162384033,
      "learning_rate": 9.97187356263526e-05,
      "loss": 0.1975,
      "step": 932
    },
    {
      "epoch": 0.03772478696412506,
      "grad_norm": 8.830099105834961,
      "learning_rate": 9.971805974026376e-05,
      "loss": 0.2496,
      "step": 933
    },
    {
      "epoch": 0.03776522081939208,
      "grad_norm": 3.2122678756713867,
      "learning_rate": 9.971738304535873e-05,
      "loss": 0.1227,
      "step": 934
    },
    {
      "epoch": 0.037805654674659094,
      "grad_norm": 3.750443458557129,
      "learning_rate": 9.971670554164856e-05,
      "loss": 0.1413,
      "step": 935
    },
    {
      "epoch": 0.03784608852992611,
      "grad_norm": 3.564455986022949,
      "learning_rate": 9.971602722914424e-05,
      "loss": 0.1932,
      "step": 936
    },
    {
      "epoch": 0.03788652238519312,
      "grad_norm": 6.608478546142578,
      "learning_rate": 9.971534810785679e-05,
      "loss": 0.1479,
      "step": 937
    },
    {
      "epoch": 0.037926956240460136,
      "grad_norm": 2.7007393836975098,
      "learning_rate": 9.97146681777973e-05,
      "loss": 0.1209,
      "step": 938
    },
    {
      "epoch": 0.03796739009572715,
      "grad_norm": 6.7665300369262695,
      "learning_rate": 9.97139874389768e-05,
      "loss": 0.4071,
      "step": 939
    },
    {
      "epoch": 0.03800782395099417,
      "grad_norm": 11.875426292419434,
      "learning_rate": 9.971330589140637e-05,
      "loss": 0.2755,
      "step": 940
    },
    {
      "epoch": 0.038048257806261185,
      "grad_norm": 4.566583633422852,
      "learning_rate": 9.97126235350971e-05,
      "loss": 0.1741,
      "step": 941
    },
    {
      "epoch": 0.038088691661528194,
      "grad_norm": 4.127447128295898,
      "learning_rate": 9.971194037006008e-05,
      "loss": 0.2307,
      "step": 942
    },
    {
      "epoch": 0.03812912551679521,
      "grad_norm": 3.277635097503662,
      "learning_rate": 9.971125639630646e-05,
      "loss": 0.1028,
      "step": 943
    },
    {
      "epoch": 0.03816955937206223,
      "grad_norm": 9.176007270812988,
      "learning_rate": 9.971057161384734e-05,
      "loss": 0.2088,
      "step": 944
    },
    {
      "epoch": 0.03820999322732924,
      "grad_norm": 4.527174949645996,
      "learning_rate": 9.970988602269384e-05,
      "loss": 0.2065,
      "step": 945
    },
    {
      "epoch": 0.03825042708259626,
      "grad_norm": 7.769873142242432,
      "learning_rate": 9.970919962285715e-05,
      "loss": 0.1629,
      "step": 946
    },
    {
      "epoch": 0.038290860937863276,
      "grad_norm": 7.199928283691406,
      "learning_rate": 9.970851241434841e-05,
      "loss": 0.1778,
      "step": 947
    },
    {
      "epoch": 0.038331294793130286,
      "grad_norm": 3.636061668395996,
      "learning_rate": 9.970782439717882e-05,
      "loss": 0.1199,
      "step": 948
    },
    {
      "epoch": 0.0383717286483973,
      "grad_norm": 8.441319465637207,
      "learning_rate": 9.970713557135956e-05,
      "loss": 0.3425,
      "step": 949
    },
    {
      "epoch": 0.03841216250366432,
      "grad_norm": 6.82443380355835,
      "learning_rate": 9.970644593690184e-05,
      "loss": 0.1196,
      "step": 950
    },
    {
      "epoch": 0.038452596358931335,
      "grad_norm": 3.8683066368103027,
      "learning_rate": 9.970575549381687e-05,
      "loss": 0.1951,
      "step": 951
    },
    {
      "epoch": 0.03849303021419835,
      "grad_norm": 7.335735321044922,
      "learning_rate": 9.97050642421159e-05,
      "loss": 0.2465,
      "step": 952
    },
    {
      "epoch": 0.03853346406946536,
      "grad_norm": 3.0663907527923584,
      "learning_rate": 9.970437218181017e-05,
      "loss": 0.2309,
      "step": 953
    },
    {
      "epoch": 0.03857389792473238,
      "grad_norm": 3.9604740142822266,
      "learning_rate": 9.970367931291094e-05,
      "loss": 0.2384,
      "step": 954
    },
    {
      "epoch": 0.03861433177999939,
      "grad_norm": 6.34889030456543,
      "learning_rate": 9.970298563542946e-05,
      "loss": 0.3172,
      "step": 955
    },
    {
      "epoch": 0.03865476563526641,
      "grad_norm": 3.1716644763946533,
      "learning_rate": 9.970229114937703e-05,
      "loss": 0.086,
      "step": 956
    },
    {
      "epoch": 0.038695199490533426,
      "grad_norm": 5.945620536804199,
      "learning_rate": 9.970159585476496e-05,
      "loss": 0.2258,
      "step": 957
    },
    {
      "epoch": 0.038735633345800435,
      "grad_norm": 4.128820419311523,
      "learning_rate": 9.970089975160453e-05,
      "loss": 0.1867,
      "step": 958
    },
    {
      "epoch": 0.03877606720106745,
      "grad_norm": 3.8432328701019287,
      "learning_rate": 9.970020283990711e-05,
      "loss": 0.1447,
      "step": 959
    },
    {
      "epoch": 0.03881650105633447,
      "grad_norm": 5.634805202484131,
      "learning_rate": 9.9699505119684e-05,
      "loss": 0.2135,
      "step": 960
    },
    {
      "epoch": 0.038856934911601484,
      "grad_norm": 5.576113224029541,
      "learning_rate": 9.969880659094655e-05,
      "loss": 0.2029,
      "step": 961
    },
    {
      "epoch": 0.0388973687668685,
      "grad_norm": 2.158010482788086,
      "learning_rate": 9.969810725370615e-05,
      "loss": 0.1496,
      "step": 962
    },
    {
      "epoch": 0.03893780262213552,
      "grad_norm": 6.7158684730529785,
      "learning_rate": 9.969740710797417e-05,
      "loss": 0.1438,
      "step": 963
    },
    {
      "epoch": 0.03897823647740253,
      "grad_norm": 9.979058265686035,
      "learning_rate": 9.969670615376196e-05,
      "loss": 0.2435,
      "step": 964
    },
    {
      "epoch": 0.03901867033266954,
      "grad_norm": 3.3732995986938477,
      "learning_rate": 9.9696004391081e-05,
      "loss": 0.1708,
      "step": 965
    },
    {
      "epoch": 0.03905910418793656,
      "grad_norm": 8.409553527832031,
      "learning_rate": 9.969530181994263e-05,
      "loss": 0.2585,
      "step": 966
    },
    {
      "epoch": 0.039099538043203576,
      "grad_norm": 3.084747076034546,
      "learning_rate": 9.969459844035833e-05,
      "loss": 0.0946,
      "step": 967
    },
    {
      "epoch": 0.03913997189847059,
      "grad_norm": 6.054865837097168,
      "learning_rate": 9.96938942523395e-05,
      "loss": 0.2942,
      "step": 968
    },
    {
      "epoch": 0.0391804057537376,
      "grad_norm": 4.952055931091309,
      "learning_rate": 9.969318925589764e-05,
      "loss": 0.2048,
      "step": 969
    },
    {
      "epoch": 0.03922083960900462,
      "grad_norm": 6.14146089553833,
      "learning_rate": 9.969248345104419e-05,
      "loss": 0.2598,
      "step": 970
    },
    {
      "epoch": 0.039261273464271634,
      "grad_norm": 6.694477558135986,
      "learning_rate": 9.969177683779065e-05,
      "loss": 0.3708,
      "step": 971
    },
    {
      "epoch": 0.03930170731953865,
      "grad_norm": 6.406944274902344,
      "learning_rate": 9.96910694161485e-05,
      "loss": 0.2011,
      "step": 972
    },
    {
      "epoch": 0.03934214117480567,
      "grad_norm": 7.077157497406006,
      "learning_rate": 9.969036118612925e-05,
      "loss": 0.2431,
      "step": 973
    },
    {
      "epoch": 0.039382575030072677,
      "grad_norm": 6.514596939086914,
      "learning_rate": 9.968965214774444e-05,
      "loss": 0.1166,
      "step": 974
    },
    {
      "epoch": 0.03942300888533969,
      "grad_norm": 2.419205904006958,
      "learning_rate": 9.968894230100559e-05,
      "loss": 0.1095,
      "step": 975
    },
    {
      "epoch": 0.03946344274060671,
      "grad_norm": 5.408444881439209,
      "learning_rate": 9.968823164592423e-05,
      "loss": 0.2548,
      "step": 976
    },
    {
      "epoch": 0.039503876595873726,
      "grad_norm": 7.297247886657715,
      "learning_rate": 9.968752018251196e-05,
      "loss": 0.206,
      "step": 977
    },
    {
      "epoch": 0.03954431045114074,
      "grad_norm": 5.55707311630249,
      "learning_rate": 9.968680791078032e-05,
      "loss": 0.1482,
      "step": 978
    },
    {
      "epoch": 0.03958474430640776,
      "grad_norm": 5.616539478302002,
      "learning_rate": 9.968609483074091e-05,
      "loss": 0.2317,
      "step": 979
    },
    {
      "epoch": 0.03962517816167477,
      "grad_norm": 5.391134262084961,
      "learning_rate": 9.968538094240534e-05,
      "loss": 0.2246,
      "step": 980
    },
    {
      "epoch": 0.039665612016941784,
      "grad_norm": 2.4739136695861816,
      "learning_rate": 9.96846662457852e-05,
      "loss": 0.0826,
      "step": 981
    },
    {
      "epoch": 0.0397060458722088,
      "grad_norm": 6.499474048614502,
      "learning_rate": 9.968395074089214e-05,
      "loss": 0.2161,
      "step": 982
    },
    {
      "epoch": 0.03974647972747582,
      "grad_norm": 12.792696952819824,
      "learning_rate": 9.968323442773779e-05,
      "loss": 0.4865,
      "step": 983
    },
    {
      "epoch": 0.03978691358274283,
      "grad_norm": 9.621682167053223,
      "learning_rate": 9.968251730633382e-05,
      "loss": 0.3283,
      "step": 984
    },
    {
      "epoch": 0.03982734743800984,
      "grad_norm": 9.544614791870117,
      "learning_rate": 9.968179937669187e-05,
      "loss": 0.3596,
      "step": 985
    },
    {
      "epoch": 0.03986778129327686,
      "grad_norm": 7.390946865081787,
      "learning_rate": 9.968108063882362e-05,
      "loss": 0.2341,
      "step": 986
    },
    {
      "epoch": 0.039908215148543875,
      "grad_norm": 4.2290215492248535,
      "learning_rate": 9.96803610927408e-05,
      "loss": 0.1831,
      "step": 987
    },
    {
      "epoch": 0.03994864900381089,
      "grad_norm": 2.9978044033050537,
      "learning_rate": 9.967964073845506e-05,
      "loss": 0.1968,
      "step": 988
    },
    {
      "epoch": 0.03998908285907791,
      "grad_norm": 5.247458457946777,
      "learning_rate": 9.967891957597815e-05,
      "loss": 0.1911,
      "step": 989
    },
    {
      "epoch": 0.04002951671434492,
      "grad_norm": 4.375576972961426,
      "learning_rate": 9.96781976053218e-05,
      "loss": 0.1204,
      "step": 990
    },
    {
      "epoch": 0.040069950569611934,
      "grad_norm": 5.390709400177002,
      "learning_rate": 9.967747482649777e-05,
      "loss": 0.2061,
      "step": 991
    },
    {
      "epoch": 0.04011038442487895,
      "grad_norm": 7.774453163146973,
      "learning_rate": 9.967675123951779e-05,
      "loss": 0.158,
      "step": 992
    },
    {
      "epoch": 0.04015081828014597,
      "grad_norm": 6.610381603240967,
      "learning_rate": 9.967602684439364e-05,
      "loss": 0.3152,
      "step": 993
    },
    {
      "epoch": 0.04019125213541298,
      "grad_norm": 3.569241523742676,
      "learning_rate": 9.967530164113711e-05,
      "loss": 0.2083,
      "step": 994
    },
    {
      "epoch": 0.04023168599068,
      "grad_norm": 5.130991458892822,
      "learning_rate": 9.967457562976e-05,
      "loss": 0.2448,
      "step": 995
    },
    {
      "epoch": 0.04027211984594701,
      "grad_norm": 4.351527690887451,
      "learning_rate": 9.967384881027411e-05,
      "loss": 0.1414,
      "step": 996
    },
    {
      "epoch": 0.040312553701214025,
      "grad_norm": 4.036287307739258,
      "learning_rate": 9.967312118269127e-05,
      "loss": 0.1657,
      "step": 997
    },
    {
      "epoch": 0.04035298755648104,
      "grad_norm": 5.094952583312988,
      "learning_rate": 9.967239274702333e-05,
      "loss": 0.1043,
      "step": 998
    },
    {
      "epoch": 0.04039342141174806,
      "grad_norm": 5.604987144470215,
      "learning_rate": 9.96716635032821e-05,
      "loss": 0.2876,
      "step": 999
    },
    {
      "epoch": 0.040433855267015074,
      "grad_norm": 6.747164726257324,
      "learning_rate": 9.967093345147951e-05,
      "loss": 0.1747,
      "step": 1000
    },
    {
      "epoch": 0.040474289122282084,
      "grad_norm": 3.242488384246826,
      "learning_rate": 9.967020259162737e-05,
      "loss": 0.1905,
      "step": 1001
    },
    {
      "epoch": 0.0405147229775491,
      "grad_norm": 5.311917781829834,
      "learning_rate": 9.96694709237376e-05,
      "loss": 0.3084,
      "step": 1002
    },
    {
      "epoch": 0.04055515683281612,
      "grad_norm": 8.132774353027344,
      "learning_rate": 9.96687384478221e-05,
      "loss": 0.2718,
      "step": 1003
    },
    {
      "epoch": 0.04059559068808313,
      "grad_norm": 8.841411590576172,
      "learning_rate": 9.96680051638928e-05,
      "loss": 0.2219,
      "step": 1004
    },
    {
      "epoch": 0.04063602454335015,
      "grad_norm": 7.7256598472595215,
      "learning_rate": 9.96672710719616e-05,
      "loss": 0.2314,
      "step": 1005
    },
    {
      "epoch": 0.04067645839861716,
      "grad_norm": 4.769709587097168,
      "learning_rate": 9.966653617204047e-05,
      "loss": 0.2761,
      "step": 1006
    },
    {
      "epoch": 0.040716892253884175,
      "grad_norm": 4.831228733062744,
      "learning_rate": 9.966580046414133e-05,
      "loss": 0.1651,
      "step": 1007
    },
    {
      "epoch": 0.04075732610915119,
      "grad_norm": 7.274237155914307,
      "learning_rate": 9.966506394827619e-05,
      "loss": 0.2205,
      "step": 1008
    },
    {
      "epoch": 0.04079775996441821,
      "grad_norm": 2.80499005317688,
      "learning_rate": 9.966432662445701e-05,
      "loss": 0.2258,
      "step": 1009
    },
    {
      "epoch": 0.040838193819685224,
      "grad_norm": 2.9447154998779297,
      "learning_rate": 9.966358849269579e-05,
      "loss": 0.1565,
      "step": 1010
    },
    {
      "epoch": 0.04087862767495224,
      "grad_norm": 3.0066587924957275,
      "learning_rate": 9.966284955300453e-05,
      "loss": 0.1429,
      "step": 1011
    },
    {
      "epoch": 0.04091906153021925,
      "grad_norm": 2.405465841293335,
      "learning_rate": 9.966210980539525e-05,
      "loss": 0.1237,
      "step": 1012
    },
    {
      "epoch": 0.040959495385486266,
      "grad_norm": 6.896819114685059,
      "learning_rate": 9.966136924988001e-05,
      "loss": 0.1838,
      "step": 1013
    },
    {
      "epoch": 0.04099992924075328,
      "grad_norm": 6.092067241668701,
      "learning_rate": 9.966062788647082e-05,
      "loss": 0.2291,
      "step": 1014
    },
    {
      "epoch": 0.0410403630960203,
      "grad_norm": 11.456220626831055,
      "learning_rate": 9.965988571517976e-05,
      "loss": 0.3312,
      "step": 1015
    },
    {
      "epoch": 0.041080796951287316,
      "grad_norm": 1.6754671335220337,
      "learning_rate": 9.965914273601893e-05,
      "loss": 0.1057,
      "step": 1016
    },
    {
      "epoch": 0.041121230806554325,
      "grad_norm": 6.922665596008301,
      "learning_rate": 9.965839894900037e-05,
      "loss": 0.1211,
      "step": 1017
    },
    {
      "epoch": 0.04116166466182134,
      "grad_norm": 6.462112903594971,
      "learning_rate": 9.965765435413618e-05,
      "loss": 0.3266,
      "step": 1018
    },
    {
      "epoch": 0.04120209851708836,
      "grad_norm": 7.278192043304443,
      "learning_rate": 9.965690895143853e-05,
      "loss": 0.2464,
      "step": 1019
    },
    {
      "epoch": 0.041242532372355374,
      "grad_norm": 5.592101097106934,
      "learning_rate": 9.965616274091949e-05,
      "loss": 0.2272,
      "step": 1020
    },
    {
      "epoch": 0.04128296622762239,
      "grad_norm": 7.2022480964660645,
      "learning_rate": 9.96554157225912e-05,
      "loss": 0.2112,
      "step": 1021
    },
    {
      "epoch": 0.0413234000828894,
      "grad_norm": 6.607624530792236,
      "learning_rate": 9.965466789646585e-05,
      "loss": 0.2253,
      "step": 1022
    },
    {
      "epoch": 0.041363833938156416,
      "grad_norm": 8.514381408691406,
      "learning_rate": 9.965391926255559e-05,
      "loss": 0.1453,
      "step": 1023
    },
    {
      "epoch": 0.04140426779342343,
      "grad_norm": 3.9110910892486572,
      "learning_rate": 9.965316982087259e-05,
      "loss": 0.1928,
      "step": 1024
    },
    {
      "epoch": 0.04144470164869045,
      "grad_norm": 6.122485637664795,
      "learning_rate": 9.965241957142906e-05,
      "loss": 0.1622,
      "step": 1025
    },
    {
      "epoch": 0.041485135503957465,
      "grad_norm": 5.429446697235107,
      "learning_rate": 9.965166851423717e-05,
      "loss": 0.1354,
      "step": 1026
    },
    {
      "epoch": 0.04152556935922448,
      "grad_norm": 8.99289321899414,
      "learning_rate": 9.965091664930916e-05,
      "loss": 0.2376,
      "step": 1027
    },
    {
      "epoch": 0.04156600321449149,
      "grad_norm": 2.773200273513794,
      "learning_rate": 9.965016397665728e-05,
      "loss": 0.0978,
      "step": 1028
    },
    {
      "epoch": 0.04160643706975851,
      "grad_norm": 6.850367069244385,
      "learning_rate": 9.964941049629374e-05,
      "loss": 0.3082,
      "step": 1029
    },
    {
      "epoch": 0.041646870925025524,
      "grad_norm": 6.860160827636719,
      "learning_rate": 9.964865620823084e-05,
      "loss": 0.2104,
      "step": 1030
    },
    {
      "epoch": 0.04168730478029254,
      "grad_norm": 9.532269477844238,
      "learning_rate": 9.96479011124808e-05,
      "loss": 0.2037,
      "step": 1031
    },
    {
      "epoch": 0.04172773863555956,
      "grad_norm": 7.938676357269287,
      "learning_rate": 9.964714520905594e-05,
      "loss": 0.1622,
      "step": 1032
    },
    {
      "epoch": 0.041768172490826566,
      "grad_norm": 13.559954643249512,
      "learning_rate": 9.964638849796855e-05,
      "loss": 0.382,
      "step": 1033
    },
    {
      "epoch": 0.04180860634609358,
      "grad_norm": 4.60775899887085,
      "learning_rate": 9.964563097923091e-05,
      "loss": 0.1087,
      "step": 1034
    },
    {
      "epoch": 0.0418490402013606,
      "grad_norm": 8.237541198730469,
      "learning_rate": 9.964487265285541e-05,
      "loss": 0.1454,
      "step": 1035
    },
    {
      "epoch": 0.041889474056627615,
      "grad_norm": 4.831090450286865,
      "learning_rate": 9.964411351885432e-05,
      "loss": 0.1125,
      "step": 1036
    },
    {
      "epoch": 0.04192990791189463,
      "grad_norm": 5.011022567749023,
      "learning_rate": 9.964335357724002e-05,
      "loss": 0.2702,
      "step": 1037
    },
    {
      "epoch": 0.04197034176716164,
      "grad_norm": 4.2666425704956055,
      "learning_rate": 9.964259282802486e-05,
      "loss": 0.1887,
      "step": 1038
    },
    {
      "epoch": 0.04201077562242866,
      "grad_norm": 3.486125946044922,
      "learning_rate": 9.964183127122124e-05,
      "loss": 0.2357,
      "step": 1039
    },
    {
      "epoch": 0.042051209477695674,
      "grad_norm": 6.843602180480957,
      "learning_rate": 9.964106890684153e-05,
      "loss": 0.18,
      "step": 1040
    },
    {
      "epoch": 0.04209164333296269,
      "grad_norm": 5.6598029136657715,
      "learning_rate": 9.964030573489813e-05,
      "loss": 0.197,
      "step": 1041
    },
    {
      "epoch": 0.04213207718822971,
      "grad_norm": 7.2295145988464355,
      "learning_rate": 9.963954175540345e-05,
      "loss": 0.2122,
      "step": 1042
    },
    {
      "epoch": 0.04217251104349672,
      "grad_norm": 4.633933067321777,
      "learning_rate": 9.963877696836994e-05,
      "loss": 0.2077,
      "step": 1043
    },
    {
      "epoch": 0.04221294489876373,
      "grad_norm": 6.034922122955322,
      "learning_rate": 9.963801137381004e-05,
      "loss": 0.2042,
      "step": 1044
    },
    {
      "epoch": 0.04225337875403075,
      "grad_norm": 5.653476238250732,
      "learning_rate": 9.963724497173618e-05,
      "loss": 0.1836,
      "step": 1045
    },
    {
      "epoch": 0.042293812609297765,
      "grad_norm": 11.303414344787598,
      "learning_rate": 9.963647776216085e-05,
      "loss": 0.2408,
      "step": 1046
    },
    {
      "epoch": 0.04233424646456478,
      "grad_norm": 2.827366828918457,
      "learning_rate": 9.963570974509653e-05,
      "loss": 0.1506,
      "step": 1047
    },
    {
      "epoch": 0.0423746803198318,
      "grad_norm": 9.231380462646484,
      "learning_rate": 9.96349409205557e-05,
      "loss": 0.2071,
      "step": 1048
    },
    {
      "epoch": 0.04241511417509881,
      "grad_norm": 1.9777097702026367,
      "learning_rate": 9.963417128855089e-05,
      "loss": 0.1023,
      "step": 1049
    },
    {
      "epoch": 0.042455548030365824,
      "grad_norm": 7.244409084320068,
      "learning_rate": 9.96334008490946e-05,
      "loss": 0.1766,
      "step": 1050
    },
    {
      "epoch": 0.04249598188563284,
      "grad_norm": 12.106108665466309,
      "learning_rate": 9.963262960219937e-05,
      "loss": 0.2759,
      "step": 1051
    },
    {
      "epoch": 0.042536415740899856,
      "grad_norm": 6.6821699142456055,
      "learning_rate": 9.963185754787774e-05,
      "loss": 0.236,
      "step": 1052
    },
    {
      "epoch": 0.04257684959616687,
      "grad_norm": 11.79532241821289,
      "learning_rate": 9.963108468614229e-05,
      "loss": 0.2613,
      "step": 1053
    },
    {
      "epoch": 0.04261728345143388,
      "grad_norm": 4.9874186515808105,
      "learning_rate": 9.963031101700555e-05,
      "loss": 0.1532,
      "step": 1054
    },
    {
      "epoch": 0.0426577173067009,
      "grad_norm": 4.2928900718688965,
      "learning_rate": 9.962953654048017e-05,
      "loss": 0.1587,
      "step": 1055
    },
    {
      "epoch": 0.042698151161967915,
      "grad_norm": 2.786557912826538,
      "learning_rate": 9.962876125657869e-05,
      "loss": 0.1478,
      "step": 1056
    },
    {
      "epoch": 0.04273858501723493,
      "grad_norm": 7.082719802856445,
      "learning_rate": 9.962798516531375e-05,
      "loss": 0.1798,
      "step": 1057
    },
    {
      "epoch": 0.04277901887250195,
      "grad_norm": 7.698630332946777,
      "learning_rate": 9.9627208266698e-05,
      "loss": 0.4274,
      "step": 1058
    },
    {
      "epoch": 0.042819452727768964,
      "grad_norm": 4.552220344543457,
      "learning_rate": 9.962643056074402e-05,
      "loss": 0.1747,
      "step": 1059
    },
    {
      "epoch": 0.042859886583035973,
      "grad_norm": 2.834768295288086,
      "learning_rate": 9.962565204746451e-05,
      "loss": 0.138,
      "step": 1060
    },
    {
      "epoch": 0.04290032043830299,
      "grad_norm": 2.6517117023468018,
      "learning_rate": 9.962487272687211e-05,
      "loss": 0.066,
      "step": 1061
    },
    {
      "epoch": 0.042940754293570006,
      "grad_norm": 8.92813491821289,
      "learning_rate": 9.96240925989795e-05,
      "loss": 0.3338,
      "step": 1062
    },
    {
      "epoch": 0.04298118814883702,
      "grad_norm": 10.537973403930664,
      "learning_rate": 9.962331166379939e-05,
      "loss": 0.2417,
      "step": 1063
    },
    {
      "epoch": 0.04302162200410404,
      "grad_norm": 3.7294211387634277,
      "learning_rate": 9.962252992134447e-05,
      "loss": 0.1287,
      "step": 1064
    },
    {
      "epoch": 0.04306205585937105,
      "grad_norm": 7.48951530456543,
      "learning_rate": 9.962174737162744e-05,
      "loss": 0.2689,
      "step": 1065
    },
    {
      "epoch": 0.043102489714638065,
      "grad_norm": 4.3161163330078125,
      "learning_rate": 9.962096401466108e-05,
      "loss": 0.2317,
      "step": 1066
    },
    {
      "epoch": 0.04314292356990508,
      "grad_norm": 5.372151851654053,
      "learning_rate": 9.962017985045808e-05,
      "loss": 0.1436,
      "step": 1067
    },
    {
      "epoch": 0.0431833574251721,
      "grad_norm": 5.337826251983643,
      "learning_rate": 9.961939487903123e-05,
      "loss": 0.2586,
      "step": 1068
    },
    {
      "epoch": 0.043223791280439114,
      "grad_norm": 7.526915550231934,
      "learning_rate": 9.961860910039327e-05,
      "loss": 0.2072,
      "step": 1069
    },
    {
      "epoch": 0.04326422513570612,
      "grad_norm": 4.610411167144775,
      "learning_rate": 9.961782251455703e-05,
      "loss": 0.2096,
      "step": 1070
    },
    {
      "epoch": 0.04330465899097314,
      "grad_norm": 6.585238933563232,
      "learning_rate": 9.961703512153526e-05,
      "loss": 0.1399,
      "step": 1071
    },
    {
      "epoch": 0.043345092846240156,
      "grad_norm": 3.7144370079040527,
      "learning_rate": 9.96162469213408e-05,
      "loss": 0.215,
      "step": 1072
    },
    {
      "epoch": 0.04338552670150717,
      "grad_norm": 4.416204452514648,
      "learning_rate": 9.961545791398646e-05,
      "loss": 0.2415,
      "step": 1073
    },
    {
      "epoch": 0.04342596055677419,
      "grad_norm": 5.932099342346191,
      "learning_rate": 9.961466809948508e-05,
      "loss": 0.2498,
      "step": 1074
    },
    {
      "epoch": 0.043466394412041205,
      "grad_norm": 3.0711920261383057,
      "learning_rate": 9.961387747784949e-05,
      "loss": 0.1935,
      "step": 1075
    },
    {
      "epoch": 0.043506828267308215,
      "grad_norm": 5.047266483306885,
      "learning_rate": 9.961308604909258e-05,
      "loss": 0.2205,
      "step": 1076
    },
    {
      "epoch": 0.04354726212257523,
      "grad_norm": 5.084931373596191,
      "learning_rate": 9.961229381322721e-05,
      "loss": 0.1411,
      "step": 1077
    },
    {
      "epoch": 0.04358769597784225,
      "grad_norm": 5.8279643058776855,
      "learning_rate": 9.961150077026627e-05,
      "loss": 0.1659,
      "step": 1078
    },
    {
      "epoch": 0.043628129833109264,
      "grad_norm": 8.403576850891113,
      "learning_rate": 9.961070692022268e-05,
      "loss": 0.2171,
      "step": 1079
    },
    {
      "epoch": 0.04366856368837628,
      "grad_norm": 4.912980556488037,
      "learning_rate": 9.960991226310931e-05,
      "loss": 0.186,
      "step": 1080
    },
    {
      "epoch": 0.04370899754364329,
      "grad_norm": 2.2110841274261475,
      "learning_rate": 9.960911679893912e-05,
      "loss": 0.1064,
      "step": 1081
    },
    {
      "epoch": 0.043749431398910306,
      "grad_norm": 7.117617130279541,
      "learning_rate": 9.960832052772503e-05,
      "loss": 0.1908,
      "step": 1082
    },
    {
      "epoch": 0.04378986525417732,
      "grad_norm": 7.387581825256348,
      "learning_rate": 9.960752344948003e-05,
      "loss": 0.1674,
      "step": 1083
    },
    {
      "epoch": 0.04383029910944434,
      "grad_norm": 6.171600341796875,
      "learning_rate": 9.960672556421706e-05,
      "loss": 0.1701,
      "step": 1084
    },
    {
      "epoch": 0.043870732964711355,
      "grad_norm": 10.121071815490723,
      "learning_rate": 9.960592687194911e-05,
      "loss": 0.1953,
      "step": 1085
    },
    {
      "epoch": 0.043911166819978364,
      "grad_norm": 11.535463333129883,
      "learning_rate": 9.960512737268916e-05,
      "loss": 0.2143,
      "step": 1086
    },
    {
      "epoch": 0.04395160067524538,
      "grad_norm": 3.3594110012054443,
      "learning_rate": 9.960432706645022e-05,
      "loss": 0.2099,
      "step": 1087
    },
    {
      "epoch": 0.0439920345305124,
      "grad_norm": 8.039867401123047,
      "learning_rate": 9.960352595324531e-05,
      "loss": 0.1974,
      "step": 1088
    },
    {
      "epoch": 0.044032468385779414,
      "grad_norm": 5.1939778327941895,
      "learning_rate": 9.960272403308749e-05,
      "loss": 0.1548,
      "step": 1089
    },
    {
      "epoch": 0.04407290224104643,
      "grad_norm": 3.5446434020996094,
      "learning_rate": 9.960192130598976e-05,
      "loss": 0.1335,
      "step": 1090
    },
    {
      "epoch": 0.044113336096313446,
      "grad_norm": 6.8553080558776855,
      "learning_rate": 9.960111777196522e-05,
      "loss": 0.188,
      "step": 1091
    },
    {
      "epoch": 0.044153769951580456,
      "grad_norm": 7.548846244812012,
      "learning_rate": 9.960031343102691e-05,
      "loss": 0.2646,
      "step": 1092
    },
    {
      "epoch": 0.04419420380684747,
      "grad_norm": 3.623375415802002,
      "learning_rate": 9.959950828318792e-05,
      "loss": 0.1446,
      "step": 1093
    },
    {
      "epoch": 0.04423463766211449,
      "grad_norm": 6.641726493835449,
      "learning_rate": 9.959870232846137e-05,
      "loss": 0.1464,
      "step": 1094
    },
    {
      "epoch": 0.044275071517381505,
      "grad_norm": 5.442054271697998,
      "learning_rate": 9.959789556686035e-05,
      "loss": 0.185,
      "step": 1095
    },
    {
      "epoch": 0.04431550537264852,
      "grad_norm": 2.7019641399383545,
      "learning_rate": 9.9597087998398e-05,
      "loss": 0.1262,
      "step": 1096
    },
    {
      "epoch": 0.04435593922791553,
      "grad_norm": 4.988285541534424,
      "learning_rate": 9.959627962308743e-05,
      "loss": 0.0692,
      "step": 1097
    },
    {
      "epoch": 0.04439637308318255,
      "grad_norm": 5.718743801116943,
      "learning_rate": 9.959547044094184e-05,
      "loss": 0.1222,
      "step": 1098
    },
    {
      "epoch": 0.04443680693844956,
      "grad_norm": 5.107194423675537,
      "learning_rate": 9.959466045197434e-05,
      "loss": 0.1858,
      "step": 1099
    },
    {
      "epoch": 0.04447724079371658,
      "grad_norm": 5.481889247894287,
      "learning_rate": 9.959384965619815e-05,
      "loss": 0.1628,
      "step": 1100
    },
    {
      "epoch": 0.044517674648983596,
      "grad_norm": 4.749547004699707,
      "learning_rate": 9.959303805362643e-05,
      "loss": 0.1674,
      "step": 1101
    },
    {
      "epoch": 0.044558108504250606,
      "grad_norm": 8.081692695617676,
      "learning_rate": 9.95922256442724e-05,
      "loss": 0.1655,
      "step": 1102
    },
    {
      "epoch": 0.04459854235951762,
      "grad_norm": 5.6508893966674805,
      "learning_rate": 9.959141242814925e-05,
      "loss": 0.1799,
      "step": 1103
    },
    {
      "epoch": 0.04463897621478464,
      "grad_norm": 8.355573654174805,
      "learning_rate": 9.959059840527026e-05,
      "loss": 0.2799,
      "step": 1104
    },
    {
      "epoch": 0.044679410070051655,
      "grad_norm": 5.109837532043457,
      "learning_rate": 9.95897835756486e-05,
      "loss": 0.2141,
      "step": 1105
    },
    {
      "epoch": 0.04471984392531867,
      "grad_norm": 5.671930313110352,
      "learning_rate": 9.958896793929761e-05,
      "loss": 0.2868,
      "step": 1106
    },
    {
      "epoch": 0.04476027778058569,
      "grad_norm": 4.721503734588623,
      "learning_rate": 9.958815149623051e-05,
      "loss": 0.1806,
      "step": 1107
    },
    {
      "epoch": 0.0448007116358527,
      "grad_norm": 8.835660934448242,
      "learning_rate": 9.958733424646057e-05,
      "loss": 0.2433,
      "step": 1108
    },
    {
      "epoch": 0.04484114549111971,
      "grad_norm": 5.736001968383789,
      "learning_rate": 9.958651619000112e-05,
      "loss": 0.0969,
      "step": 1109
    },
    {
      "epoch": 0.04488157934638673,
      "grad_norm": 3.870368242263794,
      "learning_rate": 9.958569732686544e-05,
      "loss": 0.1874,
      "step": 1110
    },
    {
      "epoch": 0.044922013201653746,
      "grad_norm": 4.785103797912598,
      "learning_rate": 9.958487765706688e-05,
      "loss": 0.2779,
      "step": 1111
    },
    {
      "epoch": 0.04496244705692076,
      "grad_norm": 10.21684741973877,
      "learning_rate": 9.958405718061873e-05,
      "loss": 0.3585,
      "step": 1112
    },
    {
      "epoch": 0.04500288091218777,
      "grad_norm": 5.960936069488525,
      "learning_rate": 9.95832358975344e-05,
      "loss": 0.2426,
      "step": 1113
    },
    {
      "epoch": 0.04504331476745479,
      "grad_norm": 3.7496578693389893,
      "learning_rate": 9.95824138078272e-05,
      "loss": 0.1529,
      "step": 1114
    },
    {
      "epoch": 0.045083748622721805,
      "grad_norm": 12.032094955444336,
      "learning_rate": 9.958159091151051e-05,
      "loss": 0.4171,
      "step": 1115
    },
    {
      "epoch": 0.04512418247798882,
      "grad_norm": 4.11959981918335,
      "learning_rate": 9.958076720859775e-05,
      "loss": 0.1639,
      "step": 1116
    },
    {
      "epoch": 0.04516461633325584,
      "grad_norm": 7.389408111572266,
      "learning_rate": 9.957994269910228e-05,
      "loss": 0.1765,
      "step": 1117
    },
    {
      "epoch": 0.04520505018852285,
      "grad_norm": 4.228634357452393,
      "learning_rate": 9.957911738303754e-05,
      "loss": 0.1192,
      "step": 1118
    },
    {
      "epoch": 0.04524548404378986,
      "grad_norm": 3.4356515407562256,
      "learning_rate": 9.957829126041694e-05,
      "loss": 0.1771,
      "step": 1119
    },
    {
      "epoch": 0.04528591789905688,
      "grad_norm": 4.2367377281188965,
      "learning_rate": 9.957746433125392e-05,
      "loss": 0.1447,
      "step": 1120
    },
    {
      "epoch": 0.045326351754323896,
      "grad_norm": 6.006786823272705,
      "learning_rate": 9.957663659556196e-05,
      "loss": 0.2534,
      "step": 1121
    },
    {
      "epoch": 0.04536678560959091,
      "grad_norm": 5.121968746185303,
      "learning_rate": 9.957580805335447e-05,
      "loss": 0.1804,
      "step": 1122
    },
    {
      "epoch": 0.04540721946485793,
      "grad_norm": 8.69858455657959,
      "learning_rate": 9.957497870464499e-05,
      "loss": 0.1869,
      "step": 1123
    },
    {
      "epoch": 0.04544765332012494,
      "grad_norm": 2.493257999420166,
      "learning_rate": 9.957414854944697e-05,
      "loss": 0.1803,
      "step": 1124
    },
    {
      "epoch": 0.045488087175391954,
      "grad_norm": 8.855361938476562,
      "learning_rate": 9.957331758777394e-05,
      "loss": 0.3452,
      "step": 1125
    },
    {
      "epoch": 0.04552852103065897,
      "grad_norm": 2.2939271926879883,
      "learning_rate": 9.95724858196394e-05,
      "loss": 0.0989,
      "step": 1126
    },
    {
      "epoch": 0.04556895488592599,
      "grad_norm": 7.932673931121826,
      "learning_rate": 9.95716532450569e-05,
      "loss": 0.2689,
      "step": 1127
    },
    {
      "epoch": 0.045609388741193004,
      "grad_norm": 5.521693706512451,
      "learning_rate": 9.957081986403995e-05,
      "loss": 0.2228,
      "step": 1128
    },
    {
      "epoch": 0.04564982259646001,
      "grad_norm": 4.8846611976623535,
      "learning_rate": 9.956998567660216e-05,
      "loss": 0.2553,
      "step": 1129
    },
    {
      "epoch": 0.04569025645172703,
      "grad_norm": 3.5398826599121094,
      "learning_rate": 9.956915068275705e-05,
      "loss": 0.1522,
      "step": 1130
    },
    {
      "epoch": 0.045730690306994046,
      "grad_norm": 3.159580945968628,
      "learning_rate": 9.956831488251825e-05,
      "loss": 0.1892,
      "step": 1131
    },
    {
      "epoch": 0.04577112416226106,
      "grad_norm": 4.694597244262695,
      "learning_rate": 9.95674782758993e-05,
      "loss": 0.1466,
      "step": 1132
    },
    {
      "epoch": 0.04581155801752808,
      "grad_norm": 6.872805595397949,
      "learning_rate": 9.956664086291386e-05,
      "loss": 0.2407,
      "step": 1133
    },
    {
      "epoch": 0.04585199187279509,
      "grad_norm": 6.970820426940918,
      "learning_rate": 9.956580264357553e-05,
      "loss": 0.1073,
      "step": 1134
    },
    {
      "epoch": 0.045892425728062104,
      "grad_norm": 7.879580020904541,
      "learning_rate": 9.956496361789796e-05,
      "loss": 0.276,
      "step": 1135
    },
    {
      "epoch": 0.04593285958332912,
      "grad_norm": 2.136270761489868,
      "learning_rate": 9.95641237858948e-05,
      "loss": 0.112,
      "step": 1136
    },
    {
      "epoch": 0.04597329343859614,
      "grad_norm": 3.757417678833008,
      "learning_rate": 9.956328314757968e-05,
      "loss": 0.0962,
      "step": 1137
    },
    {
      "epoch": 0.04601372729386315,
      "grad_norm": 2.1068403720855713,
      "learning_rate": 9.95624417029663e-05,
      "loss": 0.0996,
      "step": 1138
    },
    {
      "epoch": 0.04605416114913017,
      "grad_norm": 7.61361837387085,
      "learning_rate": 9.956159945206835e-05,
      "loss": 0.2561,
      "step": 1139
    },
    {
      "epoch": 0.04609459500439718,
      "grad_norm": 6.596360683441162,
      "learning_rate": 9.956075639489953e-05,
      "loss": 0.3618,
      "step": 1140
    },
    {
      "epoch": 0.046135028859664196,
      "grad_norm": 4.988916397094727,
      "learning_rate": 9.955991253147356e-05,
      "loss": 0.2252,
      "step": 1141
    },
    {
      "epoch": 0.04617546271493121,
      "grad_norm": 3.9377646446228027,
      "learning_rate": 9.955906786180416e-05,
      "loss": 0.2621,
      "step": 1142
    },
    {
      "epoch": 0.04621589657019823,
      "grad_norm": 3.243882894515991,
      "learning_rate": 9.955822238590507e-05,
      "loss": 0.0575,
      "step": 1143
    },
    {
      "epoch": 0.046256330425465245,
      "grad_norm": 3.271188497543335,
      "learning_rate": 9.955737610379006e-05,
      "loss": 0.1101,
      "step": 1144
    },
    {
      "epoch": 0.046296764280732254,
      "grad_norm": 10.046098709106445,
      "learning_rate": 9.955652901547286e-05,
      "loss": 0.2827,
      "step": 1145
    },
    {
      "epoch": 0.04633719813599927,
      "grad_norm": 6.393460750579834,
      "learning_rate": 9.955568112096729e-05,
      "loss": 0.2634,
      "step": 1146
    },
    {
      "epoch": 0.04637763199126629,
      "grad_norm": 6.6856255531311035,
      "learning_rate": 9.955483242028714e-05,
      "loss": 0.2175,
      "step": 1147
    },
    {
      "epoch": 0.0464180658465333,
      "grad_norm": 10.207395553588867,
      "learning_rate": 9.955398291344619e-05,
      "loss": 0.2816,
      "step": 1148
    },
    {
      "epoch": 0.04645849970180032,
      "grad_norm": 6.325708389282227,
      "learning_rate": 9.955313260045828e-05,
      "loss": 0.229,
      "step": 1149
    },
    {
      "epoch": 0.046498933557067336,
      "grad_norm": 6.779524326324463,
      "learning_rate": 9.955228148133723e-05,
      "loss": 0.221,
      "step": 1150
    },
    {
      "epoch": 0.046539367412334345,
      "grad_norm": 13.740260124206543,
      "learning_rate": 9.95514295560969e-05,
      "loss": 0.3481,
      "step": 1151
    },
    {
      "epoch": 0.04657980126760136,
      "grad_norm": 13.101494789123535,
      "learning_rate": 9.955057682475115e-05,
      "loss": 0.3274,
      "step": 1152
    },
    {
      "epoch": 0.04662023512286838,
      "grad_norm": 13.507052421569824,
      "learning_rate": 9.954972328731385e-05,
      "loss": 0.342,
      "step": 1153
    },
    {
      "epoch": 0.046660668978135394,
      "grad_norm": 14.43952465057373,
      "learning_rate": 9.954886894379886e-05,
      "loss": 0.3575,
      "step": 1154
    },
    {
      "epoch": 0.04670110283340241,
      "grad_norm": 10.427202224731445,
      "learning_rate": 9.954801379422012e-05,
      "loss": 0.3386,
      "step": 1155
    },
    {
      "epoch": 0.04674153668866942,
      "grad_norm": 3.921696424484253,
      "learning_rate": 9.95471578385915e-05,
      "loss": 0.1275,
      "step": 1156
    },
    {
      "epoch": 0.04678197054393644,
      "grad_norm": 1.9087414741516113,
      "learning_rate": 9.954630107692697e-05,
      "loss": 0.2162,
      "step": 1157
    },
    {
      "epoch": 0.04682240439920345,
      "grad_norm": 6.470560073852539,
      "learning_rate": 9.954544350924042e-05,
      "loss": 0.2445,
      "step": 1158
    },
    {
      "epoch": 0.04686283825447047,
      "grad_norm": 4.896584987640381,
      "learning_rate": 9.954458513554585e-05,
      "loss": 0.2205,
      "step": 1159
    },
    {
      "epoch": 0.046903272109737486,
      "grad_norm": 3.642158269882202,
      "learning_rate": 9.954372595585719e-05,
      "loss": 0.3128,
      "step": 1160
    },
    {
      "epoch": 0.046943705965004495,
      "grad_norm": 8.89863395690918,
      "learning_rate": 9.954286597018841e-05,
      "loss": 0.2734,
      "step": 1161
    },
    {
      "epoch": 0.04698413982027151,
      "grad_norm": 8.55456829071045,
      "learning_rate": 9.954200517855353e-05,
      "loss": 0.2106,
      "step": 1162
    },
    {
      "epoch": 0.04702457367553853,
      "grad_norm": 4.720560073852539,
      "learning_rate": 9.954114358096653e-05,
      "loss": 0.2027,
      "step": 1163
    },
    {
      "epoch": 0.047065007530805544,
      "grad_norm": 5.058065891265869,
      "learning_rate": 9.954028117744145e-05,
      "loss": 0.1602,
      "step": 1164
    },
    {
      "epoch": 0.04710544138607256,
      "grad_norm": 5.031289577484131,
      "learning_rate": 9.953941796799231e-05,
      "loss": 0.1723,
      "step": 1165
    },
    {
      "epoch": 0.04714587524133958,
      "grad_norm": 12.430282592773438,
      "learning_rate": 9.953855395263314e-05,
      "loss": 0.2763,
      "step": 1166
    },
    {
      "epoch": 0.047186309096606587,
      "grad_norm": 6.080780029296875,
      "learning_rate": 9.9537689131378e-05,
      "loss": 0.1458,
      "step": 1167
    },
    {
      "epoch": 0.0472267429518736,
      "grad_norm": 9.086959838867188,
      "learning_rate": 9.953682350424096e-05,
      "loss": 0.2415,
      "step": 1168
    },
    {
      "epoch": 0.04726717680714062,
      "grad_norm": 6.597081661224365,
      "learning_rate": 9.953595707123613e-05,
      "loss": 0.2379,
      "step": 1169
    },
    {
      "epoch": 0.047307610662407636,
      "grad_norm": 9.053053855895996,
      "learning_rate": 9.953508983237755e-05,
      "loss": 0.286,
      "step": 1170
    },
    {
      "epoch": 0.04734804451767465,
      "grad_norm": 7.845701217651367,
      "learning_rate": 9.953422178767938e-05,
      "loss": 0.2303,
      "step": 1171
    },
    {
      "epoch": 0.04738847837294166,
      "grad_norm": 4.738273620605469,
      "learning_rate": 9.953335293715571e-05,
      "loss": 0.1283,
      "step": 1172
    },
    {
      "epoch": 0.04742891222820868,
      "grad_norm": 8.572137832641602,
      "learning_rate": 9.95324832808207e-05,
      "loss": 0.1439,
      "step": 1173
    },
    {
      "epoch": 0.047469346083475694,
      "grad_norm": 3.7151081562042236,
      "learning_rate": 9.953161281868846e-05,
      "loss": 0.191,
      "step": 1174
    },
    {
      "epoch": 0.04750977993874271,
      "grad_norm": 7.805776596069336,
      "learning_rate": 9.953074155077318e-05,
      "loss": 0.2161,
      "step": 1175
    },
    {
      "epoch": 0.04755021379400973,
      "grad_norm": 3.8934736251831055,
      "learning_rate": 9.952986947708904e-05,
      "loss": 0.0971,
      "step": 1176
    },
    {
      "epoch": 0.047590647649276736,
      "grad_norm": 2.9515786170959473,
      "learning_rate": 9.95289965976502e-05,
      "loss": 0.1493,
      "step": 1177
    },
    {
      "epoch": 0.04763108150454375,
      "grad_norm": 5.796713352203369,
      "learning_rate": 9.952812291247089e-05,
      "loss": 0.3267,
      "step": 1178
    },
    {
      "epoch": 0.04767151535981077,
      "grad_norm": 9.13005256652832,
      "learning_rate": 9.952724842156529e-05,
      "loss": 0.2736,
      "step": 1179
    },
    {
      "epoch": 0.047711949215077785,
      "grad_norm": 10.248169898986816,
      "learning_rate": 9.952637312494766e-05,
      "loss": 0.4568,
      "step": 1180
    },
    {
      "epoch": 0.0477523830703448,
      "grad_norm": 9.752904891967773,
      "learning_rate": 9.95254970226322e-05,
      "loss": 0.3706,
      "step": 1181
    },
    {
      "epoch": 0.04779281692561182,
      "grad_norm": 2.9924063682556152,
      "learning_rate": 9.95246201146332e-05,
      "loss": 0.1278,
      "step": 1182
    },
    {
      "epoch": 0.04783325078087883,
      "grad_norm": 4.512125015258789,
      "learning_rate": 9.952374240096491e-05,
      "loss": 0.1414,
      "step": 1183
    },
    {
      "epoch": 0.047873684636145844,
      "grad_norm": 3.4402480125427246,
      "learning_rate": 9.952286388164161e-05,
      "loss": 0.1699,
      "step": 1184
    },
    {
      "epoch": 0.04791411849141286,
      "grad_norm": 4.2754034996032715,
      "learning_rate": 9.95219845566776e-05,
      "loss": 0.2116,
      "step": 1185
    },
    {
      "epoch": 0.04795455234667988,
      "grad_norm": 7.90944766998291,
      "learning_rate": 9.952110442608716e-05,
      "loss": 0.1844,
      "step": 1186
    },
    {
      "epoch": 0.04799498620194689,
      "grad_norm": 5.320490837097168,
      "learning_rate": 9.952022348988463e-05,
      "loss": 0.1448,
      "step": 1187
    },
    {
      "epoch": 0.0480354200572139,
      "grad_norm": 9.684764862060547,
      "learning_rate": 9.951934174808435e-05,
      "loss": 0.2688,
      "step": 1188
    },
    {
      "epoch": 0.04807585391248092,
      "grad_norm": 4.744980812072754,
      "learning_rate": 9.951845920070064e-05,
      "loss": 0.1454,
      "step": 1189
    },
    {
      "epoch": 0.048116287767747935,
      "grad_norm": 14.848072052001953,
      "learning_rate": 9.951757584774785e-05,
      "loss": 0.3577,
      "step": 1190
    },
    {
      "epoch": 0.04815672162301495,
      "grad_norm": 2.2153878211975098,
      "learning_rate": 9.951669168924038e-05,
      "loss": 0.0786,
      "step": 1191
    },
    {
      "epoch": 0.04819715547828197,
      "grad_norm": 4.202341079711914,
      "learning_rate": 9.95158067251926e-05,
      "loss": 0.1205,
      "step": 1192
    },
    {
      "epoch": 0.04823758933354898,
      "grad_norm": 2.570460081100464,
      "learning_rate": 9.951492095561892e-05,
      "loss": 0.0566,
      "step": 1193
    },
    {
      "epoch": 0.048278023188815994,
      "grad_norm": 6.833673477172852,
      "learning_rate": 9.951403438053372e-05,
      "loss": 0.2755,
      "step": 1194
    },
    {
      "epoch": 0.04831845704408301,
      "grad_norm": 7.026516914367676,
      "learning_rate": 9.951314699995146e-05,
      "loss": 0.2739,
      "step": 1195
    },
    {
      "epoch": 0.04835889089935003,
      "grad_norm": 4.6357293128967285,
      "learning_rate": 9.951225881388653e-05,
      "loss": 0.159,
      "step": 1196
    },
    {
      "epoch": 0.04839932475461704,
      "grad_norm": 4.5424418449401855,
      "learning_rate": 9.951136982235342e-05,
      "loss": 0.1222,
      "step": 1197
    },
    {
      "epoch": 0.04843975860988406,
      "grad_norm": 5.4371209144592285,
      "learning_rate": 9.951048002536658e-05,
      "loss": 0.0966,
      "step": 1198
    },
    {
      "epoch": 0.04848019246515107,
      "grad_norm": 4.059628486633301,
      "learning_rate": 9.950958942294048e-05,
      "loss": 0.0994,
      "step": 1199
    },
    {
      "epoch": 0.048520626320418085,
      "grad_norm": 13.510993003845215,
      "learning_rate": 9.950869801508961e-05,
      "loss": 0.3269,
      "step": 1200
    },
    {
      "epoch": 0.0485610601756851,
      "grad_norm": 7.80584716796875,
      "learning_rate": 9.950780580182848e-05,
      "loss": 0.2649,
      "step": 1201
    },
    {
      "epoch": 0.04860149403095212,
      "grad_norm": 8.147446632385254,
      "learning_rate": 9.950691278317159e-05,
      "loss": 0.2057,
      "step": 1202
    },
    {
      "epoch": 0.048641927886219134,
      "grad_norm": 6.4793267250061035,
      "learning_rate": 9.950601895913348e-05,
      "loss": 0.2811,
      "step": 1203
    },
    {
      "epoch": 0.048682361741486144,
      "grad_norm": 3.697185516357422,
      "learning_rate": 9.950512432972867e-05,
      "loss": 0.2235,
      "step": 1204
    },
    {
      "epoch": 0.04872279559675316,
      "grad_norm": 2.8358469009399414,
      "learning_rate": 9.950422889497176e-05,
      "loss": 0.0855,
      "step": 1205
    },
    {
      "epoch": 0.048763229452020176,
      "grad_norm": 7.000943660736084,
      "learning_rate": 9.950333265487726e-05,
      "loss": 0.1717,
      "step": 1206
    },
    {
      "epoch": 0.04880366330728719,
      "grad_norm": 3.7808680534362793,
      "learning_rate": 9.950243560945978e-05,
      "loss": 0.1806,
      "step": 1207
    },
    {
      "epoch": 0.04884409716255421,
      "grad_norm": 5.245044231414795,
      "learning_rate": 9.950153775873393e-05,
      "loss": 0.1229,
      "step": 1208
    },
    {
      "epoch": 0.04888453101782122,
      "grad_norm": 3.501779556274414,
      "learning_rate": 9.950063910271427e-05,
      "loss": 0.1418,
      "step": 1209
    },
    {
      "epoch": 0.048924964873088235,
      "grad_norm": 6.328236103057861,
      "learning_rate": 9.949973964141546e-05,
      "loss": 0.2291,
      "step": 1210
    },
    {
      "epoch": 0.04896539872835525,
      "grad_norm": 3.9596774578094482,
      "learning_rate": 9.949883937485211e-05,
      "loss": 0.2302,
      "step": 1211
    },
    {
      "epoch": 0.04900583258362227,
      "grad_norm": 6.8577094078063965,
      "learning_rate": 9.949793830303888e-05,
      "loss": 0.0848,
      "step": 1212
    },
    {
      "epoch": 0.049046266438889284,
      "grad_norm": 2.952054977416992,
      "learning_rate": 9.949703642599041e-05,
      "loss": 0.1565,
      "step": 1213
    },
    {
      "epoch": 0.0490867002941563,
      "grad_norm": 5.134357929229736,
      "learning_rate": 9.94961337437214e-05,
      "loss": 0.196,
      "step": 1214
    },
    {
      "epoch": 0.04912713414942331,
      "grad_norm": 4.069969654083252,
      "learning_rate": 9.949523025624649e-05,
      "loss": 0.2352,
      "step": 1215
    },
    {
      "epoch": 0.049167568004690326,
      "grad_norm": 5.033340930938721,
      "learning_rate": 9.949432596358043e-05,
      "loss": 0.194,
      "step": 1216
    },
    {
      "epoch": 0.04920800185995734,
      "grad_norm": 8.097273826599121,
      "learning_rate": 9.949342086573792e-05,
      "loss": 0.2537,
      "step": 1217
    },
    {
      "epoch": 0.04924843571522436,
      "grad_norm": 3.7515933513641357,
      "learning_rate": 9.949251496273366e-05,
      "loss": 0.1312,
      "step": 1218
    },
    {
      "epoch": 0.049288869570491375,
      "grad_norm": 5.392459392547607,
      "learning_rate": 9.949160825458239e-05,
      "loss": 0.145,
      "step": 1219
    },
    {
      "epoch": 0.049329303425758385,
      "grad_norm": 2.515160083770752,
      "learning_rate": 9.949070074129889e-05,
      "loss": 0.0773,
      "step": 1220
    },
    {
      "epoch": 0.0493697372810254,
      "grad_norm": 11.98266315460205,
      "learning_rate": 9.948979242289788e-05,
      "loss": 0.2075,
      "step": 1221
    },
    {
      "epoch": 0.04941017113629242,
      "grad_norm": 3.602379560470581,
      "learning_rate": 9.948888329939418e-05,
      "loss": 0.1912,
      "step": 1222
    },
    {
      "epoch": 0.049450604991559434,
      "grad_norm": 10.972463607788086,
      "learning_rate": 9.948797337080254e-05,
      "loss": 0.2325,
      "step": 1223
    },
    {
      "epoch": 0.04949103884682645,
      "grad_norm": 5.439568996429443,
      "learning_rate": 9.94870626371378e-05,
      "loss": 0.1821,
      "step": 1224
    },
    {
      "epoch": 0.04953147270209346,
      "grad_norm": 7.492272853851318,
      "learning_rate": 9.948615109841475e-05,
      "loss": 0.2873,
      "step": 1225
    },
    {
      "epoch": 0.049571906557360476,
      "grad_norm": 6.384400367736816,
      "learning_rate": 9.948523875464824e-05,
      "loss": 0.2753,
      "step": 1226
    },
    {
      "epoch": 0.04961234041262749,
      "grad_norm": 8.149813652038574,
      "learning_rate": 9.948432560585308e-05,
      "loss": 0.3669,
      "step": 1227
    },
    {
      "epoch": 0.04965277426789451,
      "grad_norm": 7.409963130950928,
      "learning_rate": 9.948341165204414e-05,
      "loss": 0.2233,
      "step": 1228
    },
    {
      "epoch": 0.049693208123161525,
      "grad_norm": 6.286019802093506,
      "learning_rate": 9.948249689323631e-05,
      "loss": 0.1972,
      "step": 1229
    },
    {
      "epoch": 0.04973364197842854,
      "grad_norm": 7.73862361907959,
      "learning_rate": 9.948158132944445e-05,
      "loss": 0.2831,
      "step": 1230
    },
    {
      "epoch": 0.04977407583369555,
      "grad_norm": 11.159944534301758,
      "learning_rate": 9.948066496068346e-05,
      "loss": 0.2571,
      "step": 1231
    },
    {
      "epoch": 0.04981450968896257,
      "grad_norm": 8.0750150680542,
      "learning_rate": 9.947974778696824e-05,
      "loss": 0.2255,
      "step": 1232
    },
    {
      "epoch": 0.049854943544229584,
      "grad_norm": 2.8113396167755127,
      "learning_rate": 9.947882980831371e-05,
      "loss": 0.2408,
      "step": 1233
    },
    {
      "epoch": 0.0498953773994966,
      "grad_norm": 5.652257919311523,
      "learning_rate": 9.947791102473483e-05,
      "loss": 0.1866,
      "step": 1234
    },
    {
      "epoch": 0.04993581125476362,
      "grad_norm": 5.821153163909912,
      "learning_rate": 9.947699143624652e-05,
      "loss": 0.2013,
      "step": 1235
    },
    {
      "epoch": 0.049976245110030626,
      "grad_norm": 9.115815162658691,
      "learning_rate": 9.947607104286374e-05,
      "loss": 0.2376,
      "step": 1236
    },
    {
      "epoch": 0.05001667896529764,
      "grad_norm": 8.036824226379395,
      "learning_rate": 9.947514984460147e-05,
      "loss": 0.2177,
      "step": 1237
    },
    {
      "epoch": 0.05005711282056466,
      "grad_norm": 9.01035213470459,
      "learning_rate": 9.94742278414747e-05,
      "loss": 0.2127,
      "step": 1238
    },
    {
      "epoch": 0.050097546675831675,
      "grad_norm": 6.903233051300049,
      "learning_rate": 9.947330503349842e-05,
      "loss": 0.21,
      "step": 1239
    },
    {
      "epoch": 0.05013798053109869,
      "grad_norm": 5.857448577880859,
      "learning_rate": 9.947238142068765e-05,
      "loss": 0.1196,
      "step": 1240
    },
    {
      "epoch": 0.0501784143863657,
      "grad_norm": 7.502307891845703,
      "learning_rate": 9.947145700305742e-05,
      "loss": 0.2402,
      "step": 1241
    },
    {
      "epoch": 0.05021884824163272,
      "grad_norm": 4.341075420379639,
      "learning_rate": 9.947053178062276e-05,
      "loss": 0.3828,
      "step": 1242
    },
    {
      "epoch": 0.050259282096899734,
      "grad_norm": 3.0865466594696045,
      "learning_rate": 9.946960575339871e-05,
      "loss": 0.1313,
      "step": 1243
    },
    {
      "epoch": 0.05029971595216675,
      "grad_norm": 4.652061462402344,
      "learning_rate": 9.946867892140036e-05,
      "loss": 0.156,
      "step": 1244
    },
    {
      "epoch": 0.050340149807433766,
      "grad_norm": 3.4580812454223633,
      "learning_rate": 9.946775128464278e-05,
      "loss": 0.1724,
      "step": 1245
    },
    {
      "epoch": 0.05038058366270078,
      "grad_norm": 2.100170373916626,
      "learning_rate": 9.946682284314104e-05,
      "loss": 0.0945,
      "step": 1246
    },
    {
      "epoch": 0.05042101751796779,
      "grad_norm": 5.890810489654541,
      "learning_rate": 9.946589359691028e-05,
      "loss": 0.1858,
      "step": 1247
    },
    {
      "epoch": 0.05046145137323481,
      "grad_norm": 3.0006163120269775,
      "learning_rate": 9.946496354596557e-05,
      "loss": 0.0841,
      "step": 1248
    },
    {
      "epoch": 0.050501885228501825,
      "grad_norm": 4.3676676750183105,
      "learning_rate": 9.94640326903221e-05,
      "loss": 0.2198,
      "step": 1249
    },
    {
      "epoch": 0.05054231908376884,
      "grad_norm": 8.32162857055664,
      "learning_rate": 9.946310102999496e-05,
      "loss": 0.3599,
      "step": 1250
    },
    {
      "epoch": 0.05058275293903586,
      "grad_norm": 12.200478553771973,
      "learning_rate": 9.946216856499932e-05,
      "loss": 0.3395,
      "step": 1251
    },
    {
      "epoch": 0.05062318679430287,
      "grad_norm": 4.876917839050293,
      "learning_rate": 9.946123529535039e-05,
      "loss": 0.1994,
      "step": 1252
    },
    {
      "epoch": 0.050663620649569883,
      "grad_norm": 4.037123680114746,
      "learning_rate": 9.94603012210633e-05,
      "loss": 0.1364,
      "step": 1253
    },
    {
      "epoch": 0.0507040545048369,
      "grad_norm": 3.035832166671753,
      "learning_rate": 9.945936634215325e-05,
      "loss": 0.1265,
      "step": 1254
    },
    {
      "epoch": 0.050744488360103916,
      "grad_norm": 5.721221446990967,
      "learning_rate": 9.945843065863547e-05,
      "loss": 0.2392,
      "step": 1255
    },
    {
      "epoch": 0.05078492221537093,
      "grad_norm": 6.27882194519043,
      "learning_rate": 9.945749417052517e-05,
      "loss": 0.1599,
      "step": 1256
    },
    {
      "epoch": 0.05082535607063794,
      "grad_norm": 2.9520926475524902,
      "learning_rate": 9.94565568778376e-05,
      "loss": 0.1381,
      "step": 1257
    },
    {
      "epoch": 0.05086578992590496,
      "grad_norm": 2.3456265926361084,
      "learning_rate": 9.9455618780588e-05,
      "loss": 0.1471,
      "step": 1258
    },
    {
      "epoch": 0.050906223781171975,
      "grad_norm": 3.6223018169403076,
      "learning_rate": 9.945467987879163e-05,
      "loss": 0.1794,
      "step": 1259
    },
    {
      "epoch": 0.05094665763643899,
      "grad_norm": 10.348517417907715,
      "learning_rate": 9.945374017246376e-05,
      "loss": 0.209,
      "step": 1260
    },
    {
      "epoch": 0.05098709149170601,
      "grad_norm": 5.132821559906006,
      "learning_rate": 9.945279966161967e-05,
      "loss": 0.2776,
      "step": 1261
    },
    {
      "epoch": 0.051027525346973024,
      "grad_norm": 4.420148849487305,
      "learning_rate": 9.945185834627468e-05,
      "loss": 0.2753,
      "step": 1262
    },
    {
      "epoch": 0.05106795920224003,
      "grad_norm": 5.077025413513184,
      "learning_rate": 9.945091622644408e-05,
      "loss": 0.1922,
      "step": 1263
    },
    {
      "epoch": 0.05110839305750705,
      "grad_norm": 6.091835975646973,
      "learning_rate": 9.944997330214323e-05,
      "loss": 0.1633,
      "step": 1264
    },
    {
      "epoch": 0.051148826912774066,
      "grad_norm": 3.8139922618865967,
      "learning_rate": 9.944902957338744e-05,
      "loss": 0.2656,
      "step": 1265
    },
    {
      "epoch": 0.05118926076804108,
      "grad_norm": 9.896492958068848,
      "learning_rate": 9.944808504019208e-05,
      "loss": 0.2268,
      "step": 1266
    },
    {
      "epoch": 0.0512296946233081,
      "grad_norm": 4.465797424316406,
      "learning_rate": 9.944713970257249e-05,
      "loss": 0.183,
      "step": 1267
    },
    {
      "epoch": 0.05127012847857511,
      "grad_norm": 3.4332938194274902,
      "learning_rate": 9.944619356054408e-05,
      "loss": 0.0861,
      "step": 1268
    },
    {
      "epoch": 0.051310562333842125,
      "grad_norm": 3.203334093093872,
      "learning_rate": 9.944524661412222e-05,
      "loss": 0.1476,
      "step": 1269
    },
    {
      "epoch": 0.05135099618910914,
      "grad_norm": 5.330435276031494,
      "learning_rate": 9.944429886332233e-05,
      "loss": 0.2245,
      "step": 1270
    },
    {
      "epoch": 0.05139143004437616,
      "grad_norm": 5.236692905426025,
      "learning_rate": 9.944335030815982e-05,
      "loss": 0.1653,
      "step": 1271
    },
    {
      "epoch": 0.051431863899643174,
      "grad_norm": 4.078321933746338,
      "learning_rate": 9.944240094865012e-05,
      "loss": 0.1038,
      "step": 1272
    },
    {
      "epoch": 0.05147229775491018,
      "grad_norm": 5.534811496734619,
      "learning_rate": 9.944145078480869e-05,
      "loss": 0.2213,
      "step": 1273
    },
    {
      "epoch": 0.0515127316101772,
      "grad_norm": 6.796341419219971,
      "learning_rate": 9.944049981665095e-05,
      "loss": 0.1793,
      "step": 1274
    },
    {
      "epoch": 0.051553165465444216,
      "grad_norm": 7.087914943695068,
      "learning_rate": 9.943954804419242e-05,
      "loss": 0.1221,
      "step": 1275
    },
    {
      "epoch": 0.05159359932071123,
      "grad_norm": 4.245282173156738,
      "learning_rate": 9.943859546744854e-05,
      "loss": 0.1277,
      "step": 1276
    },
    {
      "epoch": 0.05163403317597825,
      "grad_norm": 3.097053050994873,
      "learning_rate": 9.943764208643483e-05,
      "loss": 0.139,
      "step": 1277
    },
    {
      "epoch": 0.051674467031245265,
      "grad_norm": 2.386692762374878,
      "learning_rate": 9.94366879011668e-05,
      "loss": 0.1581,
      "step": 1278
    },
    {
      "epoch": 0.051714900886512274,
      "grad_norm": 9.312906265258789,
      "learning_rate": 9.943573291165998e-05,
      "loss": 0.2343,
      "step": 1279
    },
    {
      "epoch": 0.05175533474177929,
      "grad_norm": 5.931743621826172,
      "learning_rate": 9.943477711792986e-05,
      "loss": 0.1267,
      "step": 1280
    },
    {
      "epoch": 0.05179576859704631,
      "grad_norm": 3.2452392578125,
      "learning_rate": 9.943382051999203e-05,
      "loss": 0.155,
      "step": 1281
    },
    {
      "epoch": 0.051836202452313324,
      "grad_norm": 10.063560485839844,
      "learning_rate": 9.943286311786206e-05,
      "loss": 0.3404,
      "step": 1282
    },
    {
      "epoch": 0.05187663630758034,
      "grad_norm": 12.447166442871094,
      "learning_rate": 9.943190491155551e-05,
      "loss": 0.2858,
      "step": 1283
    },
    {
      "epoch": 0.05191707016284735,
      "grad_norm": 8.46149730682373,
      "learning_rate": 9.943094590108794e-05,
      "loss": 0.3089,
      "step": 1284
    },
    {
      "epoch": 0.051957504018114366,
      "grad_norm": 5.808350563049316,
      "learning_rate": 9.9429986086475e-05,
      "loss": 0.1632,
      "step": 1285
    },
    {
      "epoch": 0.05199793787338138,
      "grad_norm": 6.949563503265381,
      "learning_rate": 9.942902546773229e-05,
      "loss": 0.1995,
      "step": 1286
    },
    {
      "epoch": 0.0520383717286484,
      "grad_norm": 7.136388778686523,
      "learning_rate": 9.942806404487541e-05,
      "loss": 0.2824,
      "step": 1287
    },
    {
      "epoch": 0.052078805583915415,
      "grad_norm": 3.145890951156616,
      "learning_rate": 9.942710181792003e-05,
      "loss": 0.1085,
      "step": 1288
    },
    {
      "epoch": 0.052119239439182424,
      "grad_norm": 2.9668140411376953,
      "learning_rate": 9.94261387868818e-05,
      "loss": 0.1016,
      "step": 1289
    },
    {
      "epoch": 0.05215967329444944,
      "grad_norm": 5.163139343261719,
      "learning_rate": 9.942517495177638e-05,
      "loss": 0.1244,
      "step": 1290
    },
    {
      "epoch": 0.05220010714971646,
      "grad_norm": 6.910492420196533,
      "learning_rate": 9.942421031261944e-05,
      "loss": 0.1853,
      "step": 1291
    },
    {
      "epoch": 0.05224054100498347,
      "grad_norm": 14.92756175994873,
      "learning_rate": 9.94232448694267e-05,
      "loss": 0.1963,
      "step": 1292
    },
    {
      "epoch": 0.05228097486025049,
      "grad_norm": 10.046538352966309,
      "learning_rate": 9.942227862221383e-05,
      "loss": 0.29,
      "step": 1293
    },
    {
      "epoch": 0.052321408715517506,
      "grad_norm": 8.670050621032715,
      "learning_rate": 9.942131157099656e-05,
      "loss": 0.3553,
      "step": 1294
    },
    {
      "epoch": 0.052361842570784516,
      "grad_norm": 9.415162086486816,
      "learning_rate": 9.942034371579066e-05,
      "loss": 0.3174,
      "step": 1295
    },
    {
      "epoch": 0.05240227642605153,
      "grad_norm": 7.141231536865234,
      "learning_rate": 9.941937505661183e-05,
      "loss": 0.3125,
      "step": 1296
    },
    {
      "epoch": 0.05244271028131855,
      "grad_norm": 3.3727052211761475,
      "learning_rate": 9.941840559347585e-05,
      "loss": 0.1546,
      "step": 1297
    },
    {
      "epoch": 0.052483144136585565,
      "grad_norm": 4.992310523986816,
      "learning_rate": 9.941743532639848e-05,
      "loss": 0.125,
      "step": 1298
    },
    {
      "epoch": 0.05252357799185258,
      "grad_norm": 4.586063385009766,
      "learning_rate": 9.94164642553955e-05,
      "loss": 0.1318,
      "step": 1299
    },
    {
      "epoch": 0.05256401184711959,
      "grad_norm": 3.372098445892334,
      "learning_rate": 9.941549238048274e-05,
      "loss": 0.1239,
      "step": 1300
    },
    {
      "epoch": 0.05260444570238661,
      "grad_norm": 7.907415866851807,
      "learning_rate": 9.941451970167596e-05,
      "loss": 0.2554,
      "step": 1301
    },
    {
      "epoch": 0.05264487955765362,
      "grad_norm": 10.8170166015625,
      "learning_rate": 9.941354621899102e-05,
      "loss": 0.2321,
      "step": 1302
    },
    {
      "epoch": 0.05268531341292064,
      "grad_norm": 4.584695816040039,
      "learning_rate": 9.941257193244375e-05,
      "loss": 0.1961,
      "step": 1303
    },
    {
      "epoch": 0.052725747268187656,
      "grad_norm": 5.531389236450195,
      "learning_rate": 9.941159684205e-05,
      "loss": 0.1622,
      "step": 1304
    },
    {
      "epoch": 0.052766181123454665,
      "grad_norm": 2.5804152488708496,
      "learning_rate": 9.941062094782563e-05,
      "loss": 0.2143,
      "step": 1305
    },
    {
      "epoch": 0.05280661497872168,
      "grad_norm": 6.512624263763428,
      "learning_rate": 9.940964424978652e-05,
      "loss": 0.2652,
      "step": 1306
    },
    {
      "epoch": 0.0528470488339887,
      "grad_norm": 4.737202167510986,
      "learning_rate": 9.940866674794854e-05,
      "loss": 0.2276,
      "step": 1307
    },
    {
      "epoch": 0.052887482689255715,
      "grad_norm": 7.695554733276367,
      "learning_rate": 9.940768844232762e-05,
      "loss": 0.2358,
      "step": 1308
    },
    {
      "epoch": 0.05292791654452273,
      "grad_norm": 5.7882890701293945,
      "learning_rate": 9.940670933293965e-05,
      "loss": 0.2418,
      "step": 1309
    },
    {
      "epoch": 0.05296835039978975,
      "grad_norm": 8.109101295471191,
      "learning_rate": 9.940572941980058e-05,
      "loss": 0.1771,
      "step": 1310
    },
    {
      "epoch": 0.05300878425505676,
      "grad_norm": 2.8940999507904053,
      "learning_rate": 9.940474870292633e-05,
      "loss": 0.12,
      "step": 1311
    },
    {
      "epoch": 0.05304921811032377,
      "grad_norm": 8.631229400634766,
      "learning_rate": 9.940376718233289e-05,
      "loss": 0.2241,
      "step": 1312
    },
    {
      "epoch": 0.05308965196559079,
      "grad_norm": 8.017901420593262,
      "learning_rate": 9.940278485803618e-05,
      "loss": 0.2108,
      "step": 1313
    },
    {
      "epoch": 0.053130085820857806,
      "grad_norm": 4.772304058074951,
      "learning_rate": 9.94018017300522e-05,
      "loss": 0.1923,
      "step": 1314
    },
    {
      "epoch": 0.05317051967612482,
      "grad_norm": 3.255382537841797,
      "learning_rate": 9.940081779839696e-05,
      "loss": 0.1595,
      "step": 1315
    },
    {
      "epoch": 0.05321095353139183,
      "grad_norm": 3.020578145980835,
      "learning_rate": 9.939983306308647e-05,
      "loss": 0.0613,
      "step": 1316
    },
    {
      "epoch": 0.05325138738665885,
      "grad_norm": 5.327719211578369,
      "learning_rate": 9.939884752413671e-05,
      "loss": 0.1881,
      "step": 1317
    },
    {
      "epoch": 0.053291821241925864,
      "grad_norm": 3.247042655944824,
      "learning_rate": 9.939786118156373e-05,
      "loss": 0.2125,
      "step": 1318
    },
    {
      "epoch": 0.05333225509719288,
      "grad_norm": 5.178999423980713,
      "learning_rate": 9.93968740353836e-05,
      "loss": 0.2196,
      "step": 1319
    },
    {
      "epoch": 0.0533726889524599,
      "grad_norm": 3.9060566425323486,
      "learning_rate": 9.939588608561236e-05,
      "loss": 0.1556,
      "step": 1320
    },
    {
      "epoch": 0.05341312280772691,
      "grad_norm": 4.775238990783691,
      "learning_rate": 9.939489733226607e-05,
      "loss": 0.1472,
      "step": 1321
    },
    {
      "epoch": 0.05345355666299392,
      "grad_norm": 2.8667566776275635,
      "learning_rate": 9.939390777536085e-05,
      "loss": 0.1562,
      "step": 1322
    },
    {
      "epoch": 0.05349399051826094,
      "grad_norm": 4.5102715492248535,
      "learning_rate": 9.939291741491277e-05,
      "loss": 0.1165,
      "step": 1323
    },
    {
      "epoch": 0.053534424373527956,
      "grad_norm": 5.313643932342529,
      "learning_rate": 9.939192625093794e-05,
      "loss": 0.2245,
      "step": 1324
    },
    {
      "epoch": 0.05357485822879497,
      "grad_norm": 3.3393213748931885,
      "learning_rate": 9.939093428345249e-05,
      "loss": 0.1723,
      "step": 1325
    },
    {
      "epoch": 0.05361529208406199,
      "grad_norm": 5.78945779800415,
      "learning_rate": 9.938994151247257e-05,
      "loss": 0.1034,
      "step": 1326
    },
    {
      "epoch": 0.053655725939329,
      "grad_norm": 11.20936107635498,
      "learning_rate": 9.938894793801432e-05,
      "loss": 0.2913,
      "step": 1327
    },
    {
      "epoch": 0.053696159794596014,
      "grad_norm": 14.034457206726074,
      "learning_rate": 9.93879535600939e-05,
      "loss": 0.3247,
      "step": 1328
    },
    {
      "epoch": 0.05373659364986303,
      "grad_norm": 7.630880832672119,
      "learning_rate": 9.938695837872748e-05,
      "loss": 0.2317,
      "step": 1329
    },
    {
      "epoch": 0.05377702750513005,
      "grad_norm": 6.568752765655518,
      "learning_rate": 9.938596239393127e-05,
      "loss": 0.2567,
      "step": 1330
    },
    {
      "epoch": 0.05381746136039706,
      "grad_norm": 4.499589443206787,
      "learning_rate": 9.938496560572146e-05,
      "loss": 0.2225,
      "step": 1331
    },
    {
      "epoch": 0.05385789521566407,
      "grad_norm": 6.810868740081787,
      "learning_rate": 9.938396801411427e-05,
      "loss": 0.2466,
      "step": 1332
    },
    {
      "epoch": 0.05389832907093109,
      "grad_norm": 3.2281811237335205,
      "learning_rate": 9.938296961912592e-05,
      "loss": 0.154,
      "step": 1333
    },
    {
      "epoch": 0.053938762926198106,
      "grad_norm": 6.490049362182617,
      "learning_rate": 9.938197042077266e-05,
      "loss": 0.2935,
      "step": 1334
    },
    {
      "epoch": 0.05397919678146512,
      "grad_norm": 6.5003437995910645,
      "learning_rate": 9.938097041907076e-05,
      "loss": 0.3401,
      "step": 1335
    },
    {
      "epoch": 0.05401963063673214,
      "grad_norm": 5.451622486114502,
      "learning_rate": 9.937996961403647e-05,
      "loss": 0.2344,
      "step": 1336
    },
    {
      "epoch": 0.05406006449199915,
      "grad_norm": 4.084493637084961,
      "learning_rate": 9.937896800568606e-05,
      "loss": 0.18,
      "step": 1337
    },
    {
      "epoch": 0.054100498347266164,
      "grad_norm": 7.304638862609863,
      "learning_rate": 9.937796559403583e-05,
      "loss": 0.2012,
      "step": 1338
    },
    {
      "epoch": 0.05414093220253318,
      "grad_norm": 4.9891533851623535,
      "learning_rate": 9.937696237910213e-05,
      "loss": 0.2372,
      "step": 1339
    },
    {
      "epoch": 0.0541813660578002,
      "grad_norm": 6.495491981506348,
      "learning_rate": 9.93759583609012e-05,
      "loss": 0.2297,
      "step": 1340
    },
    {
      "epoch": 0.05422179991306721,
      "grad_norm": 7.487184047698975,
      "learning_rate": 9.937495353944945e-05,
      "loss": 0.3075,
      "step": 1341
    },
    {
      "epoch": 0.05426223376833423,
      "grad_norm": 4.257822036743164,
      "learning_rate": 9.937394791476319e-05,
      "loss": 0.198,
      "step": 1342
    },
    {
      "epoch": 0.05430266762360124,
      "grad_norm": 3.622641086578369,
      "learning_rate": 9.937294148685878e-05,
      "loss": 0.1591,
      "step": 1343
    },
    {
      "epoch": 0.054343101478868255,
      "grad_norm": 3.4544076919555664,
      "learning_rate": 9.93719342557526e-05,
      "loss": 0.1293,
      "step": 1344
    },
    {
      "epoch": 0.05438353533413527,
      "grad_norm": 9.25291633605957,
      "learning_rate": 9.937092622146103e-05,
      "loss": 0.2475,
      "step": 1345
    },
    {
      "epoch": 0.05442396918940229,
      "grad_norm": 5.649351119995117,
      "learning_rate": 9.936991738400047e-05,
      "loss": 0.1677,
      "step": 1346
    },
    {
      "epoch": 0.054464403044669304,
      "grad_norm": 18.042448043823242,
      "learning_rate": 9.936890774338734e-05,
      "loss": 0.382,
      "step": 1347
    },
    {
      "epoch": 0.054504836899936314,
      "grad_norm": 6.543587684631348,
      "learning_rate": 9.936789729963807e-05,
      "loss": 0.1783,
      "step": 1348
    },
    {
      "epoch": 0.05454527075520333,
      "grad_norm": 11.179071426391602,
      "learning_rate": 9.936688605276908e-05,
      "loss": 0.3154,
      "step": 1349
    },
    {
      "epoch": 0.05458570461047035,
      "grad_norm": 8.762697219848633,
      "learning_rate": 9.93658740027968e-05,
      "loss": 0.2796,
      "step": 1350
    },
    {
      "epoch": 0.05462613846573736,
      "grad_norm": 8.139274597167969,
      "learning_rate": 9.936486114973775e-05,
      "loss": 0.2199,
      "step": 1351
    },
    {
      "epoch": 0.05466657232100438,
      "grad_norm": 3.6884829998016357,
      "learning_rate": 9.936384749360836e-05,
      "loss": 0.2658,
      "step": 1352
    },
    {
      "epoch": 0.05470700617627139,
      "grad_norm": 4.405956268310547,
      "learning_rate": 9.936283303442515e-05,
      "loss": 0.1674,
      "step": 1353
    },
    {
      "epoch": 0.054747440031538405,
      "grad_norm": 10.997044563293457,
      "learning_rate": 9.936181777220461e-05,
      "loss": 0.2533,
      "step": 1354
    },
    {
      "epoch": 0.05478787388680542,
      "grad_norm": 8.021032333374023,
      "learning_rate": 9.936080170696326e-05,
      "loss": 0.3133,
      "step": 1355
    },
    {
      "epoch": 0.05482830774207244,
      "grad_norm": 8.427315711975098,
      "learning_rate": 9.935978483871764e-05,
      "loss": 0.2121,
      "step": 1356
    },
    {
      "epoch": 0.054868741597339454,
      "grad_norm": 5.5175700187683105,
      "learning_rate": 9.935876716748426e-05,
      "loss": 0.1882,
      "step": 1357
    },
    {
      "epoch": 0.05490917545260647,
      "grad_norm": 2.882260322570801,
      "learning_rate": 9.935774869327971e-05,
      "loss": 0.1694,
      "step": 1358
    },
    {
      "epoch": 0.05494960930787348,
      "grad_norm": 4.902988910675049,
      "learning_rate": 9.935672941612054e-05,
      "loss": 0.1528,
      "step": 1359
    },
    {
      "epoch": 0.054990043163140497,
      "grad_norm": 7.045770645141602,
      "learning_rate": 9.935570933602334e-05,
      "loss": 0.3137,
      "step": 1360
    },
    {
      "epoch": 0.05503047701840751,
      "grad_norm": 5.428627967834473,
      "learning_rate": 9.93546884530047e-05,
      "loss": 0.1526,
      "step": 1361
    },
    {
      "epoch": 0.05507091087367453,
      "grad_norm": 10.377058029174805,
      "learning_rate": 9.935366676708123e-05,
      "loss": 0.3017,
      "step": 1362
    },
    {
      "epoch": 0.055111344728941546,
      "grad_norm": 3.935405969619751,
      "learning_rate": 9.935264427826954e-05,
      "loss": 0.2563,
      "step": 1363
    },
    {
      "epoch": 0.055151778584208555,
      "grad_norm": 7.441375255584717,
      "learning_rate": 9.935162098658628e-05,
      "loss": 0.2906,
      "step": 1364
    },
    {
      "epoch": 0.05519221243947557,
      "grad_norm": 3.211361885070801,
      "learning_rate": 9.935059689204809e-05,
      "loss": 0.2168,
      "step": 1365
    },
    {
      "epoch": 0.05523264629474259,
      "grad_norm": 6.161634922027588,
      "learning_rate": 9.934957199467163e-05,
      "loss": 0.2579,
      "step": 1366
    },
    {
      "epoch": 0.055273080150009604,
      "grad_norm": 3.7721683979034424,
      "learning_rate": 9.934854629447356e-05,
      "loss": 0.1438,
      "step": 1367
    },
    {
      "epoch": 0.05531351400527662,
      "grad_norm": 7.3119001388549805,
      "learning_rate": 9.93475197914706e-05,
      "loss": 0.177,
      "step": 1368
    },
    {
      "epoch": 0.05535394786054363,
      "grad_norm": 3.832587957382202,
      "learning_rate": 9.934649248567941e-05,
      "loss": 0.185,
      "step": 1369
    },
    {
      "epoch": 0.055394381715810646,
      "grad_norm": 10.597411155700684,
      "learning_rate": 9.934546437711675e-05,
      "loss": 0.2543,
      "step": 1370
    },
    {
      "epoch": 0.05543481557107766,
      "grad_norm": 8.365492820739746,
      "learning_rate": 9.93444354657993e-05,
      "loss": 0.2138,
      "step": 1371
    },
    {
      "epoch": 0.05547524942634468,
      "grad_norm": 5.45442533493042,
      "learning_rate": 9.934340575174381e-05,
      "loss": 0.2551,
      "step": 1372
    },
    {
      "epoch": 0.055515683281611695,
      "grad_norm": 5.643710613250732,
      "learning_rate": 9.934237523496704e-05,
      "loss": 0.3164,
      "step": 1373
    },
    {
      "epoch": 0.05555611713687871,
      "grad_norm": 1.7948371171951294,
      "learning_rate": 9.934134391548575e-05,
      "loss": 0.0907,
      "step": 1374
    },
    {
      "epoch": 0.05559655099214572,
      "grad_norm": 3.23000431060791,
      "learning_rate": 9.934031179331673e-05,
      "loss": 0.1564,
      "step": 1375
    },
    {
      "epoch": 0.05563698484741274,
      "grad_norm": 4.509960651397705,
      "learning_rate": 9.933927886847674e-05,
      "loss": 0.1465,
      "step": 1376
    },
    {
      "epoch": 0.055677418702679754,
      "grad_norm": 9.073298454284668,
      "learning_rate": 9.933824514098262e-05,
      "loss": 0.3212,
      "step": 1377
    },
    {
      "epoch": 0.05571785255794677,
      "grad_norm": 11.130426406860352,
      "learning_rate": 9.933721061085116e-05,
      "loss": 0.3158,
      "step": 1378
    },
    {
      "epoch": 0.05575828641321379,
      "grad_norm": 7.153498649597168,
      "learning_rate": 9.93361752780992e-05,
      "loss": 0.2233,
      "step": 1379
    },
    {
      "epoch": 0.055798720268480796,
      "grad_norm": 4.102141380310059,
      "learning_rate": 9.933513914274359e-05,
      "loss": 0.2191,
      "step": 1380
    },
    {
      "epoch": 0.05583915412374781,
      "grad_norm": 4.483169078826904,
      "learning_rate": 9.933410220480118e-05,
      "loss": 0.1579,
      "step": 1381
    },
    {
      "epoch": 0.05587958797901483,
      "grad_norm": 6.991993427276611,
      "learning_rate": 9.933306446428881e-05,
      "loss": 0.3283,
      "step": 1382
    },
    {
      "epoch": 0.055920021834281845,
      "grad_norm": 6.751758098602295,
      "learning_rate": 9.933202592122341e-05,
      "loss": 0.2409,
      "step": 1383
    },
    {
      "epoch": 0.05596045568954886,
      "grad_norm": 4.376112461090088,
      "learning_rate": 9.933098657562186e-05,
      "loss": 0.1882,
      "step": 1384
    },
    {
      "epoch": 0.05600088954481587,
      "grad_norm": 3.77413010597229,
      "learning_rate": 9.932994642750104e-05,
      "loss": 0.1365,
      "step": 1385
    },
    {
      "epoch": 0.05604132340008289,
      "grad_norm": 5.57158899307251,
      "learning_rate": 9.932890547687792e-05,
      "loss": 0.3598,
      "step": 1386
    },
    {
      "epoch": 0.056081757255349904,
      "grad_norm": 4.050318717956543,
      "learning_rate": 9.932786372376939e-05,
      "loss": 0.191,
      "step": 1387
    },
    {
      "epoch": 0.05612219111061692,
      "grad_norm": 8.976958274841309,
      "learning_rate": 9.932682116819242e-05,
      "loss": 0.3214,
      "step": 1388
    },
    {
      "epoch": 0.05616262496588394,
      "grad_norm": 11.22439193725586,
      "learning_rate": 9.932577781016396e-05,
      "loss": 0.3312,
      "step": 1389
    },
    {
      "epoch": 0.05620305882115095,
      "grad_norm": 8.162606239318848,
      "learning_rate": 9.9324733649701e-05,
      "loss": 0.2167,
      "step": 1390
    },
    {
      "epoch": 0.05624349267641796,
      "grad_norm": 8.362516403198242,
      "learning_rate": 9.93236886868205e-05,
      "loss": 0.2022,
      "step": 1391
    },
    {
      "epoch": 0.05628392653168498,
      "grad_norm": 10.072547912597656,
      "learning_rate": 9.932264292153949e-05,
      "loss": 0.2989,
      "step": 1392
    },
    {
      "epoch": 0.056324360386951995,
      "grad_norm": 4.424838066101074,
      "learning_rate": 9.932159635387497e-05,
      "loss": 0.1785,
      "step": 1393
    },
    {
      "epoch": 0.05636479424221901,
      "grad_norm": 8.446402549743652,
      "learning_rate": 9.932054898384396e-05,
      "loss": 0.1676,
      "step": 1394
    },
    {
      "epoch": 0.05640522809748603,
      "grad_norm": 2.1356959342956543,
      "learning_rate": 9.931950081146348e-05,
      "loss": 0.135,
      "step": 1395
    },
    {
      "epoch": 0.05644566195275304,
      "grad_norm": 4.658906936645508,
      "learning_rate": 9.931845183675062e-05,
      "loss": 0.2174,
      "step": 1396
    },
    {
      "epoch": 0.056486095808020054,
      "grad_norm": 3.412038803100586,
      "learning_rate": 9.931740205972241e-05,
      "loss": 0.1073,
      "step": 1397
    },
    {
      "epoch": 0.05652652966328707,
      "grad_norm": 9.804624557495117,
      "learning_rate": 9.931635148039597e-05,
      "loss": 0.2256,
      "step": 1398
    },
    {
      "epoch": 0.056566963518554086,
      "grad_norm": 11.733297348022461,
      "learning_rate": 9.931530009878837e-05,
      "loss": 0.2831,
      "step": 1399
    },
    {
      "epoch": 0.0566073973738211,
      "grad_norm": 8.707399368286133,
      "learning_rate": 9.931424791491671e-05,
      "loss": 0.233,
      "step": 1400
    },
    {
      "epoch": 0.05664783122908811,
      "grad_norm": 8.939682960510254,
      "learning_rate": 9.93131949287981e-05,
      "loss": 0.2899,
      "step": 1401
    },
    {
      "epoch": 0.05668826508435513,
      "grad_norm": 7.7383131980896,
      "learning_rate": 9.931214114044967e-05,
      "loss": 0.2409,
      "step": 1402
    },
    {
      "epoch": 0.056728698939622145,
      "grad_norm": 2.5775835514068604,
      "learning_rate": 9.931108654988857e-05,
      "loss": 0.1646,
      "step": 1403
    },
    {
      "epoch": 0.05676913279488916,
      "grad_norm": 3.9005651473999023,
      "learning_rate": 9.931003115713195e-05,
      "loss": 0.1417,
      "step": 1404
    },
    {
      "epoch": 0.05680956665015618,
      "grad_norm": 4.328968524932861,
      "learning_rate": 9.930897496219701e-05,
      "loss": 0.1679,
      "step": 1405
    },
    {
      "epoch": 0.056850000505423194,
      "grad_norm": 5.930502891540527,
      "learning_rate": 9.930791796510091e-05,
      "loss": 0.1942,
      "step": 1406
    },
    {
      "epoch": 0.056890434360690204,
      "grad_norm": 7.5674614906311035,
      "learning_rate": 9.930686016586082e-05,
      "loss": 0.1954,
      "step": 1407
    },
    {
      "epoch": 0.05693086821595722,
      "grad_norm": 8.184088706970215,
      "learning_rate": 9.930580156449398e-05,
      "loss": 0.2711,
      "step": 1408
    },
    {
      "epoch": 0.056971302071224236,
      "grad_norm": 5.0398268699646,
      "learning_rate": 9.930474216101762e-05,
      "loss": 0.1167,
      "step": 1409
    },
    {
      "epoch": 0.05701173592649125,
      "grad_norm": 5.753417015075684,
      "learning_rate": 9.930368195544893e-05,
      "loss": 0.1611,
      "step": 1410
    },
    {
      "epoch": 0.05705216978175827,
      "grad_norm": 4.066885948181152,
      "learning_rate": 9.93026209478052e-05,
      "loss": 0.099,
      "step": 1411
    },
    {
      "epoch": 0.05709260363702528,
      "grad_norm": 5.4626545906066895,
      "learning_rate": 9.930155913810369e-05,
      "loss": 0.1359,
      "step": 1412
    },
    {
      "epoch": 0.057133037492292295,
      "grad_norm": 6.948180675506592,
      "learning_rate": 9.930049652636164e-05,
      "loss": 0.1023,
      "step": 1413
    },
    {
      "epoch": 0.05717347134755931,
      "grad_norm": 11.637255668640137,
      "learning_rate": 9.929943311259635e-05,
      "loss": 0.3675,
      "step": 1414
    },
    {
      "epoch": 0.05721390520282633,
      "grad_norm": 4.941865921020508,
      "learning_rate": 9.929836889682514e-05,
      "loss": 0.2023,
      "step": 1415
    },
    {
      "epoch": 0.057254339058093344,
      "grad_norm": 6.169483661651611,
      "learning_rate": 9.929730387906531e-05,
      "loss": 0.2743,
      "step": 1416
    },
    {
      "epoch": 0.05729477291336035,
      "grad_norm": 10.03034496307373,
      "learning_rate": 9.929623805933417e-05,
      "loss": 0.4325,
      "step": 1417
    },
    {
      "epoch": 0.05733520676862737,
      "grad_norm": 3.8539249897003174,
      "learning_rate": 9.929517143764909e-05,
      "loss": 0.1722,
      "step": 1418
    },
    {
      "epoch": 0.057375640623894386,
      "grad_norm": 7.403121471405029,
      "learning_rate": 9.92941040140274e-05,
      "loss": 0.168,
      "step": 1419
    },
    {
      "epoch": 0.0574160744791614,
      "grad_norm": 9.057836532592773,
      "learning_rate": 9.929303578848647e-05,
      "loss": 0.2594,
      "step": 1420
    },
    {
      "epoch": 0.05745650833442842,
      "grad_norm": 5.170660972595215,
      "learning_rate": 9.929196676104368e-05,
      "loss": 0.1862,
      "step": 1421
    },
    {
      "epoch": 0.057496942189695435,
      "grad_norm": 9.588068962097168,
      "learning_rate": 9.92908969317164e-05,
      "loss": 0.2422,
      "step": 1422
    },
    {
      "epoch": 0.057537376044962445,
      "grad_norm": 3.5277092456817627,
      "learning_rate": 9.928982630052208e-05,
      "loss": 0.1458,
      "step": 1423
    },
    {
      "epoch": 0.05757780990022946,
      "grad_norm": 5.695284843444824,
      "learning_rate": 9.92887548674781e-05,
      "loss": 0.1183,
      "step": 1424
    },
    {
      "epoch": 0.05761824375549648,
      "grad_norm": 3.3188178539276123,
      "learning_rate": 9.928768263260188e-05,
      "loss": 0.1681,
      "step": 1425
    },
    {
      "epoch": 0.057658677610763494,
      "grad_norm": 3.9397125244140625,
      "learning_rate": 9.928660959591091e-05,
      "loss": 0.2259,
      "step": 1426
    },
    {
      "epoch": 0.05769911146603051,
      "grad_norm": 2.24654483795166,
      "learning_rate": 9.928553575742261e-05,
      "loss": 0.1414,
      "step": 1427
    },
    {
      "epoch": 0.05773954532129752,
      "grad_norm": 8.203660011291504,
      "learning_rate": 9.928446111715446e-05,
      "loss": 0.4127,
      "step": 1428
    },
    {
      "epoch": 0.057779979176564536,
      "grad_norm": 7.670619487762451,
      "learning_rate": 9.928338567512394e-05,
      "loss": 0.169,
      "step": 1429
    },
    {
      "epoch": 0.05782041303183155,
      "grad_norm": 3.348825216293335,
      "learning_rate": 9.928230943134855e-05,
      "loss": 0.1458,
      "step": 1430
    },
    {
      "epoch": 0.05786084688709857,
      "grad_norm": 1.7959498167037964,
      "learning_rate": 9.92812323858458e-05,
      "loss": 0.1395,
      "step": 1431
    },
    {
      "epoch": 0.057901280742365585,
      "grad_norm": 2.8355765342712402,
      "learning_rate": 9.928015453863318e-05,
      "loss": 0.0981,
      "step": 1432
    },
    {
      "epoch": 0.057941714597632595,
      "grad_norm": 4.64672327041626,
      "learning_rate": 9.927907588972827e-05,
      "loss": 0.0885,
      "step": 1433
    },
    {
      "epoch": 0.05798214845289961,
      "grad_norm": 3.3282997608184814,
      "learning_rate": 9.927799643914859e-05,
      "loss": 0.1714,
      "step": 1434
    },
    {
      "epoch": 0.05802258230816663,
      "grad_norm": 6.719245433807373,
      "learning_rate": 9.927691618691171e-05,
      "loss": 0.1497,
      "step": 1435
    },
    {
      "epoch": 0.058063016163433644,
      "grad_norm": 6.3323445320129395,
      "learning_rate": 9.927583513303521e-05,
      "loss": 0.2015,
      "step": 1436
    },
    {
      "epoch": 0.05810345001870066,
      "grad_norm": 4.704412937164307,
      "learning_rate": 9.927475327753667e-05,
      "loss": 0.1728,
      "step": 1437
    },
    {
      "epoch": 0.058143883873967676,
      "grad_norm": 2.7862377166748047,
      "learning_rate": 9.927367062043368e-05,
      "loss": 0.101,
      "step": 1438
    },
    {
      "epoch": 0.058184317729234686,
      "grad_norm": 1.9709141254425049,
      "learning_rate": 9.927258716174386e-05,
      "loss": 0.1326,
      "step": 1439
    },
    {
      "epoch": 0.0582247515845017,
      "grad_norm": 4.307905197143555,
      "learning_rate": 9.927150290148485e-05,
      "loss": 0.2021,
      "step": 1440
    },
    {
      "epoch": 0.05826518543976872,
      "grad_norm": 2.1209099292755127,
      "learning_rate": 9.927041783967427e-05,
      "loss": 0.1241,
      "step": 1441
    },
    {
      "epoch": 0.058305619295035735,
      "grad_norm": 7.222928524017334,
      "learning_rate": 9.926933197632977e-05,
      "loss": 0.2214,
      "step": 1442
    },
    {
      "epoch": 0.05834605315030275,
      "grad_norm": 4.048924922943115,
      "learning_rate": 9.926824531146903e-05,
      "loss": 0.2578,
      "step": 1443
    },
    {
      "epoch": 0.05838648700556976,
      "grad_norm": 2.198029041290283,
      "learning_rate": 9.926715784510971e-05,
      "loss": 0.1963,
      "step": 1444
    },
    {
      "epoch": 0.05842692086083678,
      "grad_norm": 4.3168721199035645,
      "learning_rate": 9.926606957726953e-05,
      "loss": 0.1238,
      "step": 1445
    },
    {
      "epoch": 0.058467354716103793,
      "grad_norm": 4.095746994018555,
      "learning_rate": 9.926498050796618e-05,
      "loss": 0.0527,
      "step": 1446
    },
    {
      "epoch": 0.05850778857137081,
      "grad_norm": 4.362798690795898,
      "learning_rate": 9.926389063721735e-05,
      "loss": 0.1552,
      "step": 1447
    },
    {
      "epoch": 0.058548222426637826,
      "grad_norm": 3.0533945560455322,
      "learning_rate": 9.92627999650408e-05,
      "loss": 0.1144,
      "step": 1448
    },
    {
      "epoch": 0.058588656281904836,
      "grad_norm": 8.477404594421387,
      "learning_rate": 9.926170849145426e-05,
      "loss": 0.1464,
      "step": 1449
    },
    {
      "epoch": 0.05862909013717185,
      "grad_norm": 11.283618927001953,
      "learning_rate": 9.92606162164755e-05,
      "loss": 0.1912,
      "step": 1450
    },
    {
      "epoch": 0.05866952399243887,
      "grad_norm": 6.521129131317139,
      "learning_rate": 9.925952314012229e-05,
      "loss": 0.2289,
      "step": 1451
    },
    {
      "epoch": 0.058709957847705885,
      "grad_norm": 4.541840076446533,
      "learning_rate": 9.92584292624124e-05,
      "loss": 0.1178,
      "step": 1452
    },
    {
      "epoch": 0.0587503917029729,
      "grad_norm": 3.871896982192993,
      "learning_rate": 9.925733458336362e-05,
      "loss": 0.2178,
      "step": 1453
    },
    {
      "epoch": 0.05879082555823992,
      "grad_norm": 2.072091817855835,
      "learning_rate": 9.925623910299378e-05,
      "loss": 0.0856,
      "step": 1454
    },
    {
      "epoch": 0.05883125941350693,
      "grad_norm": 8.085548400878906,
      "learning_rate": 9.925514282132067e-05,
      "loss": 0.3337,
      "step": 1455
    },
    {
      "epoch": 0.05887169326877394,
      "grad_norm": 6.197680950164795,
      "learning_rate": 9.925404573836215e-05,
      "loss": 0.1803,
      "step": 1456
    },
    {
      "epoch": 0.05891212712404096,
      "grad_norm": 9.838767051696777,
      "learning_rate": 9.925294785413606e-05,
      "loss": 0.2453,
      "step": 1457
    },
    {
      "epoch": 0.058952560979307976,
      "grad_norm": 9.262682914733887,
      "learning_rate": 9.925184916866026e-05,
      "loss": 0.2567,
      "step": 1458
    },
    {
      "epoch": 0.05899299483457499,
      "grad_norm": 14.757526397705078,
      "learning_rate": 9.925074968195262e-05,
      "loss": 0.4045,
      "step": 1459
    },
    {
      "epoch": 0.059033428689842,
      "grad_norm": 2.0353305339813232,
      "learning_rate": 9.924964939403105e-05,
      "loss": 0.0885,
      "step": 1460
    },
    {
      "epoch": 0.05907386254510902,
      "grad_norm": 6.083865642547607,
      "learning_rate": 9.924854830491342e-05,
      "loss": 0.2426,
      "step": 1461
    },
    {
      "epoch": 0.059114296400376035,
      "grad_norm": 4.020568370819092,
      "learning_rate": 9.924744641461764e-05,
      "loss": 0.285,
      "step": 1462
    },
    {
      "epoch": 0.05915473025564305,
      "grad_norm": 11.384716987609863,
      "learning_rate": 9.924634372316166e-05,
      "loss": 0.3668,
      "step": 1463
    },
    {
      "epoch": 0.05919516411091007,
      "grad_norm": 6.951832294464111,
      "learning_rate": 9.92452402305634e-05,
      "loss": 0.1985,
      "step": 1464
    },
    {
      "epoch": 0.05923559796617708,
      "grad_norm": 2.5312283039093018,
      "learning_rate": 9.924413593684082e-05,
      "loss": 0.1462,
      "step": 1465
    },
    {
      "epoch": 0.05927603182144409,
      "grad_norm": 2.6452925205230713,
      "learning_rate": 9.924303084201189e-05,
      "loss": 0.1376,
      "step": 1466
    },
    {
      "epoch": 0.05931646567671111,
      "grad_norm": 4.163340091705322,
      "learning_rate": 9.924192494609458e-05,
      "loss": 0.1567,
      "step": 1467
    },
    {
      "epoch": 0.059356899531978126,
      "grad_norm": 3.9507229328155518,
      "learning_rate": 9.924081824910688e-05,
      "loss": 0.1649,
      "step": 1468
    },
    {
      "epoch": 0.05939733338724514,
      "grad_norm": 5.151828765869141,
      "learning_rate": 9.92397107510668e-05,
      "loss": 0.1646,
      "step": 1469
    },
    {
      "epoch": 0.05943776724251216,
      "grad_norm": 2.422424793243408,
      "learning_rate": 9.923860245199233e-05,
      "loss": 0.0993,
      "step": 1470
    },
    {
      "epoch": 0.05947820109777917,
      "grad_norm": 13.692816734313965,
      "learning_rate": 9.923749335190154e-05,
      "loss": 0.4179,
      "step": 1471
    },
    {
      "epoch": 0.059518634953046184,
      "grad_norm": 5.033790588378906,
      "learning_rate": 9.923638345081245e-05,
      "loss": 0.141,
      "step": 1472
    },
    {
      "epoch": 0.0595590688083132,
      "grad_norm": 6.910784721374512,
      "learning_rate": 9.923527274874314e-05,
      "loss": 0.2975,
      "step": 1473
    },
    {
      "epoch": 0.05959950266358022,
      "grad_norm": 8.2840576171875,
      "learning_rate": 9.923416124571163e-05,
      "loss": 0.1782,
      "step": 1474
    },
    {
      "epoch": 0.059639936518847234,
      "grad_norm": 3.0392425060272217,
      "learning_rate": 9.923304894173606e-05,
      "loss": 0.1464,
      "step": 1475
    },
    {
      "epoch": 0.05968037037411424,
      "grad_norm": 3.891530752182007,
      "learning_rate": 9.923193583683447e-05,
      "loss": 0.1956,
      "step": 1476
    },
    {
      "epoch": 0.05972080422938126,
      "grad_norm": 5.259338855743408,
      "learning_rate": 9.923082193102501e-05,
      "loss": 0.1461,
      "step": 1477
    },
    {
      "epoch": 0.059761238084648276,
      "grad_norm": 11.630255699157715,
      "learning_rate": 9.92297072243258e-05,
      "loss": 0.3171,
      "step": 1478
    },
    {
      "epoch": 0.05980167193991529,
      "grad_norm": 10.16624927520752,
      "learning_rate": 9.922859171675493e-05,
      "loss": 0.2637,
      "step": 1479
    },
    {
      "epoch": 0.05984210579518231,
      "grad_norm": 6.842409133911133,
      "learning_rate": 9.922747540833061e-05,
      "loss": 0.26,
      "step": 1480
    },
    {
      "epoch": 0.05988253965044932,
      "grad_norm": 5.393866539001465,
      "learning_rate": 9.922635829907094e-05,
      "loss": 0.2625,
      "step": 1481
    },
    {
      "epoch": 0.059922973505716334,
      "grad_norm": 7.717151641845703,
      "learning_rate": 9.922524038899415e-05,
      "loss": 0.2138,
      "step": 1482
    },
    {
      "epoch": 0.05996340736098335,
      "grad_norm": 5.877342700958252,
      "learning_rate": 9.922412167811837e-05,
      "loss": 0.1896,
      "step": 1483
    },
    {
      "epoch": 0.06000384121625037,
      "grad_norm": 7.007465839385986,
      "learning_rate": 9.922300216646184e-05,
      "loss": 0.1285,
      "step": 1484
    },
    {
      "epoch": 0.06004427507151738,
      "grad_norm": 3.2064719200134277,
      "learning_rate": 9.922188185404276e-05,
      "loss": 0.2048,
      "step": 1485
    },
    {
      "epoch": 0.0600847089267844,
      "grad_norm": 8.557111740112305,
      "learning_rate": 9.922076074087934e-05,
      "loss": 0.2923,
      "step": 1486
    },
    {
      "epoch": 0.06012514278205141,
      "grad_norm": 3.4697537422180176,
      "learning_rate": 9.921963882698984e-05,
      "loss": 0.1082,
      "step": 1487
    },
    {
      "epoch": 0.060165576637318426,
      "grad_norm": 2.8538320064544678,
      "learning_rate": 9.92185161123925e-05,
      "loss": 0.0936,
      "step": 1488
    },
    {
      "epoch": 0.06020601049258544,
      "grad_norm": 11.081395149230957,
      "learning_rate": 9.92173925971056e-05,
      "loss": 0.2956,
      "step": 1489
    },
    {
      "epoch": 0.06024644434785246,
      "grad_norm": 4.404125690460205,
      "learning_rate": 9.921626828114741e-05,
      "loss": 0.2355,
      "step": 1490
    },
    {
      "epoch": 0.060286878203119475,
      "grad_norm": 2.5893630981445312,
      "learning_rate": 9.921514316453619e-05,
      "loss": 0.1523,
      "step": 1491
    },
    {
      "epoch": 0.060327312058386484,
      "grad_norm": 7.941740989685059,
      "learning_rate": 9.92140172472903e-05,
      "loss": 0.2048,
      "step": 1492
    },
    {
      "epoch": 0.0603677459136535,
      "grad_norm": 6.021244525909424,
      "learning_rate": 9.921289052942799e-05,
      "loss": 0.3113,
      "step": 1493
    },
    {
      "epoch": 0.06040817976892052,
      "grad_norm": 3.044511318206787,
      "learning_rate": 9.921176301096765e-05,
      "loss": 0.1092,
      "step": 1494
    },
    {
      "epoch": 0.06044861362418753,
      "grad_norm": 3.796078681945801,
      "learning_rate": 9.921063469192759e-05,
      "loss": 0.1036,
      "step": 1495
    },
    {
      "epoch": 0.06048904747945455,
      "grad_norm": 5.367008686065674,
      "learning_rate": 9.920950557232618e-05,
      "loss": 0.1653,
      "step": 1496
    },
    {
      "epoch": 0.06052948133472156,
      "grad_norm": 6.031381130218506,
      "learning_rate": 9.920837565218177e-05,
      "loss": 0.2653,
      "step": 1497
    },
    {
      "epoch": 0.060569915189988575,
      "grad_norm": 2.304837465286255,
      "learning_rate": 9.920724493151276e-05,
      "loss": 0.0897,
      "step": 1498
    },
    {
      "epoch": 0.06061034904525559,
      "grad_norm": 6.305426120758057,
      "learning_rate": 9.920611341033754e-05,
      "loss": 0.1886,
      "step": 1499
    },
    {
      "epoch": 0.06065078290052261,
      "grad_norm": 10.211784362792969,
      "learning_rate": 9.920498108867453e-05,
      "loss": 0.2481,
      "step": 1500
    },
    {
      "epoch": 0.060691216755789625,
      "grad_norm": 9.631447792053223,
      "learning_rate": 9.920384796654212e-05,
      "loss": 0.3012,
      "step": 1501
    },
    {
      "epoch": 0.06073165061105664,
      "grad_norm": 7.636146068572998,
      "learning_rate": 9.920271404395876e-05,
      "loss": 0.2313,
      "step": 1502
    },
    {
      "epoch": 0.06077208446632365,
      "grad_norm": 9.774024963378906,
      "learning_rate": 9.920157932094289e-05,
      "loss": 0.2497,
      "step": 1503
    },
    {
      "epoch": 0.06081251832159067,
      "grad_norm": 6.658890724182129,
      "learning_rate": 9.920044379751299e-05,
      "loss": 0.1993,
      "step": 1504
    },
    {
      "epoch": 0.06085295217685768,
      "grad_norm": 4.256870269775391,
      "learning_rate": 9.919930747368753e-05,
      "loss": 0.1279,
      "step": 1505
    },
    {
      "epoch": 0.0608933860321247,
      "grad_norm": 7.223304748535156,
      "learning_rate": 9.919817034948496e-05,
      "loss": 0.1967,
      "step": 1506
    },
    {
      "epoch": 0.060933819887391716,
      "grad_norm": 1.5365058183670044,
      "learning_rate": 9.919703242492382e-05,
      "loss": 0.1257,
      "step": 1507
    },
    {
      "epoch": 0.060974253742658725,
      "grad_norm": 3.982435464859009,
      "learning_rate": 9.919589370002258e-05,
      "loss": 0.1583,
      "step": 1508
    },
    {
      "epoch": 0.06101468759792574,
      "grad_norm": 11.779656410217285,
      "learning_rate": 9.919475417479983e-05,
      "loss": 0.3247,
      "step": 1509
    },
    {
      "epoch": 0.06105512145319276,
      "grad_norm": 4.34445858001709,
      "learning_rate": 9.919361384927403e-05,
      "loss": 0.0989,
      "step": 1510
    },
    {
      "epoch": 0.061095555308459774,
      "grad_norm": 3.006352663040161,
      "learning_rate": 9.91924727234638e-05,
      "loss": 0.1706,
      "step": 1511
    },
    {
      "epoch": 0.06113598916372679,
      "grad_norm": 4.386690616607666,
      "learning_rate": 9.919133079738767e-05,
      "loss": 0.3358,
      "step": 1512
    },
    {
      "epoch": 0.0611764230189938,
      "grad_norm": 5.080314636230469,
      "learning_rate": 9.91901880710642e-05,
      "loss": 0.2074,
      "step": 1513
    },
    {
      "epoch": 0.06121685687426082,
      "grad_norm": 1.7615193128585815,
      "learning_rate": 9.918904454451202e-05,
      "loss": 0.0594,
      "step": 1514
    },
    {
      "epoch": 0.06125729072952783,
      "grad_norm": 7.6022748947143555,
      "learning_rate": 9.918790021774971e-05,
      "loss": 0.1702,
      "step": 1515
    },
    {
      "epoch": 0.06129772458479485,
      "grad_norm": 9.466252326965332,
      "learning_rate": 9.918675509079587e-05,
      "loss": 0.229,
      "step": 1516
    },
    {
      "epoch": 0.061338158440061866,
      "grad_norm": 5.846565246582031,
      "learning_rate": 9.918560916366917e-05,
      "loss": 0.1991,
      "step": 1517
    },
    {
      "epoch": 0.06137859229532888,
      "grad_norm": 6.043490886688232,
      "learning_rate": 9.918446243638823e-05,
      "loss": 0.1338,
      "step": 1518
    },
    {
      "epoch": 0.06141902615059589,
      "grad_norm": 5.366283416748047,
      "learning_rate": 9.91833149089717e-05,
      "loss": 0.1546,
      "step": 1519
    },
    {
      "epoch": 0.06145946000586291,
      "grad_norm": 3.590197801589966,
      "learning_rate": 9.918216658143825e-05,
      "loss": 0.1606,
      "step": 1520
    },
    {
      "epoch": 0.061499893861129924,
      "grad_norm": 5.756260395050049,
      "learning_rate": 9.918101745380657e-05,
      "loss": 0.1722,
      "step": 1521
    },
    {
      "epoch": 0.06154032771639694,
      "grad_norm": 4.020991802215576,
      "learning_rate": 9.917986752609533e-05,
      "loss": 0.1861,
      "step": 1522
    },
    {
      "epoch": 0.06158076157166396,
      "grad_norm": 3.200106143951416,
      "learning_rate": 9.917871679832327e-05,
      "loss": 0.3272,
      "step": 1523
    },
    {
      "epoch": 0.061621195426930966,
      "grad_norm": 4.977615833282471,
      "learning_rate": 9.91775652705091e-05,
      "loss": 0.1475,
      "step": 1524
    },
    {
      "epoch": 0.06166162928219798,
      "grad_norm": 5.720094203948975,
      "learning_rate": 9.917641294267155e-05,
      "loss": 0.1817,
      "step": 1525
    },
    {
      "epoch": 0.061702063137465,
      "grad_norm": 2.6053671836853027,
      "learning_rate": 9.917525981482936e-05,
      "loss": 0.1405,
      "step": 1526
    },
    {
      "epoch": 0.061742496992732016,
      "grad_norm": 13.017812728881836,
      "learning_rate": 9.917410588700128e-05,
      "loss": 0.3419,
      "step": 1527
    },
    {
      "epoch": 0.06178293084799903,
      "grad_norm": 8.22148323059082,
      "learning_rate": 9.91729511592061e-05,
      "loss": 0.3517,
      "step": 1528
    },
    {
      "epoch": 0.06182336470326604,
      "grad_norm": 8.00568675994873,
      "learning_rate": 9.917179563146261e-05,
      "loss": 0.2648,
      "step": 1529
    },
    {
      "epoch": 0.06186379855853306,
      "grad_norm": 7.261849880218506,
      "learning_rate": 9.917063930378961e-05,
      "loss": 0.2412,
      "step": 1530
    },
    {
      "epoch": 0.061904232413800074,
      "grad_norm": 11.258759498596191,
      "learning_rate": 9.91694821762059e-05,
      "loss": 0.2501,
      "step": 1531
    },
    {
      "epoch": 0.06194466626906709,
      "grad_norm": 1.9563425779342651,
      "learning_rate": 9.916832424873028e-05,
      "loss": 0.1194,
      "step": 1532
    },
    {
      "epoch": 0.06198510012433411,
      "grad_norm": 6.295723915100098,
      "learning_rate": 9.916716552138163e-05,
      "loss": 0.229,
      "step": 1533
    },
    {
      "epoch": 0.06202553397960112,
      "grad_norm": 10.11221981048584,
      "learning_rate": 9.916600599417878e-05,
      "loss": 0.2334,
      "step": 1534
    },
    {
      "epoch": 0.06206596783486813,
      "grad_norm": 8.048049926757812,
      "learning_rate": 9.916484566714059e-05,
      "loss": 0.1724,
      "step": 1535
    },
    {
      "epoch": 0.06210640169013515,
      "grad_norm": 5.623824596405029,
      "learning_rate": 9.916368454028595e-05,
      "loss": 0.2264,
      "step": 1536
    },
    {
      "epoch": 0.062146835545402165,
      "grad_norm": 2.9101154804229736,
      "learning_rate": 9.916252261363374e-05,
      "loss": 0.1053,
      "step": 1537
    },
    {
      "epoch": 0.06218726940066918,
      "grad_norm": 4.124871730804443,
      "learning_rate": 9.916135988720285e-05,
      "loss": 0.1669,
      "step": 1538
    },
    {
      "epoch": 0.0622277032559362,
      "grad_norm": 6.295670032501221,
      "learning_rate": 9.916019636101223e-05,
      "loss": 0.2339,
      "step": 1539
    },
    {
      "epoch": 0.06226813711120321,
      "grad_norm": 3.094670057296753,
      "learning_rate": 9.915903203508077e-05,
      "loss": 0.1192,
      "step": 1540
    },
    {
      "epoch": 0.062308570966470224,
      "grad_norm": 3.416536569595337,
      "learning_rate": 9.915786690942744e-05,
      "loss": 0.1507,
      "step": 1541
    },
    {
      "epoch": 0.06234900482173724,
      "grad_norm": 7.417731761932373,
      "learning_rate": 9.915670098407117e-05,
      "loss": 0.1872,
      "step": 1542
    },
    {
      "epoch": 0.06238943867700426,
      "grad_norm": 3.502101421356201,
      "learning_rate": 9.915553425903094e-05,
      "loss": 0.2272,
      "step": 1543
    },
    {
      "epoch": 0.06242987253227127,
      "grad_norm": 3.8697011470794678,
      "learning_rate": 9.915436673432575e-05,
      "loss": 0.1,
      "step": 1544
    },
    {
      "epoch": 0.06247030638753828,
      "grad_norm": 8.228521347045898,
      "learning_rate": 9.915319840997456e-05,
      "loss": 0.3646,
      "step": 1545
    },
    {
      "epoch": 0.0625107402428053,
      "grad_norm": 3.2267842292785645,
      "learning_rate": 9.915202928599639e-05,
      "loss": 0.077,
      "step": 1546
    },
    {
      "epoch": 0.06255117409807232,
      "grad_norm": 5.761537551879883,
      "learning_rate": 9.915085936241026e-05,
      "loss": 0.1401,
      "step": 1547
    },
    {
      "epoch": 0.06259160795333933,
      "grad_norm": 11.487232208251953,
      "learning_rate": 9.914968863923519e-05,
      "loss": 0.2338,
      "step": 1548
    },
    {
      "epoch": 0.06263204180860635,
      "grad_norm": 5.205531597137451,
      "learning_rate": 9.914851711649026e-05,
      "loss": 0.074,
      "step": 1549
    },
    {
      "epoch": 0.06267247566387336,
      "grad_norm": 35.287845611572266,
      "learning_rate": 9.91473447941945e-05,
      "loss": 0.4223,
      "step": 1550
    },
    {
      "epoch": 0.06271290951914038,
      "grad_norm": 2.6794731616973877,
      "learning_rate": 9.914617167236697e-05,
      "loss": 0.1818,
      "step": 1551
    },
    {
      "epoch": 0.0627533433744074,
      "grad_norm": 10.774479866027832,
      "learning_rate": 9.914499775102679e-05,
      "loss": 0.2969,
      "step": 1552
    },
    {
      "epoch": 0.0627937772296744,
      "grad_norm": 6.784860610961914,
      "learning_rate": 9.914382303019303e-05,
      "loss": 0.2064,
      "step": 1553
    },
    {
      "epoch": 0.06283421108494142,
      "grad_norm": 5.522778034210205,
      "learning_rate": 9.914264750988483e-05,
      "loss": 0.3077,
      "step": 1554
    },
    {
      "epoch": 0.06287464494020843,
      "grad_norm": 7.515862941741943,
      "learning_rate": 9.914147119012125e-05,
      "loss": 0.1589,
      "step": 1555
    },
    {
      "epoch": 0.06291507879547545,
      "grad_norm": 3.4086835384368896,
      "learning_rate": 9.91402940709215e-05,
      "loss": 0.1107,
      "step": 1556
    },
    {
      "epoch": 0.06295551265074247,
      "grad_norm": 10.977673530578613,
      "learning_rate": 9.91391161523047e-05,
      "loss": 0.2069,
      "step": 1557
    },
    {
      "epoch": 0.06299594650600948,
      "grad_norm": 3.3024075031280518,
      "learning_rate": 9.913793743429e-05,
      "loss": 0.1102,
      "step": 1558
    },
    {
      "epoch": 0.0630363803612765,
      "grad_norm": 7.54313325881958,
      "learning_rate": 9.913675791689658e-05,
      "loss": 0.1487,
      "step": 1559
    },
    {
      "epoch": 0.06307681421654351,
      "grad_norm": 5.960170269012451,
      "learning_rate": 9.913557760014365e-05,
      "loss": 0.2668,
      "step": 1560
    },
    {
      "epoch": 0.06311724807181053,
      "grad_norm": 4.662247180938721,
      "learning_rate": 9.913439648405039e-05,
      "loss": 0.2511,
      "step": 1561
    },
    {
      "epoch": 0.06315768192707755,
      "grad_norm": 7.298117637634277,
      "learning_rate": 9.913321456863603e-05,
      "loss": 0.1799,
      "step": 1562
    },
    {
      "epoch": 0.06319811578234456,
      "grad_norm": 4.205918788909912,
      "learning_rate": 9.913203185391977e-05,
      "loss": 0.268,
      "step": 1563
    },
    {
      "epoch": 0.06323854963761157,
      "grad_norm": 4.2428717613220215,
      "learning_rate": 9.913084833992088e-05,
      "loss": 0.1913,
      "step": 1564
    },
    {
      "epoch": 0.06327898349287858,
      "grad_norm": 6.33016300201416,
      "learning_rate": 9.91296640266586e-05,
      "loss": 0.2132,
      "step": 1565
    },
    {
      "epoch": 0.0633194173481456,
      "grad_norm": 8.462950706481934,
      "learning_rate": 9.91284789141522e-05,
      "loss": 0.2202,
      "step": 1566
    },
    {
      "epoch": 0.06335985120341261,
      "grad_norm": 11.97440242767334,
      "learning_rate": 9.912729300242097e-05,
      "loss": 0.2719,
      "step": 1567
    },
    {
      "epoch": 0.06340028505867963,
      "grad_norm": 9.846284866333008,
      "learning_rate": 9.912610629148417e-05,
      "loss": 0.2954,
      "step": 1568
    },
    {
      "epoch": 0.06344071891394665,
      "grad_norm": 5.3169169425964355,
      "learning_rate": 9.912491878136113e-05,
      "loss": 0.1866,
      "step": 1569
    },
    {
      "epoch": 0.06348115276921366,
      "grad_norm": 5.82018518447876,
      "learning_rate": 9.912373047207117e-05,
      "loss": 0.2513,
      "step": 1570
    },
    {
      "epoch": 0.06352158662448068,
      "grad_norm": 2.8800771236419678,
      "learning_rate": 9.912254136363361e-05,
      "loss": 0.1459,
      "step": 1571
    },
    {
      "epoch": 0.0635620204797477,
      "grad_norm": 2.6133639812469482,
      "learning_rate": 9.912135145606781e-05,
      "loss": 0.2235,
      "step": 1572
    },
    {
      "epoch": 0.06360245433501471,
      "grad_norm": 6.5553693771362305,
      "learning_rate": 9.912016074939313e-05,
      "loss": 0.2042,
      "step": 1573
    },
    {
      "epoch": 0.06364288819028173,
      "grad_norm": 2.9834353923797607,
      "learning_rate": 9.91189692436289e-05,
      "loss": 0.2238,
      "step": 1574
    },
    {
      "epoch": 0.06368332204554873,
      "grad_norm": 7.165900230407715,
      "learning_rate": 9.911777693879454e-05,
      "loss": 0.1955,
      "step": 1575
    },
    {
      "epoch": 0.06372375590081575,
      "grad_norm": 2.384648084640503,
      "learning_rate": 9.911658383490945e-05,
      "loss": 0.0874,
      "step": 1576
    },
    {
      "epoch": 0.06376418975608276,
      "grad_norm": 3.6222381591796875,
      "learning_rate": 9.911538993199301e-05,
      "loss": 0.1327,
      "step": 1577
    },
    {
      "epoch": 0.06380462361134978,
      "grad_norm": 11.3930025100708,
      "learning_rate": 9.911419523006469e-05,
      "loss": 0.2581,
      "step": 1578
    },
    {
      "epoch": 0.0638450574666168,
      "grad_norm": 10.176639556884766,
      "learning_rate": 9.911299972914388e-05,
      "loss": 0.2224,
      "step": 1579
    },
    {
      "epoch": 0.06388549132188381,
      "grad_norm": 7.479902267456055,
      "learning_rate": 9.911180342925003e-05,
      "loss": 0.1739,
      "step": 1580
    },
    {
      "epoch": 0.06392592517715083,
      "grad_norm": 3.561979055404663,
      "learning_rate": 9.911060633040264e-05,
      "loss": 0.274,
      "step": 1581
    },
    {
      "epoch": 0.06396635903241785,
      "grad_norm": 2.2039783000946045,
      "learning_rate": 9.910940843262113e-05,
      "loss": 0.0829,
      "step": 1582
    },
    {
      "epoch": 0.06400679288768486,
      "grad_norm": 7.87419319152832,
      "learning_rate": 9.910820973592504e-05,
      "loss": 0.2534,
      "step": 1583
    },
    {
      "epoch": 0.06404722674295188,
      "grad_norm": 4.467629432678223,
      "learning_rate": 9.910701024033385e-05,
      "loss": 0.1102,
      "step": 1584
    },
    {
      "epoch": 0.06408766059821888,
      "grad_norm": 7.177118301391602,
      "learning_rate": 9.910580994586707e-05,
      "loss": 0.1431,
      "step": 1585
    },
    {
      "epoch": 0.0641280944534859,
      "grad_norm": 9.04709529876709,
      "learning_rate": 9.910460885254422e-05,
      "loss": 0.192,
      "step": 1586
    },
    {
      "epoch": 0.06416852830875291,
      "grad_norm": 8.07638931274414,
      "learning_rate": 9.910340696038484e-05,
      "loss": 0.2409,
      "step": 1587
    },
    {
      "epoch": 0.06420896216401993,
      "grad_norm": 9.383169174194336,
      "learning_rate": 9.91022042694085e-05,
      "loss": 0.2978,
      "step": 1588
    },
    {
      "epoch": 0.06424939601928695,
      "grad_norm": 7.415613651275635,
      "learning_rate": 9.910100077963476e-05,
      "loss": 0.1524,
      "step": 1589
    },
    {
      "epoch": 0.06428982987455396,
      "grad_norm": 5.076672554016113,
      "learning_rate": 9.90997964910832e-05,
      "loss": 0.1673,
      "step": 1590
    },
    {
      "epoch": 0.06433026372982098,
      "grad_norm": 3.1432385444641113,
      "learning_rate": 9.90985914037734e-05,
      "loss": 0.1279,
      "step": 1591
    },
    {
      "epoch": 0.064370697585088,
      "grad_norm": 3.2620697021484375,
      "learning_rate": 9.909738551772497e-05,
      "loss": 0.1129,
      "step": 1592
    },
    {
      "epoch": 0.06441113144035501,
      "grad_norm": 6.301144599914551,
      "learning_rate": 9.909617883295752e-05,
      "loss": 0.2329,
      "step": 1593
    },
    {
      "epoch": 0.06445156529562203,
      "grad_norm": 3.075220823287964,
      "learning_rate": 9.909497134949068e-05,
      "loss": 0.0656,
      "step": 1594
    },
    {
      "epoch": 0.06449199915088905,
      "grad_norm": 8.724457740783691,
      "learning_rate": 9.909376306734413e-05,
      "loss": 0.1451,
      "step": 1595
    },
    {
      "epoch": 0.06453243300615605,
      "grad_norm": 9.219529151916504,
      "learning_rate": 9.909255398653748e-05,
      "loss": 0.2838,
      "step": 1596
    },
    {
      "epoch": 0.06457286686142306,
      "grad_norm": 5.19163703918457,
      "learning_rate": 9.909134410709043e-05,
      "loss": 0.1349,
      "step": 1597
    },
    {
      "epoch": 0.06461330071669008,
      "grad_norm": 3.038506269454956,
      "learning_rate": 9.909013342902262e-05,
      "loss": 0.1632,
      "step": 1598
    },
    {
      "epoch": 0.0646537345719571,
      "grad_norm": 5.4024176597595215,
      "learning_rate": 9.90889219523538e-05,
      "loss": 0.1299,
      "step": 1599
    },
    {
      "epoch": 0.06469416842722411,
      "grad_norm": 5.959799766540527,
      "learning_rate": 9.908770967710363e-05,
      "loss": 0.1883,
      "step": 1600
    },
    {
      "epoch": 0.06473460228249113,
      "grad_norm": 16.51708984375,
      "learning_rate": 9.908649660329187e-05,
      "loss": 0.3803,
      "step": 1601
    },
    {
      "epoch": 0.06477503613775815,
      "grad_norm": 9.996596336364746,
      "learning_rate": 9.908528273093825e-05,
      "loss": 0.2642,
      "step": 1602
    },
    {
      "epoch": 0.06481546999302516,
      "grad_norm": 8.556817054748535,
      "learning_rate": 9.90840680600625e-05,
      "loss": 0.3024,
      "step": 1603
    },
    {
      "epoch": 0.06485590384829218,
      "grad_norm": 7.400540828704834,
      "learning_rate": 9.908285259068437e-05,
      "loss": 0.2172,
      "step": 1604
    },
    {
      "epoch": 0.0648963377035592,
      "grad_norm": 2.0541484355926514,
      "learning_rate": 9.908163632282365e-05,
      "loss": 0.0993,
      "step": 1605
    },
    {
      "epoch": 0.06493677155882621,
      "grad_norm": 5.620724678039551,
      "learning_rate": 9.908041925650014e-05,
      "loss": 0.2555,
      "step": 1606
    },
    {
      "epoch": 0.06497720541409321,
      "grad_norm": 3.188631534576416,
      "learning_rate": 9.907920139173363e-05,
      "loss": 0.1665,
      "step": 1607
    },
    {
      "epoch": 0.06501763926936023,
      "grad_norm": 3.680882453918457,
      "learning_rate": 9.907798272854391e-05,
      "loss": 0.1436,
      "step": 1608
    },
    {
      "epoch": 0.06505807312462725,
      "grad_norm": 4.789421558380127,
      "learning_rate": 9.907676326695083e-05,
      "loss": 0.1378,
      "step": 1609
    },
    {
      "epoch": 0.06509850697989426,
      "grad_norm": 8.919291496276855,
      "learning_rate": 9.907554300697423e-05,
      "loss": 0.2964,
      "step": 1610
    },
    {
      "epoch": 0.06513894083516128,
      "grad_norm": 2.2851107120513916,
      "learning_rate": 9.907432194863395e-05,
      "loss": 0.1065,
      "step": 1611
    },
    {
      "epoch": 0.0651793746904283,
      "grad_norm": 3.1836793422698975,
      "learning_rate": 9.907310009194986e-05,
      "loss": 0.1897,
      "step": 1612
    },
    {
      "epoch": 0.06521980854569531,
      "grad_norm": 3.3415918350219727,
      "learning_rate": 9.907187743694182e-05,
      "loss": 0.1206,
      "step": 1613
    },
    {
      "epoch": 0.06526024240096233,
      "grad_norm": 7.148499011993408,
      "learning_rate": 9.907065398362974e-05,
      "loss": 0.1352,
      "step": 1614
    },
    {
      "epoch": 0.06530067625622935,
      "grad_norm": 7.901660442352295,
      "learning_rate": 9.906942973203352e-05,
      "loss": 0.312,
      "step": 1615
    },
    {
      "epoch": 0.06534111011149636,
      "grad_norm": 4.787680149078369,
      "learning_rate": 9.906820468217307e-05,
      "loss": 0.2079,
      "step": 1616
    },
    {
      "epoch": 0.06538154396676336,
      "grad_norm": 5.763666152954102,
      "learning_rate": 9.906697883406833e-05,
      "loss": 0.2347,
      "step": 1617
    },
    {
      "epoch": 0.06542197782203038,
      "grad_norm": 5.864205360412598,
      "learning_rate": 9.906575218773923e-05,
      "loss": 0.2155,
      "step": 1618
    },
    {
      "epoch": 0.0654624116772974,
      "grad_norm": 10.06261157989502,
      "learning_rate": 9.906452474320573e-05,
      "loss": 0.278,
      "step": 1619
    },
    {
      "epoch": 0.06550284553256441,
      "grad_norm": 7.752874374389648,
      "learning_rate": 9.90632965004878e-05,
      "loss": 0.1589,
      "step": 1620
    },
    {
      "epoch": 0.06554327938783143,
      "grad_norm": 4.4440813064575195,
      "learning_rate": 9.906206745960541e-05,
      "loss": 0.1731,
      "step": 1621
    },
    {
      "epoch": 0.06558371324309845,
      "grad_norm": 9.636917114257812,
      "learning_rate": 9.906083762057857e-05,
      "loss": 0.2725,
      "step": 1622
    },
    {
      "epoch": 0.06562414709836546,
      "grad_norm": 5.5359063148498535,
      "learning_rate": 9.90596069834273e-05,
      "loss": 0.1477,
      "step": 1623
    },
    {
      "epoch": 0.06566458095363248,
      "grad_norm": 7.7632246017456055,
      "learning_rate": 9.905837554817157e-05,
      "loss": 0.2011,
      "step": 1624
    },
    {
      "epoch": 0.0657050148088995,
      "grad_norm": 2.873377561569214,
      "learning_rate": 9.905714331483145e-05,
      "loss": 0.0594,
      "step": 1625
    },
    {
      "epoch": 0.06574544866416651,
      "grad_norm": 3.547520160675049,
      "learning_rate": 9.905591028342698e-05,
      "loss": 0.1559,
      "step": 1626
    },
    {
      "epoch": 0.06578588251943353,
      "grad_norm": 5.6297502517700195,
      "learning_rate": 9.905467645397823e-05,
      "loss": 0.1471,
      "step": 1627
    },
    {
      "epoch": 0.06582631637470053,
      "grad_norm": 11.934308052062988,
      "learning_rate": 9.905344182650526e-05,
      "loss": 0.3012,
      "step": 1628
    },
    {
      "epoch": 0.06586675022996755,
      "grad_norm": 6.839045524597168,
      "learning_rate": 9.905220640102815e-05,
      "loss": 0.1455,
      "step": 1629
    },
    {
      "epoch": 0.06590718408523456,
      "grad_norm": 10.07431697845459,
      "learning_rate": 9.905097017756701e-05,
      "loss": 0.2819,
      "step": 1630
    },
    {
      "epoch": 0.06594761794050158,
      "grad_norm": 7.67207670211792,
      "learning_rate": 9.904973315614194e-05,
      "loss": 0.2458,
      "step": 1631
    },
    {
      "epoch": 0.0659880517957686,
      "grad_norm": 8.905278205871582,
      "learning_rate": 9.904849533677307e-05,
      "loss": 0.1691,
      "step": 1632
    },
    {
      "epoch": 0.06602848565103561,
      "grad_norm": 2.4351353645324707,
      "learning_rate": 9.904725671948051e-05,
      "loss": 0.092,
      "step": 1633
    },
    {
      "epoch": 0.06606891950630263,
      "grad_norm": 15.901311874389648,
      "learning_rate": 9.904601730428447e-05,
      "loss": 0.1594,
      "step": 1634
    },
    {
      "epoch": 0.06610935336156964,
      "grad_norm": 3.999016523361206,
      "learning_rate": 9.904477709120506e-05,
      "loss": 0.1686,
      "step": 1635
    },
    {
      "epoch": 0.06614978721683666,
      "grad_norm": 8.251566886901855,
      "learning_rate": 9.904353608026249e-05,
      "loss": 0.2432,
      "step": 1636
    },
    {
      "epoch": 0.06619022107210368,
      "grad_norm": 2.7484471797943115,
      "learning_rate": 9.904229427147691e-05,
      "loss": 0.1064,
      "step": 1637
    },
    {
      "epoch": 0.0662306549273707,
      "grad_norm": 6.759425163269043,
      "learning_rate": 9.904105166486857e-05,
      "loss": 0.263,
      "step": 1638
    },
    {
      "epoch": 0.0662710887826377,
      "grad_norm": 4.501479625701904,
      "learning_rate": 9.903980826045764e-05,
      "loss": 0.2494,
      "step": 1639
    },
    {
      "epoch": 0.06631152263790471,
      "grad_norm": 8.279570579528809,
      "learning_rate": 9.903856405826438e-05,
      "loss": 0.1235,
      "step": 1640
    },
    {
      "epoch": 0.06635195649317173,
      "grad_norm": 6.758943557739258,
      "learning_rate": 9.903731905830902e-05,
      "loss": 0.1352,
      "step": 1641
    },
    {
      "epoch": 0.06639239034843875,
      "grad_norm": 1.671072244644165,
      "learning_rate": 9.90360732606118e-05,
      "loss": 0.0635,
      "step": 1642
    },
    {
      "epoch": 0.06643282420370576,
      "grad_norm": 4.482504367828369,
      "learning_rate": 9.903482666519302e-05,
      "loss": 0.1386,
      "step": 1643
    },
    {
      "epoch": 0.06647325805897278,
      "grad_norm": 5.925358295440674,
      "learning_rate": 9.903357927207291e-05,
      "loss": 0.2365,
      "step": 1644
    },
    {
      "epoch": 0.0665136919142398,
      "grad_norm": 9.290237426757812,
      "learning_rate": 9.90323310812718e-05,
      "loss": 0.1911,
      "step": 1645
    },
    {
      "epoch": 0.06655412576950681,
      "grad_norm": 3.521745443344116,
      "learning_rate": 9.903108209280999e-05,
      "loss": 0.1607,
      "step": 1646
    },
    {
      "epoch": 0.06659455962477383,
      "grad_norm": 11.321200370788574,
      "learning_rate": 9.90298323067078e-05,
      "loss": 0.4138,
      "step": 1647
    },
    {
      "epoch": 0.06663499348004084,
      "grad_norm": 2.4181947708129883,
      "learning_rate": 9.902858172298555e-05,
      "loss": 0.1004,
      "step": 1648
    },
    {
      "epoch": 0.06667542733530785,
      "grad_norm": 7.962637424468994,
      "learning_rate": 9.90273303416636e-05,
      "loss": 0.2716,
      "step": 1649
    },
    {
      "epoch": 0.06671586119057486,
      "grad_norm": 4.06016731262207,
      "learning_rate": 9.902607816276228e-05,
      "loss": 0.1432,
      "step": 1650
    },
    {
      "epoch": 0.06675629504584188,
      "grad_norm": 3.3262860774993896,
      "learning_rate": 9.9024825186302e-05,
      "loss": 0.2074,
      "step": 1651
    },
    {
      "epoch": 0.0667967289011089,
      "grad_norm": 8.232795715332031,
      "learning_rate": 9.902357141230311e-05,
      "loss": 0.2653,
      "step": 1652
    },
    {
      "epoch": 0.06683716275637591,
      "grad_norm": 4.0189409255981445,
      "learning_rate": 9.902231684078602e-05,
      "loss": 0.1527,
      "step": 1653
    },
    {
      "epoch": 0.06687759661164293,
      "grad_norm": 4.051592826843262,
      "learning_rate": 9.902106147177115e-05,
      "loss": 0.1293,
      "step": 1654
    },
    {
      "epoch": 0.06691803046690994,
      "grad_norm": 7.4501166343688965,
      "learning_rate": 9.90198053052789e-05,
      "loss": 0.2079,
      "step": 1655
    },
    {
      "epoch": 0.06695846432217696,
      "grad_norm": 4.766932010650635,
      "learning_rate": 9.90185483413297e-05,
      "loss": 0.3418,
      "step": 1656
    },
    {
      "epoch": 0.06699889817744398,
      "grad_norm": 9.435149192810059,
      "learning_rate": 9.901729057994405e-05,
      "loss": 0.1878,
      "step": 1657
    },
    {
      "epoch": 0.067039332032711,
      "grad_norm": 4.314088821411133,
      "learning_rate": 9.901603202114236e-05,
      "loss": 0.1466,
      "step": 1658
    },
    {
      "epoch": 0.06707976588797801,
      "grad_norm": 4.428193092346191,
      "learning_rate": 9.901477266494513e-05,
      "loss": 0.1777,
      "step": 1659
    },
    {
      "epoch": 0.06712019974324501,
      "grad_norm": 4.708783149719238,
      "learning_rate": 9.901351251137281e-05,
      "loss": 0.2496,
      "step": 1660
    },
    {
      "epoch": 0.06716063359851203,
      "grad_norm": 5.278815746307373,
      "learning_rate": 9.901225156044596e-05,
      "loss": 0.145,
      "step": 1661
    },
    {
      "epoch": 0.06720106745377905,
      "grad_norm": 5.300556182861328,
      "learning_rate": 9.901098981218504e-05,
      "loss": 0.1607,
      "step": 1662
    },
    {
      "epoch": 0.06724150130904606,
      "grad_norm": 5.328469753265381,
      "learning_rate": 9.900972726661061e-05,
      "loss": 0.2154,
      "step": 1663
    },
    {
      "epoch": 0.06728193516431308,
      "grad_norm": 4.050581932067871,
      "learning_rate": 9.900846392374318e-05,
      "loss": 0.1327,
      "step": 1664
    },
    {
      "epoch": 0.0673223690195801,
      "grad_norm": 5.562441349029541,
      "learning_rate": 9.900719978360334e-05,
      "loss": 0.2336,
      "step": 1665
    },
    {
      "epoch": 0.06736280287484711,
      "grad_norm": 5.909516334533691,
      "learning_rate": 9.900593484621164e-05,
      "loss": 0.228,
      "step": 1666
    },
    {
      "epoch": 0.06740323673011413,
      "grad_norm": 6.110733985900879,
      "learning_rate": 9.900466911158863e-05,
      "loss": 0.1245,
      "step": 1667
    },
    {
      "epoch": 0.06744367058538114,
      "grad_norm": 4.894416332244873,
      "learning_rate": 9.900340257975493e-05,
      "loss": 0.2508,
      "step": 1668
    },
    {
      "epoch": 0.06748410444064816,
      "grad_norm": 4.260928630828857,
      "learning_rate": 9.900213525073113e-05,
      "loss": 0.1818,
      "step": 1669
    },
    {
      "epoch": 0.06752453829591518,
      "grad_norm": 4.472171783447266,
      "learning_rate": 9.900086712453787e-05,
      "loss": 0.1774,
      "step": 1670
    },
    {
      "epoch": 0.06756497215118218,
      "grad_norm": 4.681826591491699,
      "learning_rate": 9.899959820119575e-05,
      "loss": 0.1326,
      "step": 1671
    },
    {
      "epoch": 0.0676054060064492,
      "grad_norm": 4.482563495635986,
      "learning_rate": 9.899832848072544e-05,
      "loss": 0.1417,
      "step": 1672
    },
    {
      "epoch": 0.06764583986171621,
      "grad_norm": 3.6906306743621826,
      "learning_rate": 9.899705796314756e-05,
      "loss": 0.1001,
      "step": 1673
    },
    {
      "epoch": 0.06768627371698323,
      "grad_norm": 1.5792145729064941,
      "learning_rate": 9.899578664848281e-05,
      "loss": 0.0806,
      "step": 1674
    },
    {
      "epoch": 0.06772670757225024,
      "grad_norm": 4.083103656768799,
      "learning_rate": 9.899451453675188e-05,
      "loss": 0.3086,
      "step": 1675
    },
    {
      "epoch": 0.06776714142751726,
      "grad_norm": 2.6956233978271484,
      "learning_rate": 9.899324162797543e-05,
      "loss": 0.1453,
      "step": 1676
    },
    {
      "epoch": 0.06780757528278428,
      "grad_norm": 2.557671070098877,
      "learning_rate": 9.89919679221742e-05,
      "loss": 0.0926,
      "step": 1677
    },
    {
      "epoch": 0.0678480091380513,
      "grad_norm": 5.007843494415283,
      "learning_rate": 9.899069341936888e-05,
      "loss": 0.1921,
      "step": 1678
    },
    {
      "epoch": 0.06788844299331831,
      "grad_norm": 5.179987907409668,
      "learning_rate": 9.898941811958024e-05,
      "loss": 0.2611,
      "step": 1679
    },
    {
      "epoch": 0.06792887684858533,
      "grad_norm": 6.86572790145874,
      "learning_rate": 9.8988142022829e-05,
      "loss": 0.1496,
      "step": 1680
    },
    {
      "epoch": 0.06796931070385233,
      "grad_norm": 4.812069892883301,
      "learning_rate": 9.898686512913592e-05,
      "loss": 0.1766,
      "step": 1681
    },
    {
      "epoch": 0.06800974455911935,
      "grad_norm": 3.84023380279541,
      "learning_rate": 9.898558743852179e-05,
      "loss": 0.113,
      "step": 1682
    },
    {
      "epoch": 0.06805017841438636,
      "grad_norm": 3.952836513519287,
      "learning_rate": 9.898430895100737e-05,
      "loss": 0.1918,
      "step": 1683
    },
    {
      "epoch": 0.06809061226965338,
      "grad_norm": 4.784387588500977,
      "learning_rate": 9.898302966661348e-05,
      "loss": 0.1841,
      "step": 1684
    },
    {
      "epoch": 0.0681310461249204,
      "grad_norm": 8.189531326293945,
      "learning_rate": 9.898174958536092e-05,
      "loss": 0.2744,
      "step": 1685
    },
    {
      "epoch": 0.06817147998018741,
      "grad_norm": 6.333024978637695,
      "learning_rate": 9.898046870727054e-05,
      "loss": 0.1276,
      "step": 1686
    },
    {
      "epoch": 0.06821191383545443,
      "grad_norm": 3.9589357376098633,
      "learning_rate": 9.897918703236312e-05,
      "loss": 0.2104,
      "step": 1687
    },
    {
      "epoch": 0.06825234769072144,
      "grad_norm": 6.971314430236816,
      "learning_rate": 9.897790456065958e-05,
      "loss": 0.4909,
      "step": 1688
    },
    {
      "epoch": 0.06829278154598846,
      "grad_norm": 4.812754154205322,
      "learning_rate": 9.897662129218074e-05,
      "loss": 0.1917,
      "step": 1689
    },
    {
      "epoch": 0.06833321540125548,
      "grad_norm": 3.245166778564453,
      "learning_rate": 9.897533722694748e-05,
      "loss": 0.1622,
      "step": 1690
    },
    {
      "epoch": 0.06837364925652249,
      "grad_norm": 4.595429420471191,
      "learning_rate": 9.897405236498071e-05,
      "loss": 0.1888,
      "step": 1691
    },
    {
      "epoch": 0.0684140831117895,
      "grad_norm": 2.741359233856201,
      "learning_rate": 9.897276670630131e-05,
      "loss": 0.173,
      "step": 1692
    },
    {
      "epoch": 0.06845451696705651,
      "grad_norm": 3.515491008758545,
      "learning_rate": 9.89714802509302e-05,
      "loss": 0.1991,
      "step": 1693
    },
    {
      "epoch": 0.06849495082232353,
      "grad_norm": 4.728665351867676,
      "learning_rate": 9.897019299888831e-05,
      "loss": 0.1738,
      "step": 1694
    },
    {
      "epoch": 0.06853538467759054,
      "grad_norm": 4.878288269042969,
      "learning_rate": 9.896890495019658e-05,
      "loss": 0.223,
      "step": 1695
    },
    {
      "epoch": 0.06857581853285756,
      "grad_norm": 6.2309088706970215,
      "learning_rate": 9.896761610487597e-05,
      "loss": 0.2616,
      "step": 1696
    },
    {
      "epoch": 0.06861625238812458,
      "grad_norm": 4.523493766784668,
      "learning_rate": 9.896632646294746e-05,
      "loss": 0.1321,
      "step": 1697
    },
    {
      "epoch": 0.0686566862433916,
      "grad_norm": 1.3658608198165894,
      "learning_rate": 9.896503602443198e-05,
      "loss": 0.0774,
      "step": 1698
    },
    {
      "epoch": 0.06869712009865861,
      "grad_norm": 8.51968002319336,
      "learning_rate": 9.896374478935058e-05,
      "loss": 0.2471,
      "step": 1699
    },
    {
      "epoch": 0.06873755395392563,
      "grad_norm": 3.4419479370117188,
      "learning_rate": 9.896245275772425e-05,
      "loss": 0.1941,
      "step": 1700
    },
    {
      "epoch": 0.06877798780919264,
      "grad_norm": 5.730888843536377,
      "learning_rate": 9.896115992957398e-05,
      "loss": 0.1819,
      "step": 1701
    },
    {
      "epoch": 0.06881842166445966,
      "grad_norm": 4.877986907958984,
      "learning_rate": 9.895986630492082e-05,
      "loss": 0.1917,
      "step": 1702
    },
    {
      "epoch": 0.06885885551972666,
      "grad_norm": 8.888043403625488,
      "learning_rate": 9.895857188378583e-05,
      "loss": 0.2105,
      "step": 1703
    },
    {
      "epoch": 0.06889928937499368,
      "grad_norm": 4.914479732513428,
      "learning_rate": 9.895727666619006e-05,
      "loss": 0.1268,
      "step": 1704
    },
    {
      "epoch": 0.0689397232302607,
      "grad_norm": 8.112187385559082,
      "learning_rate": 9.895598065215456e-05,
      "loss": 0.1783,
      "step": 1705
    },
    {
      "epoch": 0.06898015708552771,
      "grad_norm": 2.6471598148345947,
      "learning_rate": 9.895468384170044e-05,
      "loss": 0.1038,
      "step": 1706
    },
    {
      "epoch": 0.06902059094079473,
      "grad_norm": 5.051476955413818,
      "learning_rate": 9.89533862348488e-05,
      "loss": 0.3676,
      "step": 1707
    },
    {
      "epoch": 0.06906102479606174,
      "grad_norm": 13.665288925170898,
      "learning_rate": 9.89520878316207e-05,
      "loss": 0.4902,
      "step": 1708
    },
    {
      "epoch": 0.06910145865132876,
      "grad_norm": 2.598788022994995,
      "learning_rate": 9.895078863203733e-05,
      "loss": 0.1067,
      "step": 1709
    },
    {
      "epoch": 0.06914189250659578,
      "grad_norm": 3.8667001724243164,
      "learning_rate": 9.89494886361198e-05,
      "loss": 0.2384,
      "step": 1710
    },
    {
      "epoch": 0.06918232636186279,
      "grad_norm": 7.378573417663574,
      "learning_rate": 9.894818784388922e-05,
      "loss": 0.2095,
      "step": 1711
    },
    {
      "epoch": 0.06922276021712981,
      "grad_norm": 6.007424354553223,
      "learning_rate": 9.89468862553668e-05,
      "loss": 0.291,
      "step": 1712
    },
    {
      "epoch": 0.06926319407239681,
      "grad_norm": 11.56386947631836,
      "learning_rate": 9.894558387057368e-05,
      "loss": 0.3018,
      "step": 1713
    },
    {
      "epoch": 0.06930362792766383,
      "grad_norm": 4.819494247436523,
      "learning_rate": 9.89442806895311e-05,
      "loss": 0.2336,
      "step": 1714
    },
    {
      "epoch": 0.06934406178293084,
      "grad_norm": 2.284614324569702,
      "learning_rate": 9.89429767122602e-05,
      "loss": 0.2113,
      "step": 1715
    },
    {
      "epoch": 0.06938449563819786,
      "grad_norm": 2.8937432765960693,
      "learning_rate": 9.894167193878222e-05,
      "loss": 0.2065,
      "step": 1716
    },
    {
      "epoch": 0.06942492949346488,
      "grad_norm": 4.092549800872803,
      "learning_rate": 9.894036636911839e-05,
      "loss": 0.1079,
      "step": 1717
    },
    {
      "epoch": 0.06946536334873189,
      "grad_norm": 7.1499457359313965,
      "learning_rate": 9.893906000328994e-05,
      "loss": 0.1694,
      "step": 1718
    },
    {
      "epoch": 0.06950579720399891,
      "grad_norm": 5.406733512878418,
      "learning_rate": 9.893775284131813e-05,
      "loss": 0.1679,
      "step": 1719
    },
    {
      "epoch": 0.06954623105926593,
      "grad_norm": 3.489213705062866,
      "learning_rate": 9.893644488322422e-05,
      "loss": 0.1687,
      "step": 1720
    },
    {
      "epoch": 0.06958666491453294,
      "grad_norm": 2.1845614910125732,
      "learning_rate": 9.893513612902949e-05,
      "loss": 0.1908,
      "step": 1721
    },
    {
      "epoch": 0.06962709876979996,
      "grad_norm": 2.856152057647705,
      "learning_rate": 9.893382657875523e-05,
      "loss": 0.084,
      "step": 1722
    },
    {
      "epoch": 0.06966753262506697,
      "grad_norm": 4.4165143966674805,
      "learning_rate": 9.893251623242273e-05,
      "loss": 0.1795,
      "step": 1723
    },
    {
      "epoch": 0.06970796648033398,
      "grad_norm": 4.603332996368408,
      "learning_rate": 9.893120509005333e-05,
      "loss": 0.1812,
      "step": 1724
    },
    {
      "epoch": 0.069748400335601,
      "grad_norm": 5.745676040649414,
      "learning_rate": 9.892989315166835e-05,
      "loss": 0.1969,
      "step": 1725
    },
    {
      "epoch": 0.06978883419086801,
      "grad_norm": 6.590615272521973,
      "learning_rate": 9.892858041728914e-05,
      "loss": 0.2675,
      "step": 1726
    },
    {
      "epoch": 0.06982926804613503,
      "grad_norm": 4.123551368713379,
      "learning_rate": 9.892726688693703e-05,
      "loss": 0.2545,
      "step": 1727
    },
    {
      "epoch": 0.06986970190140204,
      "grad_norm": 7.049654006958008,
      "learning_rate": 9.892595256063342e-05,
      "loss": 0.2877,
      "step": 1728
    },
    {
      "epoch": 0.06991013575666906,
      "grad_norm": 1.425628423690796,
      "learning_rate": 9.892463743839968e-05,
      "loss": 0.0593,
      "step": 1729
    },
    {
      "epoch": 0.06995056961193608,
      "grad_norm": 7.095800399780273,
      "learning_rate": 9.89233215202572e-05,
      "loss": 0.2271,
      "step": 1730
    },
    {
      "epoch": 0.06999100346720309,
      "grad_norm": 4.90399169921875,
      "learning_rate": 9.892200480622737e-05,
      "loss": 0.1929,
      "step": 1731
    },
    {
      "epoch": 0.07003143732247011,
      "grad_norm": 7.958353519439697,
      "learning_rate": 9.892068729633166e-05,
      "loss": 0.4683,
      "step": 1732
    },
    {
      "epoch": 0.07007187117773712,
      "grad_norm": 6.288543224334717,
      "learning_rate": 9.891936899059147e-05,
      "loss": 0.2399,
      "step": 1733
    },
    {
      "epoch": 0.07011230503300414,
      "grad_norm": 7.064364433288574,
      "learning_rate": 9.891804988902824e-05,
      "loss": 0.203,
      "step": 1734
    },
    {
      "epoch": 0.07015273888827114,
      "grad_norm": 4.235506534576416,
      "learning_rate": 9.891672999166343e-05,
      "loss": 0.1049,
      "step": 1735
    },
    {
      "epoch": 0.07019317274353816,
      "grad_norm": 6.186842441558838,
      "learning_rate": 9.891540929851855e-05,
      "loss": 0.2688,
      "step": 1736
    },
    {
      "epoch": 0.07023360659880518,
      "grad_norm": 5.31261682510376,
      "learning_rate": 9.891408780961503e-05,
      "loss": 0.254,
      "step": 1737
    },
    {
      "epoch": 0.07027404045407219,
      "grad_norm": 6.877195358276367,
      "learning_rate": 9.89127655249744e-05,
      "loss": 0.1693,
      "step": 1738
    },
    {
      "epoch": 0.07031447430933921,
      "grad_norm": 7.427832126617432,
      "learning_rate": 9.891144244461818e-05,
      "loss": 0.2231,
      "step": 1739
    },
    {
      "epoch": 0.07035490816460623,
      "grad_norm": 11.206794738769531,
      "learning_rate": 9.891011856856786e-05,
      "loss": 0.3556,
      "step": 1740
    },
    {
      "epoch": 0.07039534201987324,
      "grad_norm": 3.773075819015503,
      "learning_rate": 9.890879389684502e-05,
      "loss": 0.0794,
      "step": 1741
    },
    {
      "epoch": 0.07043577587514026,
      "grad_norm": 11.188713073730469,
      "learning_rate": 9.890746842947116e-05,
      "loss": 0.2577,
      "step": 1742
    },
    {
      "epoch": 0.07047620973040727,
      "grad_norm": 11.410466194152832,
      "learning_rate": 9.890614216646788e-05,
      "loss": 0.3283,
      "step": 1743
    },
    {
      "epoch": 0.07051664358567429,
      "grad_norm": 6.486424922943115,
      "learning_rate": 9.890481510785675e-05,
      "loss": 0.1716,
      "step": 1744
    },
    {
      "epoch": 0.07055707744094131,
      "grad_norm": 6.52285623550415,
      "learning_rate": 9.890348725365935e-05,
      "loss": 0.2409,
      "step": 1745
    },
    {
      "epoch": 0.07059751129620831,
      "grad_norm": 4.26342248916626,
      "learning_rate": 9.890215860389729e-05,
      "loss": 0.1132,
      "step": 1746
    },
    {
      "epoch": 0.07063794515147533,
      "grad_norm": 4.388311862945557,
      "learning_rate": 9.890082915859217e-05,
      "loss": 0.0892,
      "step": 1747
    },
    {
      "epoch": 0.07067837900674234,
      "grad_norm": 2.906832695007324,
      "learning_rate": 9.889949891776564e-05,
      "loss": 0.1828,
      "step": 1748
    },
    {
      "epoch": 0.07071881286200936,
      "grad_norm": 8.845560073852539,
      "learning_rate": 9.88981678814393e-05,
      "loss": 0.289,
      "step": 1749
    },
    {
      "epoch": 0.07075924671727638,
      "grad_norm": 7.173463821411133,
      "learning_rate": 9.889683604963486e-05,
      "loss": 0.1476,
      "step": 1750
    },
    {
      "epoch": 0.07079968057254339,
      "grad_norm": 4.896988868713379,
      "learning_rate": 9.889550342237395e-05,
      "loss": 0.2741,
      "step": 1751
    },
    {
      "epoch": 0.07084011442781041,
      "grad_norm": 3.463782548904419,
      "learning_rate": 9.889416999967825e-05,
      "loss": 0.1893,
      "step": 1752
    },
    {
      "epoch": 0.07088054828307742,
      "grad_norm": 6.596735000610352,
      "learning_rate": 9.889283578156947e-05,
      "loss": 0.1837,
      "step": 1753
    },
    {
      "epoch": 0.07092098213834444,
      "grad_norm": 10.155313491821289,
      "learning_rate": 9.88915007680693e-05,
      "loss": 0.2983,
      "step": 1754
    },
    {
      "epoch": 0.07096141599361146,
      "grad_norm": 9.27284049987793,
      "learning_rate": 9.889016495919945e-05,
      "loss": 0.241,
      "step": 1755
    },
    {
      "epoch": 0.07100184984887846,
      "grad_norm": 4.435075283050537,
      "learning_rate": 9.888882835498167e-05,
      "loss": 0.1861,
      "step": 1756
    },
    {
      "epoch": 0.07104228370414548,
      "grad_norm": 5.619365692138672,
      "learning_rate": 9.88874909554377e-05,
      "loss": 0.1971,
      "step": 1757
    },
    {
      "epoch": 0.07108271755941249,
      "grad_norm": 5.913852214813232,
      "learning_rate": 9.88861527605893e-05,
      "loss": 0.1867,
      "step": 1758
    },
    {
      "epoch": 0.07112315141467951,
      "grad_norm": 1.912200927734375,
      "learning_rate": 9.888481377045822e-05,
      "loss": 0.1504,
      "step": 1759
    },
    {
      "epoch": 0.07116358526994652,
      "grad_norm": 7.763233184814453,
      "learning_rate": 9.888347398506625e-05,
      "loss": 0.242,
      "step": 1760
    },
    {
      "epoch": 0.07120401912521354,
      "grad_norm": 8.070950508117676,
      "learning_rate": 9.888213340443522e-05,
      "loss": 0.1932,
      "step": 1761
    },
    {
      "epoch": 0.07124445298048056,
      "grad_norm": 7.547504425048828,
      "learning_rate": 9.888079202858688e-05,
      "loss": 0.1767,
      "step": 1762
    },
    {
      "epoch": 0.07128488683574757,
      "grad_norm": 4.408570289611816,
      "learning_rate": 9.887944985754312e-05,
      "loss": 0.1265,
      "step": 1763
    },
    {
      "epoch": 0.07132532069101459,
      "grad_norm": 7.289517402648926,
      "learning_rate": 9.88781068913257e-05,
      "loss": 0.219,
      "step": 1764
    },
    {
      "epoch": 0.0713657545462816,
      "grad_norm": 5.639768600463867,
      "learning_rate": 9.887676312995653e-05,
      "loss": 0.1574,
      "step": 1765
    },
    {
      "epoch": 0.07140618840154862,
      "grad_norm": 5.720702171325684,
      "learning_rate": 9.887541857345742e-05,
      "loss": 0.1803,
      "step": 1766
    },
    {
      "epoch": 0.07144662225681563,
      "grad_norm": 4.559646129608154,
      "learning_rate": 9.88740732218503e-05,
      "loss": 0.189,
      "step": 1767
    },
    {
      "epoch": 0.07148705611208264,
      "grad_norm": 3.4875025749206543,
      "learning_rate": 9.8872727075157e-05,
      "loss": 0.1879,
      "step": 1768
    },
    {
      "epoch": 0.07152748996734966,
      "grad_norm": 3.639653444290161,
      "learning_rate": 9.887138013339944e-05,
      "loss": 0.1976,
      "step": 1769
    },
    {
      "epoch": 0.07156792382261667,
      "grad_norm": 3.2622907161712646,
      "learning_rate": 9.887003239659954e-05,
      "loss": 0.1959,
      "step": 1770
    },
    {
      "epoch": 0.07160835767788369,
      "grad_norm": 5.338065147399902,
      "learning_rate": 9.886868386477923e-05,
      "loss": 0.1814,
      "step": 1771
    },
    {
      "epoch": 0.07164879153315071,
      "grad_norm": 4.034976005554199,
      "learning_rate": 9.886733453796043e-05,
      "loss": 0.1247,
      "step": 1772
    },
    {
      "epoch": 0.07168922538841772,
      "grad_norm": 6.7332892417907715,
      "learning_rate": 9.88659844161651e-05,
      "loss": 0.1246,
      "step": 1773
    },
    {
      "epoch": 0.07172965924368474,
      "grad_norm": 6.367109298706055,
      "learning_rate": 9.886463349941522e-05,
      "loss": 0.2251,
      "step": 1774
    },
    {
      "epoch": 0.07177009309895176,
      "grad_norm": 5.285366535186768,
      "learning_rate": 9.886328178773273e-05,
      "loss": 0.1216,
      "step": 1775
    },
    {
      "epoch": 0.07181052695421877,
      "grad_norm": 6.861854553222656,
      "learning_rate": 9.886192928113964e-05,
      "loss": 0.2946,
      "step": 1776
    },
    {
      "epoch": 0.07185096080948579,
      "grad_norm": 8.776226043701172,
      "learning_rate": 9.886057597965797e-05,
      "loss": 0.2908,
      "step": 1777
    },
    {
      "epoch": 0.07189139466475279,
      "grad_norm": 6.314785957336426,
      "learning_rate": 9.88592218833097e-05,
      "loss": 0.1597,
      "step": 1778
    },
    {
      "epoch": 0.07193182852001981,
      "grad_norm": 5.901515007019043,
      "learning_rate": 9.885786699211687e-05,
      "loss": 0.2348,
      "step": 1779
    },
    {
      "epoch": 0.07197226237528682,
      "grad_norm": 2.105095386505127,
      "learning_rate": 9.885651130610156e-05,
      "loss": 0.0945,
      "step": 1780
    },
    {
      "epoch": 0.07201269623055384,
      "grad_norm": 8.823195457458496,
      "learning_rate": 9.885515482528577e-05,
      "loss": 0.1881,
      "step": 1781
    },
    {
      "epoch": 0.07205313008582086,
      "grad_norm": 6.465571880340576,
      "learning_rate": 9.88537975496916e-05,
      "loss": 0.3334,
      "step": 1782
    },
    {
      "epoch": 0.07209356394108787,
      "grad_norm": 4.3808817863464355,
      "learning_rate": 9.885243947934111e-05,
      "loss": 0.0981,
      "step": 1783
    },
    {
      "epoch": 0.07213399779635489,
      "grad_norm": 10.881611824035645,
      "learning_rate": 9.88510806142564e-05,
      "loss": 0.225,
      "step": 1784
    },
    {
      "epoch": 0.0721744316516219,
      "grad_norm": 6.170853137969971,
      "learning_rate": 9.884972095445959e-05,
      "loss": 0.1647,
      "step": 1785
    },
    {
      "epoch": 0.07221486550688892,
      "grad_norm": 8.502425193786621,
      "learning_rate": 9.88483604999728e-05,
      "loss": 0.2078,
      "step": 1786
    },
    {
      "epoch": 0.07225529936215594,
      "grad_norm": 7.589778423309326,
      "learning_rate": 9.884699925081811e-05,
      "loss": 0.1635,
      "step": 1787
    },
    {
      "epoch": 0.07229573321742294,
      "grad_norm": 4.747900009155273,
      "learning_rate": 9.884563720701773e-05,
      "loss": 0.1358,
      "step": 1788
    },
    {
      "epoch": 0.07233616707268996,
      "grad_norm": 7.487351894378662,
      "learning_rate": 9.884427436859381e-05,
      "loss": 0.2566,
      "step": 1789
    },
    {
      "epoch": 0.07237660092795697,
      "grad_norm": 6.2125067710876465,
      "learning_rate": 9.884291073556849e-05,
      "loss": 0.3197,
      "step": 1790
    },
    {
      "epoch": 0.07241703478322399,
      "grad_norm": 3.934527635574341,
      "learning_rate": 9.884154630796395e-05,
      "loss": 0.1374,
      "step": 1791
    },
    {
      "epoch": 0.07245746863849101,
      "grad_norm": 3.6763224601745605,
      "learning_rate": 9.884018108580242e-05,
      "loss": 0.1355,
      "step": 1792
    },
    {
      "epoch": 0.07249790249375802,
      "grad_norm": 5.618278980255127,
      "learning_rate": 9.883881506910609e-05,
      "loss": 0.1686,
      "step": 1793
    },
    {
      "epoch": 0.07253833634902504,
      "grad_norm": 4.192503452301025,
      "learning_rate": 9.883744825789718e-05,
      "loss": 0.1548,
      "step": 1794
    },
    {
      "epoch": 0.07257877020429206,
      "grad_norm": 3.6635074615478516,
      "learning_rate": 9.883608065219793e-05,
      "loss": 0.2257,
      "step": 1795
    },
    {
      "epoch": 0.07261920405955907,
      "grad_norm": 2.716395378112793,
      "learning_rate": 9.883471225203061e-05,
      "loss": 0.0854,
      "step": 1796
    },
    {
      "epoch": 0.07265963791482609,
      "grad_norm": 15.645360946655273,
      "learning_rate": 9.883334305741746e-05,
      "loss": 0.2916,
      "step": 1797
    },
    {
      "epoch": 0.0727000717700931,
      "grad_norm": 3.3946714401245117,
      "learning_rate": 9.883197306838073e-05,
      "loss": 0.1074,
      "step": 1798
    },
    {
      "epoch": 0.07274050562536011,
      "grad_norm": 1.6748510599136353,
      "learning_rate": 9.883060228494274e-05,
      "loss": 0.037,
      "step": 1799
    },
    {
      "epoch": 0.07278093948062712,
      "grad_norm": 10.79991626739502,
      "learning_rate": 9.882923070712579e-05,
      "loss": 0.2289,
      "step": 1800
    },
    {
      "epoch": 0.07282137333589414,
      "grad_norm": 5.487301826477051,
      "learning_rate": 9.882785833495218e-05,
      "loss": 0.1929,
      "step": 1801
    },
    {
      "epoch": 0.07286180719116116,
      "grad_norm": 5.086686134338379,
      "learning_rate": 9.882648516844424e-05,
      "loss": 0.2871,
      "step": 1802
    },
    {
      "epoch": 0.07290224104642817,
      "grad_norm": 4.274926662445068,
      "learning_rate": 9.882511120762432e-05,
      "loss": 0.1636,
      "step": 1803
    },
    {
      "epoch": 0.07294267490169519,
      "grad_norm": 5.261094570159912,
      "learning_rate": 9.882373645251475e-05,
      "loss": 0.2125,
      "step": 1804
    },
    {
      "epoch": 0.0729831087569622,
      "grad_norm": 6.359926700592041,
      "learning_rate": 9.88223609031379e-05,
      "loss": 0.1475,
      "step": 1805
    },
    {
      "epoch": 0.07302354261222922,
      "grad_norm": 6.117772579193115,
      "learning_rate": 9.882098455951617e-05,
      "loss": 0.1571,
      "step": 1806
    },
    {
      "epoch": 0.07306397646749624,
      "grad_norm": 5.28400182723999,
      "learning_rate": 9.881960742167192e-05,
      "loss": 0.1574,
      "step": 1807
    },
    {
      "epoch": 0.07310441032276326,
      "grad_norm": 2.850966453552246,
      "learning_rate": 9.881822948962758e-05,
      "loss": 0.1363,
      "step": 1808
    },
    {
      "epoch": 0.07314484417803027,
      "grad_norm": 8.033350944519043,
      "learning_rate": 9.881685076340555e-05,
      "loss": 0.3996,
      "step": 1809
    },
    {
      "epoch": 0.07318527803329727,
      "grad_norm": 4.820026874542236,
      "learning_rate": 9.881547124302825e-05,
      "loss": 0.1059,
      "step": 1810
    },
    {
      "epoch": 0.07322571188856429,
      "grad_norm": 11.148386001586914,
      "learning_rate": 9.881409092851814e-05,
      "loss": 0.2694,
      "step": 1811
    },
    {
      "epoch": 0.0732661457438313,
      "grad_norm": 4.292884826660156,
      "learning_rate": 9.881270981989767e-05,
      "loss": 0.1593,
      "step": 1812
    },
    {
      "epoch": 0.07330657959909832,
      "grad_norm": 6.642562389373779,
      "learning_rate": 9.88113279171893e-05,
      "loss": 0.1495,
      "step": 1813
    },
    {
      "epoch": 0.07334701345436534,
      "grad_norm": 10.558499336242676,
      "learning_rate": 9.880994522041554e-05,
      "loss": 0.1776,
      "step": 1814
    },
    {
      "epoch": 0.07338744730963236,
      "grad_norm": 8.492796897888184,
      "learning_rate": 9.880856172959885e-05,
      "loss": 0.2976,
      "step": 1815
    },
    {
      "epoch": 0.07342788116489937,
      "grad_norm": 5.0179643630981445,
      "learning_rate": 9.880717744476173e-05,
      "loss": 0.1151,
      "step": 1816
    },
    {
      "epoch": 0.07346831502016639,
      "grad_norm": 3.9338204860687256,
      "learning_rate": 9.880579236592675e-05,
      "loss": 0.2098,
      "step": 1817
    },
    {
      "epoch": 0.0735087488754334,
      "grad_norm": 7.972073554992676,
      "learning_rate": 9.88044064931164e-05,
      "loss": 0.2283,
      "step": 1818
    },
    {
      "epoch": 0.07354918273070042,
      "grad_norm": 4.294610500335693,
      "learning_rate": 9.880301982635323e-05,
      "loss": 0.1677,
      "step": 1819
    },
    {
      "epoch": 0.07358961658596742,
      "grad_norm": 3.342832565307617,
      "learning_rate": 9.88016323656598e-05,
      "loss": 0.0881,
      "step": 1820
    },
    {
      "epoch": 0.07363005044123444,
      "grad_norm": 4.807898998260498,
      "learning_rate": 9.880024411105871e-05,
      "loss": 0.1065,
      "step": 1821
    },
    {
      "epoch": 0.07367048429650146,
      "grad_norm": 5.5678791999816895,
      "learning_rate": 9.879885506257249e-05,
      "loss": 0.3009,
      "step": 1822
    },
    {
      "epoch": 0.07371091815176847,
      "grad_norm": 6.502264499664307,
      "learning_rate": 9.879746522022379e-05,
      "loss": 0.2889,
      "step": 1823
    },
    {
      "epoch": 0.07375135200703549,
      "grad_norm": 4.883254051208496,
      "learning_rate": 9.879607458403519e-05,
      "loss": 0.1979,
      "step": 1824
    },
    {
      "epoch": 0.0737917858623025,
      "grad_norm": 4.912976264953613,
      "learning_rate": 9.879468315402931e-05,
      "loss": 0.1988,
      "step": 1825
    },
    {
      "epoch": 0.07383221971756952,
      "grad_norm": 8.4829683303833,
      "learning_rate": 9.879329093022882e-05,
      "loss": 0.1673,
      "step": 1826
    },
    {
      "epoch": 0.07387265357283654,
      "grad_norm": 4.379247665405273,
      "learning_rate": 9.879189791265633e-05,
      "loss": 0.295,
      "step": 1827
    },
    {
      "epoch": 0.07391308742810355,
      "grad_norm": 4.821085453033447,
      "learning_rate": 9.879050410133451e-05,
      "loss": 0.2234,
      "step": 1828
    },
    {
      "epoch": 0.07395352128337057,
      "grad_norm": 6.617440700531006,
      "learning_rate": 9.878910949628605e-05,
      "loss": 0.169,
      "step": 1829
    },
    {
      "epoch": 0.07399395513863759,
      "grad_norm": 4.656126022338867,
      "learning_rate": 9.878771409753363e-05,
      "loss": 0.2596,
      "step": 1830
    },
    {
      "epoch": 0.07403438899390459,
      "grad_norm": 7.026053428649902,
      "learning_rate": 9.878631790509996e-05,
      "loss": 0.2149,
      "step": 1831
    },
    {
      "epoch": 0.0740748228491716,
      "grad_norm": 5.384998798370361,
      "learning_rate": 9.878492091900771e-05,
      "loss": 0.1947,
      "step": 1832
    },
    {
      "epoch": 0.07411525670443862,
      "grad_norm": 2.0235700607299805,
      "learning_rate": 9.878352313927968e-05,
      "loss": 0.1117,
      "step": 1833
    },
    {
      "epoch": 0.07415569055970564,
      "grad_norm": 6.138808727264404,
      "learning_rate": 9.878212456593854e-05,
      "loss": 0.2326,
      "step": 1834
    },
    {
      "epoch": 0.07419612441497266,
      "grad_norm": 3.449012279510498,
      "learning_rate": 9.878072519900707e-05,
      "loss": 0.1397,
      "step": 1835
    },
    {
      "epoch": 0.07423655827023967,
      "grad_norm": 5.557662010192871,
      "learning_rate": 9.877932503850804e-05,
      "loss": 0.1734,
      "step": 1836
    },
    {
      "epoch": 0.07427699212550669,
      "grad_norm": 6.8185577392578125,
      "learning_rate": 9.877792408446423e-05,
      "loss": 0.172,
      "step": 1837
    },
    {
      "epoch": 0.0743174259807737,
      "grad_norm": 2.0890164375305176,
      "learning_rate": 9.877652233689843e-05,
      "loss": 0.1338,
      "step": 1838
    },
    {
      "epoch": 0.07435785983604072,
      "grad_norm": 2.2277419567108154,
      "learning_rate": 9.877511979583341e-05,
      "loss": 0.1603,
      "step": 1839
    },
    {
      "epoch": 0.07439829369130774,
      "grad_norm": 4.193663120269775,
      "learning_rate": 9.877371646129203e-05,
      "loss": 0.1688,
      "step": 1840
    },
    {
      "epoch": 0.07443872754657475,
      "grad_norm": 2.9034907817840576,
      "learning_rate": 9.87723123332971e-05,
      "loss": 0.1904,
      "step": 1841
    },
    {
      "epoch": 0.07447916140184176,
      "grad_norm": 4.024078845977783,
      "learning_rate": 9.877090741187148e-05,
      "loss": 0.1095,
      "step": 1842
    },
    {
      "epoch": 0.07451959525710877,
      "grad_norm": 2.67682147026062,
      "learning_rate": 9.8769501697038e-05,
      "loss": 0.1007,
      "step": 1843
    },
    {
      "epoch": 0.07456002911237579,
      "grad_norm": 4.456399917602539,
      "learning_rate": 9.876809518881954e-05,
      "loss": 0.2081,
      "step": 1844
    },
    {
      "epoch": 0.0746004629676428,
      "grad_norm": 3.163957357406616,
      "learning_rate": 9.8766687887239e-05,
      "loss": 0.095,
      "step": 1845
    },
    {
      "epoch": 0.07464089682290982,
      "grad_norm": 9.046263694763184,
      "learning_rate": 9.876527979231922e-05,
      "loss": 0.2232,
      "step": 1846
    },
    {
      "epoch": 0.07468133067817684,
      "grad_norm": 6.235287189483643,
      "learning_rate": 9.876387090408316e-05,
      "loss": 0.3407,
      "step": 1847
    },
    {
      "epoch": 0.07472176453344385,
      "grad_norm": 8.367833137512207,
      "learning_rate": 9.876246122255372e-05,
      "loss": 0.1951,
      "step": 1848
    },
    {
      "epoch": 0.07476219838871087,
      "grad_norm": 6.345910549163818,
      "learning_rate": 9.876105074775383e-05,
      "loss": 0.1104,
      "step": 1849
    },
    {
      "epoch": 0.07480263224397789,
      "grad_norm": 7.167973041534424,
      "learning_rate": 9.875963947970646e-05,
      "loss": 0.2139,
      "step": 1850
    },
    {
      "epoch": 0.0748430660992449,
      "grad_norm": 8.131857872009277,
      "learning_rate": 9.875822741843453e-05,
      "loss": 0.2087,
      "step": 1851
    },
    {
      "epoch": 0.0748834999545119,
      "grad_norm": 5.76434850692749,
      "learning_rate": 9.875681456396103e-05,
      "loss": 0.1122,
      "step": 1852
    },
    {
      "epoch": 0.07492393380977892,
      "grad_norm": 7.049228191375732,
      "learning_rate": 9.875540091630895e-05,
      "loss": 0.2883,
      "step": 1853
    },
    {
      "epoch": 0.07496436766504594,
      "grad_norm": 7.535574436187744,
      "learning_rate": 9.875398647550128e-05,
      "loss": 0.1939,
      "step": 1854
    },
    {
      "epoch": 0.07500480152031296,
      "grad_norm": 4.074461936950684,
      "learning_rate": 9.875257124156103e-05,
      "loss": 0.1422,
      "step": 1855
    },
    {
      "epoch": 0.07504523537557997,
      "grad_norm": 4.21569299697876,
      "learning_rate": 9.875115521451124e-05,
      "loss": 0.2564,
      "step": 1856
    },
    {
      "epoch": 0.07508566923084699,
      "grad_norm": 2.96022629737854,
      "learning_rate": 9.874973839437491e-05,
      "loss": 0.2066,
      "step": 1857
    },
    {
      "epoch": 0.075126103086114,
      "grad_norm": 3.6212940216064453,
      "learning_rate": 9.874832078117513e-05,
      "loss": 0.2222,
      "step": 1858
    },
    {
      "epoch": 0.07516653694138102,
      "grad_norm": 6.8688530921936035,
      "learning_rate": 9.874690237493491e-05,
      "loss": 0.1846,
      "step": 1859
    },
    {
      "epoch": 0.07520697079664804,
      "grad_norm": 12.21948528289795,
      "learning_rate": 9.874548317567738e-05,
      "loss": 0.3241,
      "step": 1860
    },
    {
      "epoch": 0.07524740465191505,
      "grad_norm": 8.776249885559082,
      "learning_rate": 9.874406318342562e-05,
      "loss": 0.3113,
      "step": 1861
    },
    {
      "epoch": 0.07528783850718207,
      "grad_norm": 5.742241859436035,
      "learning_rate": 9.874264239820269e-05,
      "loss": 0.1789,
      "step": 1862
    },
    {
      "epoch": 0.07532827236244907,
      "grad_norm": 6.469368934631348,
      "learning_rate": 9.874122082003173e-05,
      "loss": 0.2396,
      "step": 1863
    },
    {
      "epoch": 0.07536870621771609,
      "grad_norm": 7.525454044342041,
      "learning_rate": 9.873979844893587e-05,
      "loss": 0.2176,
      "step": 1864
    },
    {
      "epoch": 0.0754091400729831,
      "grad_norm": 6.981338977813721,
      "learning_rate": 9.873837528493825e-05,
      "loss": 0.1842,
      "step": 1865
    },
    {
      "epoch": 0.07544957392825012,
      "grad_norm": 10.51588249206543,
      "learning_rate": 9.8736951328062e-05,
      "loss": 0.2452,
      "step": 1866
    },
    {
      "epoch": 0.07549000778351714,
      "grad_norm": 9.818849563598633,
      "learning_rate": 9.873552657833032e-05,
      "loss": 0.2102,
      "step": 1867
    },
    {
      "epoch": 0.07553044163878415,
      "grad_norm": 7.915666580200195,
      "learning_rate": 9.873410103576636e-05,
      "loss": 0.2595,
      "step": 1868
    },
    {
      "epoch": 0.07557087549405117,
      "grad_norm": 5.426427364349365,
      "learning_rate": 9.873267470039331e-05,
      "loss": 0.1371,
      "step": 1869
    },
    {
      "epoch": 0.07561130934931819,
      "grad_norm": 6.121982097625732,
      "learning_rate": 9.87312475722344e-05,
      "loss": 0.2061,
      "step": 1870
    },
    {
      "epoch": 0.0756517432045852,
      "grad_norm": 7.255677700042725,
      "learning_rate": 9.872981965131282e-05,
      "loss": 0.2106,
      "step": 1871
    },
    {
      "epoch": 0.07569217705985222,
      "grad_norm": 3.2915902137756348,
      "learning_rate": 9.87283909376518e-05,
      "loss": 0.0985,
      "step": 1872
    },
    {
      "epoch": 0.07573261091511924,
      "grad_norm": 2.166973114013672,
      "learning_rate": 9.872696143127462e-05,
      "loss": 0.1333,
      "step": 1873
    },
    {
      "epoch": 0.07577304477038624,
      "grad_norm": 3.8447041511535645,
      "learning_rate": 9.872553113220449e-05,
      "loss": 0.2124,
      "step": 1874
    },
    {
      "epoch": 0.07581347862565326,
      "grad_norm": 6.448738098144531,
      "learning_rate": 9.872410004046468e-05,
      "loss": 0.3089,
      "step": 1875
    },
    {
      "epoch": 0.07585391248092027,
      "grad_norm": 4.350379467010498,
      "learning_rate": 9.872266815607851e-05,
      "loss": 0.1726,
      "step": 1876
    },
    {
      "epoch": 0.07589434633618729,
      "grad_norm": 7.949905872344971,
      "learning_rate": 9.872123547906925e-05,
      "loss": 0.1599,
      "step": 1877
    },
    {
      "epoch": 0.0759347801914543,
      "grad_norm": 8.193962097167969,
      "learning_rate": 9.871980200946019e-05,
      "loss": 0.1903,
      "step": 1878
    },
    {
      "epoch": 0.07597521404672132,
      "grad_norm": 8.916481971740723,
      "learning_rate": 9.871836774727467e-05,
      "loss": 0.2599,
      "step": 1879
    },
    {
      "epoch": 0.07601564790198834,
      "grad_norm": 2.2683231830596924,
      "learning_rate": 9.871693269253603e-05,
      "loss": 0.0752,
      "step": 1880
    },
    {
      "epoch": 0.07605608175725535,
      "grad_norm": 8.631729125976562,
      "learning_rate": 9.87154968452676e-05,
      "loss": 0.2436,
      "step": 1881
    },
    {
      "epoch": 0.07609651561252237,
      "grad_norm": 3.8100814819335938,
      "learning_rate": 9.871406020549274e-05,
      "loss": 0.1195,
      "step": 1882
    },
    {
      "epoch": 0.07613694946778939,
      "grad_norm": 8.080832481384277,
      "learning_rate": 9.871262277323484e-05,
      "loss": 0.2223,
      "step": 1883
    },
    {
      "epoch": 0.07617738332305639,
      "grad_norm": 6.448827266693115,
      "learning_rate": 9.871118454851724e-05,
      "loss": 0.2387,
      "step": 1884
    },
    {
      "epoch": 0.0762178171783234,
      "grad_norm": 7.2695817947387695,
      "learning_rate": 9.870974553136339e-05,
      "loss": 0.2215,
      "step": 1885
    },
    {
      "epoch": 0.07625825103359042,
      "grad_norm": 5.329368591308594,
      "learning_rate": 9.870830572179668e-05,
      "loss": 0.2653,
      "step": 1886
    },
    {
      "epoch": 0.07629868488885744,
      "grad_norm": 4.881958484649658,
      "learning_rate": 9.870686511984051e-05,
      "loss": 0.0832,
      "step": 1887
    },
    {
      "epoch": 0.07633911874412445,
      "grad_norm": 3.633270025253296,
      "learning_rate": 9.870542372551835e-05,
      "loss": 0.0816,
      "step": 1888
    },
    {
      "epoch": 0.07637955259939147,
      "grad_norm": 8.26182746887207,
      "learning_rate": 9.870398153885363e-05,
      "loss": 0.2355,
      "step": 1889
    },
    {
      "epoch": 0.07641998645465849,
      "grad_norm": 9.037226676940918,
      "learning_rate": 9.870253855986981e-05,
      "loss": 0.2217,
      "step": 1890
    },
    {
      "epoch": 0.0764604203099255,
      "grad_norm": 5.426405906677246,
      "learning_rate": 9.870109478859038e-05,
      "loss": 0.3152,
      "step": 1891
    },
    {
      "epoch": 0.07650085416519252,
      "grad_norm": 6.070871353149414,
      "learning_rate": 9.86996502250388e-05,
      "loss": 0.1766,
      "step": 1892
    },
    {
      "epoch": 0.07654128802045954,
      "grad_norm": 13.136993408203125,
      "learning_rate": 9.86982048692386e-05,
      "loss": 0.296,
      "step": 1893
    },
    {
      "epoch": 0.07658172187572655,
      "grad_norm": 6.958423614501953,
      "learning_rate": 9.869675872121328e-05,
      "loss": 0.1987,
      "step": 1894
    },
    {
      "epoch": 0.07662215573099355,
      "grad_norm": 5.5860419273376465,
      "learning_rate": 9.869531178098637e-05,
      "loss": 0.1481,
      "step": 1895
    },
    {
      "epoch": 0.07666258958626057,
      "grad_norm": 4.680636882781982,
      "learning_rate": 9.86938640485814e-05,
      "loss": 0.1864,
      "step": 1896
    },
    {
      "epoch": 0.07670302344152759,
      "grad_norm": 5.774404525756836,
      "learning_rate": 9.869241552402193e-05,
      "loss": 0.1436,
      "step": 1897
    },
    {
      "epoch": 0.0767434572967946,
      "grad_norm": 5.695173740386963,
      "learning_rate": 9.869096620733152e-05,
      "loss": 0.1373,
      "step": 1898
    },
    {
      "epoch": 0.07678389115206162,
      "grad_norm": 4.333402633666992,
      "learning_rate": 9.868951609853374e-05,
      "loss": 0.1955,
      "step": 1899
    },
    {
      "epoch": 0.07682432500732864,
      "grad_norm": 7.468690395355225,
      "learning_rate": 9.86880651976522e-05,
      "loss": 0.2164,
      "step": 1900
    },
    {
      "epoch": 0.07686475886259565,
      "grad_norm": 11.378255844116211,
      "learning_rate": 9.868661350471049e-05,
      "loss": 0.2477,
      "step": 1901
    },
    {
      "epoch": 0.07690519271786267,
      "grad_norm": 13.516058921813965,
      "learning_rate": 9.868516101973224e-05,
      "loss": 0.3947,
      "step": 1902
    },
    {
      "epoch": 0.07694562657312969,
      "grad_norm": 12.219511032104492,
      "learning_rate": 9.868370774274104e-05,
      "loss": 0.2722,
      "step": 1903
    },
    {
      "epoch": 0.0769860604283967,
      "grad_norm": 6.368983745574951,
      "learning_rate": 9.868225367376059e-05,
      "loss": 0.2531,
      "step": 1904
    },
    {
      "epoch": 0.07702649428366372,
      "grad_norm": 6.1280035972595215,
      "learning_rate": 9.86807988128145e-05,
      "loss": 0.2378,
      "step": 1905
    },
    {
      "epoch": 0.07706692813893072,
      "grad_norm": 11.431299209594727,
      "learning_rate": 9.867934315992647e-05,
      "loss": 0.3338,
      "step": 1906
    },
    {
      "epoch": 0.07710736199419774,
      "grad_norm": 4.489409446716309,
      "learning_rate": 9.867788671512014e-05,
      "loss": 0.1598,
      "step": 1907
    },
    {
      "epoch": 0.07714779584946475,
      "grad_norm": 5.5228776931762695,
      "learning_rate": 9.867642947841925e-05,
      "loss": 0.1395,
      "step": 1908
    },
    {
      "epoch": 0.07718822970473177,
      "grad_norm": 6.186862945556641,
      "learning_rate": 9.867497144984747e-05,
      "loss": 0.2272,
      "step": 1909
    },
    {
      "epoch": 0.07722866355999879,
      "grad_norm": 7.870133876800537,
      "learning_rate": 9.867351262942853e-05,
      "loss": 0.253,
      "step": 1910
    },
    {
      "epoch": 0.0772690974152658,
      "grad_norm": 7.376178741455078,
      "learning_rate": 9.867205301718616e-05,
      "loss": 0.2204,
      "step": 1911
    },
    {
      "epoch": 0.07730953127053282,
      "grad_norm": 3.548290967941284,
      "learning_rate": 9.867059261314413e-05,
      "loss": 0.1417,
      "step": 1912
    },
    {
      "epoch": 0.07734996512579984,
      "grad_norm": 3.480271816253662,
      "learning_rate": 9.866913141732617e-05,
      "loss": 0.1143,
      "step": 1913
    },
    {
      "epoch": 0.07739039898106685,
      "grad_norm": 3.5696704387664795,
      "learning_rate": 9.866766942975606e-05,
      "loss": 0.1275,
      "step": 1914
    },
    {
      "epoch": 0.07743083283633387,
      "grad_norm": 12.819522857666016,
      "learning_rate": 9.866620665045758e-05,
      "loss": 0.3233,
      "step": 1915
    },
    {
      "epoch": 0.07747126669160087,
      "grad_norm": 2.7526416778564453,
      "learning_rate": 9.866474307945452e-05,
      "loss": 0.1667,
      "step": 1916
    },
    {
      "epoch": 0.07751170054686789,
      "grad_norm": 3.3243532180786133,
      "learning_rate": 9.866327871677072e-05,
      "loss": 0.0851,
      "step": 1917
    },
    {
      "epoch": 0.0775521344021349,
      "grad_norm": 4.425264358520508,
      "learning_rate": 9.866181356242996e-05,
      "loss": 0.2794,
      "step": 1918
    },
    {
      "epoch": 0.07759256825740192,
      "grad_norm": 4.070652008056641,
      "learning_rate": 9.866034761645611e-05,
      "loss": 0.2194,
      "step": 1919
    },
    {
      "epoch": 0.07763300211266894,
      "grad_norm": 4.414148807525635,
      "learning_rate": 9.865888087887301e-05,
      "loss": 0.1904,
      "step": 1920
    },
    {
      "epoch": 0.07767343596793595,
      "grad_norm": 6.562784671783447,
      "learning_rate": 9.86574133497045e-05,
      "loss": 0.2115,
      "step": 1921
    },
    {
      "epoch": 0.07771386982320297,
      "grad_norm": 4.1848039627075195,
      "learning_rate": 9.865594502897448e-05,
      "loss": 0.1739,
      "step": 1922
    },
    {
      "epoch": 0.07775430367846999,
      "grad_norm": 7.314671039581299,
      "learning_rate": 9.865447591670681e-05,
      "loss": 0.1581,
      "step": 1923
    },
    {
      "epoch": 0.077794737533737,
      "grad_norm": 3.4102509021759033,
      "learning_rate": 9.865300601292542e-05,
      "loss": 0.1536,
      "step": 1924
    },
    {
      "epoch": 0.07783517138900402,
      "grad_norm": 3.35837721824646,
      "learning_rate": 9.86515353176542e-05,
      "loss": 0.1575,
      "step": 1925
    },
    {
      "epoch": 0.07787560524427103,
      "grad_norm": 6.699120044708252,
      "learning_rate": 9.865006383091709e-05,
      "loss": 0.2858,
      "step": 1926
    },
    {
      "epoch": 0.07791603909953804,
      "grad_norm": 5.780545234680176,
      "learning_rate": 9.864859155273802e-05,
      "loss": 0.2634,
      "step": 1927
    },
    {
      "epoch": 0.07795647295480505,
      "grad_norm": 5.3845038414001465,
      "learning_rate": 9.864711848314095e-05,
      "loss": 0.1616,
      "step": 1928
    },
    {
      "epoch": 0.07799690681007207,
      "grad_norm": 6.841928482055664,
      "learning_rate": 9.864564462214982e-05,
      "loss": 0.1643,
      "step": 1929
    },
    {
      "epoch": 0.07803734066533909,
      "grad_norm": 3.033644199371338,
      "learning_rate": 9.864416996978862e-05,
      "loss": 0.1129,
      "step": 1930
    },
    {
      "epoch": 0.0780777745206061,
      "grad_norm": 2.717914342880249,
      "learning_rate": 9.864269452608138e-05,
      "loss": 0.1134,
      "step": 1931
    },
    {
      "epoch": 0.07811820837587312,
      "grad_norm": 7.480720520019531,
      "learning_rate": 9.864121829105202e-05,
      "loss": 0.2096,
      "step": 1932
    },
    {
      "epoch": 0.07815864223114014,
      "grad_norm": 5.181901931762695,
      "learning_rate": 9.863974126472463e-05,
      "loss": 0.2146,
      "step": 1933
    },
    {
      "epoch": 0.07819907608640715,
      "grad_norm": 6.845294952392578,
      "learning_rate": 9.863826344712319e-05,
      "loss": 0.1819,
      "step": 1934
    },
    {
      "epoch": 0.07823950994167417,
      "grad_norm": 9.297314643859863,
      "learning_rate": 9.863678483827177e-05,
      "loss": 0.2413,
      "step": 1935
    },
    {
      "epoch": 0.07827994379694118,
      "grad_norm": 6.202345371246338,
      "learning_rate": 9.863530543819442e-05,
      "loss": 0.2361,
      "step": 1936
    },
    {
      "epoch": 0.0783203776522082,
      "grad_norm": 7.726655006408691,
      "learning_rate": 9.86338252469152e-05,
      "loss": 0.1715,
      "step": 1937
    },
    {
      "epoch": 0.0783608115074752,
      "grad_norm": 6.419961929321289,
      "learning_rate": 9.86323442644582e-05,
      "loss": 0.1953,
      "step": 1938
    },
    {
      "epoch": 0.07840124536274222,
      "grad_norm": 6.476056098937988,
      "learning_rate": 9.863086249084748e-05,
      "loss": 0.2112,
      "step": 1939
    },
    {
      "epoch": 0.07844167921800924,
      "grad_norm": 3.8232007026672363,
      "learning_rate": 9.86293799261072e-05,
      "loss": 0.1095,
      "step": 1940
    },
    {
      "epoch": 0.07848211307327625,
      "grad_norm": 3.960050106048584,
      "learning_rate": 9.862789657026142e-05,
      "loss": 0.2794,
      "step": 1941
    },
    {
      "epoch": 0.07852254692854327,
      "grad_norm": 5.467953205108643,
      "learning_rate": 9.862641242333432e-05,
      "loss": 0.3054,
      "step": 1942
    },
    {
      "epoch": 0.07856298078381029,
      "grad_norm": 3.9826231002807617,
      "learning_rate": 9.862492748535002e-05,
      "loss": 0.1384,
      "step": 1943
    },
    {
      "epoch": 0.0786034146390773,
      "grad_norm": 10.507331848144531,
      "learning_rate": 9.862344175633267e-05,
      "loss": 0.2954,
      "step": 1944
    },
    {
      "epoch": 0.07864384849434432,
      "grad_norm": 3.2512004375457764,
      "learning_rate": 9.862195523630647e-05,
      "loss": 0.1382,
      "step": 1945
    },
    {
      "epoch": 0.07868428234961133,
      "grad_norm": 4.9798383712768555,
      "learning_rate": 9.862046792529557e-05,
      "loss": 0.1402,
      "step": 1946
    },
    {
      "epoch": 0.07872471620487835,
      "grad_norm": 10.973594665527344,
      "learning_rate": 9.861897982332417e-05,
      "loss": 0.2331,
      "step": 1947
    },
    {
      "epoch": 0.07876515006014535,
      "grad_norm": 2.550144910812378,
      "learning_rate": 9.86174909304165e-05,
      "loss": 0.1573,
      "step": 1948
    },
    {
      "epoch": 0.07880558391541237,
      "grad_norm": 6.443003177642822,
      "learning_rate": 9.861600124659679e-05,
      "loss": 0.2377,
      "step": 1949
    },
    {
      "epoch": 0.07884601777067939,
      "grad_norm": 4.986809730529785,
      "learning_rate": 9.861451077188921e-05,
      "loss": 0.1932,
      "step": 1950
    },
    {
      "epoch": 0.0788864516259464,
      "grad_norm": 5.080024242401123,
      "learning_rate": 9.861301950631808e-05,
      "loss": 0.3017,
      "step": 1951
    },
    {
      "epoch": 0.07892688548121342,
      "grad_norm": 9.663859367370605,
      "learning_rate": 9.861152744990762e-05,
      "loss": 0.2299,
      "step": 1952
    },
    {
      "epoch": 0.07896731933648043,
      "grad_norm": 4.211429595947266,
      "learning_rate": 9.861003460268212e-05,
      "loss": 0.3619,
      "step": 1953
    },
    {
      "epoch": 0.07900775319174745,
      "grad_norm": 7.227161407470703,
      "learning_rate": 9.860854096466585e-05,
      "loss": 0.2804,
      "step": 1954
    },
    {
      "epoch": 0.07904818704701447,
      "grad_norm": 4.010702133178711,
      "learning_rate": 9.860704653588312e-05,
      "loss": 0.2369,
      "step": 1955
    },
    {
      "epoch": 0.07908862090228148,
      "grad_norm": 2.6659843921661377,
      "learning_rate": 9.860555131635826e-05,
      "loss": 0.1661,
      "step": 1956
    },
    {
      "epoch": 0.0791290547575485,
      "grad_norm": 1.7327635288238525,
      "learning_rate": 9.860405530611556e-05,
      "loss": 0.1163,
      "step": 1957
    },
    {
      "epoch": 0.07916948861281552,
      "grad_norm": 5.511391639709473,
      "learning_rate": 9.860255850517936e-05,
      "loss": 0.1723,
      "step": 1958
    },
    {
      "epoch": 0.07920992246808252,
      "grad_norm": 11.65819263458252,
      "learning_rate": 9.860106091357404e-05,
      "loss": 0.2132,
      "step": 1959
    },
    {
      "epoch": 0.07925035632334954,
      "grad_norm": 3.8906733989715576,
      "learning_rate": 9.859956253132391e-05,
      "loss": 0.1439,
      "step": 1960
    },
    {
      "epoch": 0.07929079017861655,
      "grad_norm": 4.15351676940918,
      "learning_rate": 9.859806335845342e-05,
      "loss": 0.1234,
      "step": 1961
    },
    {
      "epoch": 0.07933122403388357,
      "grad_norm": 11.174996376037598,
      "learning_rate": 9.85965633949869e-05,
      "loss": 0.3297,
      "step": 1962
    },
    {
      "epoch": 0.07937165788915058,
      "grad_norm": 2.268824338912964,
      "learning_rate": 9.859506264094877e-05,
      "loss": 0.1459,
      "step": 1963
    },
    {
      "epoch": 0.0794120917444176,
      "grad_norm": 7.434964656829834,
      "learning_rate": 9.859356109636345e-05,
      "loss": 0.2018,
      "step": 1964
    },
    {
      "epoch": 0.07945252559968462,
      "grad_norm": 7.46022367477417,
      "learning_rate": 9.859205876125535e-05,
      "loss": 0.2161,
      "step": 1965
    },
    {
      "epoch": 0.07949295945495163,
      "grad_norm": 3.4465866088867188,
      "learning_rate": 9.859055563564892e-05,
      "loss": 0.2464,
      "step": 1966
    },
    {
      "epoch": 0.07953339331021865,
      "grad_norm": 8.266654968261719,
      "learning_rate": 9.858905171956862e-05,
      "loss": 0.1503,
      "step": 1967
    },
    {
      "epoch": 0.07957382716548567,
      "grad_norm": 9.251333236694336,
      "learning_rate": 9.858754701303892e-05,
      "loss": 0.2703,
      "step": 1968
    },
    {
      "epoch": 0.07961426102075268,
      "grad_norm": 5.681820392608643,
      "learning_rate": 9.858604151608427e-05,
      "loss": 0.1131,
      "step": 1969
    },
    {
      "epoch": 0.07965469487601969,
      "grad_norm": 3.017091989517212,
      "learning_rate": 9.858453522872919e-05,
      "loss": 0.1579,
      "step": 1970
    },
    {
      "epoch": 0.0796951287312867,
      "grad_norm": 2.4424352645874023,
      "learning_rate": 9.858302815099818e-05,
      "loss": 0.1477,
      "step": 1971
    },
    {
      "epoch": 0.07973556258655372,
      "grad_norm": 6.790644645690918,
      "learning_rate": 9.858152028291575e-05,
      "loss": 0.1749,
      "step": 1972
    },
    {
      "epoch": 0.07977599644182073,
      "grad_norm": 6.082481861114502,
      "learning_rate": 9.858001162450642e-05,
      "loss": 0.164,
      "step": 1973
    },
    {
      "epoch": 0.07981643029708775,
      "grad_norm": 3.881479263305664,
      "learning_rate": 9.857850217579474e-05,
      "loss": 0.101,
      "step": 1974
    },
    {
      "epoch": 0.07985686415235477,
      "grad_norm": 2.3473894596099854,
      "learning_rate": 9.857699193680529e-05,
      "loss": 0.1324,
      "step": 1975
    },
    {
      "epoch": 0.07989729800762178,
      "grad_norm": 4.265580177307129,
      "learning_rate": 9.857548090756262e-05,
      "loss": 0.1471,
      "step": 1976
    },
    {
      "epoch": 0.0799377318628888,
      "grad_norm": 7.518443584442139,
      "learning_rate": 9.857396908809129e-05,
      "loss": 0.2019,
      "step": 1977
    },
    {
      "epoch": 0.07997816571815582,
      "grad_norm": 9.019708633422852,
      "learning_rate": 9.857245647841593e-05,
      "loss": 0.2607,
      "step": 1978
    },
    {
      "epoch": 0.08001859957342283,
      "grad_norm": 12.568745613098145,
      "learning_rate": 9.857094307856113e-05,
      "loss": 0.2192,
      "step": 1979
    },
    {
      "epoch": 0.08005903342868984,
      "grad_norm": 4.3192338943481445,
      "learning_rate": 9.856942888855152e-05,
      "loss": 0.1378,
      "step": 1980
    },
    {
      "epoch": 0.08009946728395685,
      "grad_norm": 3.3349099159240723,
      "learning_rate": 9.856791390841171e-05,
      "loss": 0.0956,
      "step": 1981
    },
    {
      "epoch": 0.08013990113922387,
      "grad_norm": 7.628266334533691,
      "learning_rate": 9.856639813816637e-05,
      "loss": 0.1767,
      "step": 1982
    },
    {
      "epoch": 0.08018033499449088,
      "grad_norm": 6.676673889160156,
      "learning_rate": 9.856488157784013e-05,
      "loss": 0.1805,
      "step": 1983
    },
    {
      "epoch": 0.0802207688497579,
      "grad_norm": 7.444279193878174,
      "learning_rate": 9.856336422745771e-05,
      "loss": 0.2732,
      "step": 1984
    },
    {
      "epoch": 0.08026120270502492,
      "grad_norm": 4.757753372192383,
      "learning_rate": 9.856184608704375e-05,
      "loss": 0.133,
      "step": 1985
    },
    {
      "epoch": 0.08030163656029193,
      "grad_norm": 8.14673900604248,
      "learning_rate": 9.856032715662298e-05,
      "loss": 0.2439,
      "step": 1986
    },
    {
      "epoch": 0.08034207041555895,
      "grad_norm": 10.812795639038086,
      "learning_rate": 9.855880743622009e-05,
      "loss": 0.3277,
      "step": 1987
    },
    {
      "epoch": 0.08038250427082597,
      "grad_norm": 2.711475133895874,
      "learning_rate": 9.855728692585978e-05,
      "loss": 0.0681,
      "step": 1988
    },
    {
      "epoch": 0.08042293812609298,
      "grad_norm": 6.004952430725098,
      "learning_rate": 9.855576562556683e-05,
      "loss": 0.2107,
      "step": 1989
    },
    {
      "epoch": 0.08046337198136,
      "grad_norm": 6.3270087242126465,
      "learning_rate": 9.855424353536596e-05,
      "loss": 0.2065,
      "step": 1990
    },
    {
      "epoch": 0.080503805836627,
      "grad_norm": 6.491931438446045,
      "learning_rate": 9.855272065528196e-05,
      "loss": 0.2259,
      "step": 1991
    },
    {
      "epoch": 0.08054423969189402,
      "grad_norm": 4.481473922729492,
      "learning_rate": 9.855119698533957e-05,
      "loss": 0.1823,
      "step": 1992
    },
    {
      "epoch": 0.08058467354716103,
      "grad_norm": 6.603989124298096,
      "learning_rate": 9.85496725255636e-05,
      "loss": 0.2123,
      "step": 1993
    },
    {
      "epoch": 0.08062510740242805,
      "grad_norm": 3.2952308654785156,
      "learning_rate": 9.854814727597883e-05,
      "loss": 0.308,
      "step": 1994
    },
    {
      "epoch": 0.08066554125769507,
      "grad_norm": 5.813769340515137,
      "learning_rate": 9.854662123661009e-05,
      "loss": 0.1056,
      "step": 1995
    },
    {
      "epoch": 0.08070597511296208,
      "grad_norm": 5.937270164489746,
      "learning_rate": 9.854509440748221e-05,
      "loss": 0.1551,
      "step": 1996
    },
    {
      "epoch": 0.0807464089682291,
      "grad_norm": 2.959695339202881,
      "learning_rate": 9.854356678862002e-05,
      "loss": 0.1908,
      "step": 1997
    },
    {
      "epoch": 0.08078684282349612,
      "grad_norm": 2.71689772605896,
      "learning_rate": 9.854203838004837e-05,
      "loss": 0.0846,
      "step": 1998
    },
    {
      "epoch": 0.08082727667876313,
      "grad_norm": 4.306998252868652,
      "learning_rate": 9.854050918179212e-05,
      "loss": 0.1374,
      "step": 1999
    },
    {
      "epoch": 0.08086771053403015,
      "grad_norm": 6.189534664154053,
      "learning_rate": 9.853897919387615e-05,
      "loss": 0.2276,
      "step": 2000
    },
    {
      "epoch": 0.08090814438929717,
      "grad_norm": 10.162153244018555,
      "learning_rate": 9.853744841632537e-05,
      "loss": 0.2569,
      "step": 2001
    },
    {
      "epoch": 0.08094857824456417,
      "grad_norm": 5.484414100646973,
      "learning_rate": 9.853591684916465e-05,
      "loss": 0.1386,
      "step": 2002
    },
    {
      "epoch": 0.08098901209983118,
      "grad_norm": 2.9369618892669678,
      "learning_rate": 9.853438449241893e-05,
      "loss": 0.1093,
      "step": 2003
    },
    {
      "epoch": 0.0810294459550982,
      "grad_norm": 5.589948654174805,
      "learning_rate": 9.853285134611312e-05,
      "loss": 0.192,
      "step": 2004
    },
    {
      "epoch": 0.08106987981036522,
      "grad_norm": 2.7465031147003174,
      "learning_rate": 9.853131741027219e-05,
      "loss": 0.1386,
      "step": 2005
    },
    {
      "epoch": 0.08111031366563223,
      "grad_norm": 1.8206450939178467,
      "learning_rate": 9.852978268492106e-05,
      "loss": 0.1033,
      "step": 2006
    },
    {
      "epoch": 0.08115074752089925,
      "grad_norm": 6.228503704071045,
      "learning_rate": 9.852824717008471e-05,
      "loss": 0.2123,
      "step": 2007
    },
    {
      "epoch": 0.08119118137616627,
      "grad_norm": 5.479308128356934,
      "learning_rate": 9.852671086578812e-05,
      "loss": 0.3059,
      "step": 2008
    },
    {
      "epoch": 0.08123161523143328,
      "grad_norm": 3.001014232635498,
      "learning_rate": 9.852517377205628e-05,
      "loss": 0.1584,
      "step": 2009
    },
    {
      "epoch": 0.0812720490867003,
      "grad_norm": 3.6984784603118896,
      "learning_rate": 9.852363588891423e-05,
      "loss": 0.1869,
      "step": 2010
    },
    {
      "epoch": 0.08131248294196732,
      "grad_norm": 9.835447311401367,
      "learning_rate": 9.852209721638692e-05,
      "loss": 0.3244,
      "step": 2011
    },
    {
      "epoch": 0.08135291679723432,
      "grad_norm": 4.0184645652771,
      "learning_rate": 9.852055775449945e-05,
      "loss": 0.1012,
      "step": 2012
    },
    {
      "epoch": 0.08139335065250133,
      "grad_norm": 2.8352630138397217,
      "learning_rate": 9.85190175032768e-05,
      "loss": 0.1988,
      "step": 2013
    },
    {
      "epoch": 0.08143378450776835,
      "grad_norm": 11.761627197265625,
      "learning_rate": 9.85174764627441e-05,
      "loss": 0.3633,
      "step": 2014
    },
    {
      "epoch": 0.08147421836303537,
      "grad_norm": 4.718667507171631,
      "learning_rate": 9.851593463292635e-05,
      "loss": 0.159,
      "step": 2015
    },
    {
      "epoch": 0.08151465221830238,
      "grad_norm": 5.085806846618652,
      "learning_rate": 9.851439201384868e-05,
      "loss": 0.3949,
      "step": 2016
    },
    {
      "epoch": 0.0815550860735694,
      "grad_norm": 6.529993534088135,
      "learning_rate": 9.851284860553615e-05,
      "loss": 0.2185,
      "step": 2017
    },
    {
      "epoch": 0.08159551992883642,
      "grad_norm": 4.843364238739014,
      "learning_rate": 9.851130440801391e-05,
      "loss": 0.1629,
      "step": 2018
    },
    {
      "epoch": 0.08163595378410343,
      "grad_norm": 5.056021690368652,
      "learning_rate": 9.850975942130705e-05,
      "loss": 0.265,
      "step": 2019
    },
    {
      "epoch": 0.08167638763937045,
      "grad_norm": 4.813655376434326,
      "learning_rate": 9.85082136454407e-05,
      "loss": 0.2045,
      "step": 2020
    },
    {
      "epoch": 0.08171682149463746,
      "grad_norm": 4.050836086273193,
      "learning_rate": 9.850666708044004e-05,
      "loss": 0.1171,
      "step": 2021
    },
    {
      "epoch": 0.08175725534990448,
      "grad_norm": 4.60107946395874,
      "learning_rate": 9.850511972633019e-05,
      "loss": 0.2661,
      "step": 2022
    },
    {
      "epoch": 0.08179768920517148,
      "grad_norm": 6.356388092041016,
      "learning_rate": 9.850357158313633e-05,
      "loss": 0.229,
      "step": 2023
    },
    {
      "epoch": 0.0818381230604385,
      "grad_norm": 6.965513229370117,
      "learning_rate": 9.850202265088368e-05,
      "loss": 0.2007,
      "step": 2024
    },
    {
      "epoch": 0.08187855691570552,
      "grad_norm": 3.989459753036499,
      "learning_rate": 9.85004729295974e-05,
      "loss": 0.1776,
      "step": 2025
    },
    {
      "epoch": 0.08191899077097253,
      "grad_norm": 1.5606330633163452,
      "learning_rate": 9.849892241930272e-05,
      "loss": 0.1012,
      "step": 2026
    },
    {
      "epoch": 0.08195942462623955,
      "grad_norm": 4.602168560028076,
      "learning_rate": 9.849737112002485e-05,
      "loss": 0.1744,
      "step": 2027
    },
    {
      "epoch": 0.08199985848150657,
      "grad_norm": 8.503775596618652,
      "learning_rate": 9.849581903178904e-05,
      "loss": 0.2269,
      "step": 2028
    },
    {
      "epoch": 0.08204029233677358,
      "grad_norm": 2.850799798965454,
      "learning_rate": 9.849426615462054e-05,
      "loss": 0.1284,
      "step": 2029
    },
    {
      "epoch": 0.0820807261920406,
      "grad_norm": 3.4044389724731445,
      "learning_rate": 9.849271248854461e-05,
      "loss": 0.1002,
      "step": 2030
    },
    {
      "epoch": 0.08212116004730761,
      "grad_norm": 4.26686954498291,
      "learning_rate": 9.84911580335865e-05,
      "loss": 0.2273,
      "step": 2031
    },
    {
      "epoch": 0.08216159390257463,
      "grad_norm": 5.36971378326416,
      "learning_rate": 9.848960278977155e-05,
      "loss": 0.1471,
      "step": 2032
    },
    {
      "epoch": 0.08220202775784165,
      "grad_norm": 6.537599563598633,
      "learning_rate": 9.848804675712501e-05,
      "loss": 0.1415,
      "step": 2033
    },
    {
      "epoch": 0.08224246161310865,
      "grad_norm": 2.9048447608947754,
      "learning_rate": 9.848648993567222e-05,
      "loss": 0.1388,
      "step": 2034
    },
    {
      "epoch": 0.08228289546837567,
      "grad_norm": 6.558468818664551,
      "learning_rate": 9.848493232543849e-05,
      "loss": 0.2851,
      "step": 2035
    },
    {
      "epoch": 0.08232332932364268,
      "grad_norm": 8.457770347595215,
      "learning_rate": 9.84833739264492e-05,
      "loss": 0.1355,
      "step": 2036
    },
    {
      "epoch": 0.0823637631789097,
      "grad_norm": 8.950994491577148,
      "learning_rate": 9.848181473872964e-05,
      "loss": 0.1079,
      "step": 2037
    },
    {
      "epoch": 0.08240419703417672,
      "grad_norm": 6.763830661773682,
      "learning_rate": 9.848025476230522e-05,
      "loss": 0.2789,
      "step": 2038
    },
    {
      "epoch": 0.08244463088944373,
      "grad_norm": 5.5244035720825195,
      "learning_rate": 9.847869399720131e-05,
      "loss": 0.2907,
      "step": 2039
    },
    {
      "epoch": 0.08248506474471075,
      "grad_norm": 3.8820509910583496,
      "learning_rate": 9.847713244344328e-05,
      "loss": 0.105,
      "step": 2040
    },
    {
      "epoch": 0.08252549859997776,
      "grad_norm": 5.047769069671631,
      "learning_rate": 9.847557010105657e-05,
      "loss": 0.2156,
      "step": 2041
    },
    {
      "epoch": 0.08256593245524478,
      "grad_norm": 5.500354290008545,
      "learning_rate": 9.847400697006657e-05,
      "loss": 0.4111,
      "step": 2042
    },
    {
      "epoch": 0.0826063663105118,
      "grad_norm": 4.490057468414307,
      "learning_rate": 9.84724430504987e-05,
      "loss": 0.1629,
      "step": 2043
    },
    {
      "epoch": 0.0826468001657788,
      "grad_norm": 3.6551764011383057,
      "learning_rate": 9.847087834237843e-05,
      "loss": 0.1735,
      "step": 2044
    },
    {
      "epoch": 0.08268723402104582,
      "grad_norm": 5.812078475952148,
      "learning_rate": 9.84693128457312e-05,
      "loss": 0.2341,
      "step": 2045
    },
    {
      "epoch": 0.08272766787631283,
      "grad_norm": 10.099764823913574,
      "learning_rate": 9.846774656058247e-05,
      "loss": 0.3812,
      "step": 2046
    },
    {
      "epoch": 0.08276810173157985,
      "grad_norm": 4.545963764190674,
      "learning_rate": 9.846617948695773e-05,
      "loss": 0.2365,
      "step": 2047
    },
    {
      "epoch": 0.08280853558684687,
      "grad_norm": 2.7030625343322754,
      "learning_rate": 9.846461162488248e-05,
      "loss": 0.106,
      "step": 2048
    },
    {
      "epoch": 0.08284896944211388,
      "grad_norm": 7.548708915710449,
      "learning_rate": 9.846304297438221e-05,
      "loss": 0.2142,
      "step": 2049
    },
    {
      "epoch": 0.0828894032973809,
      "grad_norm": 3.514312744140625,
      "learning_rate": 9.846147353548245e-05,
      "loss": 0.1556,
      "step": 2050
    },
    {
      "epoch": 0.08292983715264791,
      "grad_norm": 11.31811237335205,
      "learning_rate": 9.845990330820874e-05,
      "loss": 0.244,
      "step": 2051
    },
    {
      "epoch": 0.08297027100791493,
      "grad_norm": 6.01167106628418,
      "learning_rate": 9.84583322925866e-05,
      "loss": 0.2436,
      "step": 2052
    },
    {
      "epoch": 0.08301070486318195,
      "grad_norm": 9.018636703491211,
      "learning_rate": 9.84567604886416e-05,
      "loss": 0.1915,
      "step": 2053
    },
    {
      "epoch": 0.08305113871844896,
      "grad_norm": 10.899344444274902,
      "learning_rate": 9.845518789639932e-05,
      "loss": 0.2765,
      "step": 2054
    },
    {
      "epoch": 0.08309157257371597,
      "grad_norm": 6.704720497131348,
      "learning_rate": 9.845361451588532e-05,
      "loss": 0.2176,
      "step": 2055
    },
    {
      "epoch": 0.08313200642898298,
      "grad_norm": 4.200408935546875,
      "learning_rate": 9.845204034712523e-05,
      "loss": 0.2695,
      "step": 2056
    },
    {
      "epoch": 0.08317244028425,
      "grad_norm": 4.209378719329834,
      "learning_rate": 9.845046539014463e-05,
      "loss": 0.1212,
      "step": 2057
    },
    {
      "epoch": 0.08321287413951702,
      "grad_norm": 3.029438018798828,
      "learning_rate": 9.844888964496915e-05,
      "loss": 0.2587,
      "step": 2058
    },
    {
      "epoch": 0.08325330799478403,
      "grad_norm": 2.959108352661133,
      "learning_rate": 9.844731311162443e-05,
      "loss": 0.1284,
      "step": 2059
    },
    {
      "epoch": 0.08329374185005105,
      "grad_norm": 4.754903316497803,
      "learning_rate": 9.844573579013609e-05,
      "loss": 0.2288,
      "step": 2060
    },
    {
      "epoch": 0.08333417570531806,
      "grad_norm": 6.476150989532471,
      "learning_rate": 9.844415768052984e-05,
      "loss": 0.2936,
      "step": 2061
    },
    {
      "epoch": 0.08337460956058508,
      "grad_norm": 10.344002723693848,
      "learning_rate": 9.84425787828313e-05,
      "loss": 0.3315,
      "step": 2062
    },
    {
      "epoch": 0.0834150434158521,
      "grad_norm": 2.400029420852661,
      "learning_rate": 9.84409990970662e-05,
      "loss": 0.109,
      "step": 2063
    },
    {
      "epoch": 0.08345547727111911,
      "grad_norm": 7.955074787139893,
      "learning_rate": 9.84394186232602e-05,
      "loss": 0.3681,
      "step": 2064
    },
    {
      "epoch": 0.08349591112638613,
      "grad_norm": 3.293494462966919,
      "learning_rate": 9.843783736143905e-05,
      "loss": 0.1243,
      "step": 2065
    },
    {
      "epoch": 0.08353634498165313,
      "grad_norm": 4.036622524261475,
      "learning_rate": 9.843625531162844e-05,
      "loss": 0.2418,
      "step": 2066
    },
    {
      "epoch": 0.08357677883692015,
      "grad_norm": 3.4086384773254395,
      "learning_rate": 9.843467247385412e-05,
      "loss": 0.1051,
      "step": 2067
    },
    {
      "epoch": 0.08361721269218717,
      "grad_norm": 4.609098434448242,
      "learning_rate": 9.843308884814184e-05,
      "loss": 0.2261,
      "step": 2068
    },
    {
      "epoch": 0.08365764654745418,
      "grad_norm": 5.242365837097168,
      "learning_rate": 9.843150443451737e-05,
      "loss": 0.1879,
      "step": 2069
    },
    {
      "epoch": 0.0836980804027212,
      "grad_norm": 3.9991543292999268,
      "learning_rate": 9.842991923300649e-05,
      "loss": 0.1327,
      "step": 2070
    },
    {
      "epoch": 0.08373851425798821,
      "grad_norm": 2.914424180984497,
      "learning_rate": 9.842833324363494e-05,
      "loss": 0.2872,
      "step": 2071
    },
    {
      "epoch": 0.08377894811325523,
      "grad_norm": 6.414231777191162,
      "learning_rate": 9.842674646642858e-05,
      "loss": 0.1712,
      "step": 2072
    },
    {
      "epoch": 0.08381938196852225,
      "grad_norm": 8.35725212097168,
      "learning_rate": 9.84251589014132e-05,
      "loss": 0.3129,
      "step": 2073
    },
    {
      "epoch": 0.08385981582378926,
      "grad_norm": 4.311469078063965,
      "learning_rate": 9.842357054861462e-05,
      "loss": 0.1792,
      "step": 2074
    },
    {
      "epoch": 0.08390024967905628,
      "grad_norm": 2.361933708190918,
      "learning_rate": 9.84219814080587e-05,
      "loss": 0.1658,
      "step": 2075
    },
    {
      "epoch": 0.08394068353432328,
      "grad_norm": 4.285914421081543,
      "learning_rate": 9.842039147977127e-05,
      "loss": 0.1887,
      "step": 2076
    },
    {
      "epoch": 0.0839811173895903,
      "grad_norm": 3.146038293838501,
      "learning_rate": 9.841880076377821e-05,
      "loss": 0.1399,
      "step": 2077
    },
    {
      "epoch": 0.08402155124485731,
      "grad_norm": 5.117367744445801,
      "learning_rate": 9.841720926010538e-05,
      "loss": 0.173,
      "step": 2078
    },
    {
      "epoch": 0.08406198510012433,
      "grad_norm": 5.328029632568359,
      "learning_rate": 9.84156169687787e-05,
      "loss": 0.1313,
      "step": 2079
    },
    {
      "epoch": 0.08410241895539135,
      "grad_norm": 2.0733063220977783,
      "learning_rate": 9.841402388982403e-05,
      "loss": 0.0653,
      "step": 2080
    },
    {
      "epoch": 0.08414285281065836,
      "grad_norm": 12.627921104431152,
      "learning_rate": 9.841243002326734e-05,
      "loss": 0.3155,
      "step": 2081
    },
    {
      "epoch": 0.08418328666592538,
      "grad_norm": 5.5360941886901855,
      "learning_rate": 9.841083536913452e-05,
      "loss": 0.169,
      "step": 2082
    },
    {
      "epoch": 0.0842237205211924,
      "grad_norm": 16.499582290649414,
      "learning_rate": 9.840923992745152e-05,
      "loss": 0.3491,
      "step": 2083
    },
    {
      "epoch": 0.08426415437645941,
      "grad_norm": 6.222485065460205,
      "learning_rate": 9.84076436982443e-05,
      "loss": 0.1752,
      "step": 2084
    },
    {
      "epoch": 0.08430458823172643,
      "grad_norm": 5.30979061126709,
      "learning_rate": 9.840604668153884e-05,
      "loss": 0.1599,
      "step": 2085
    },
    {
      "epoch": 0.08434502208699345,
      "grad_norm": 6.1610212326049805,
      "learning_rate": 9.840444887736109e-05,
      "loss": 0.2413,
      "step": 2086
    },
    {
      "epoch": 0.08438545594226045,
      "grad_norm": 2.9072132110595703,
      "learning_rate": 9.840285028573705e-05,
      "loss": 0.2525,
      "step": 2087
    },
    {
      "epoch": 0.08442588979752746,
      "grad_norm": 8.364328384399414,
      "learning_rate": 9.840125090669277e-05,
      "loss": 0.2367,
      "step": 2088
    },
    {
      "epoch": 0.08446632365279448,
      "grad_norm": 7.627289772033691,
      "learning_rate": 9.839965074025421e-05,
      "loss": 0.173,
      "step": 2089
    },
    {
      "epoch": 0.0845067575080615,
      "grad_norm": 6.73826789855957,
      "learning_rate": 9.839804978644743e-05,
      "loss": 0.1763,
      "step": 2090
    },
    {
      "epoch": 0.08454719136332851,
      "grad_norm": 7.696340084075928,
      "learning_rate": 9.839644804529846e-05,
      "loss": 0.1736,
      "step": 2091
    },
    {
      "epoch": 0.08458762521859553,
      "grad_norm": 7.93285608291626,
      "learning_rate": 9.839484551683338e-05,
      "loss": 0.1971,
      "step": 2092
    },
    {
      "epoch": 0.08462805907386255,
      "grad_norm": 4.4259419441223145,
      "learning_rate": 9.839324220107824e-05,
      "loss": 0.12,
      "step": 2093
    },
    {
      "epoch": 0.08466849292912956,
      "grad_norm": 6.055801868438721,
      "learning_rate": 9.839163809805913e-05,
      "loss": 0.1357,
      "step": 2094
    },
    {
      "epoch": 0.08470892678439658,
      "grad_norm": 4.739110469818115,
      "learning_rate": 9.839003320780216e-05,
      "loss": 0.1841,
      "step": 2095
    },
    {
      "epoch": 0.0847493606396636,
      "grad_norm": 7.680644989013672,
      "learning_rate": 9.838842753033341e-05,
      "loss": 0.2986,
      "step": 2096
    },
    {
      "epoch": 0.08478979449493061,
      "grad_norm": 4.097894668579102,
      "learning_rate": 9.838682106567902e-05,
      "loss": 0.1163,
      "step": 2097
    },
    {
      "epoch": 0.08483022835019761,
      "grad_norm": 6.28085470199585,
      "learning_rate": 9.838521381386512e-05,
      "loss": 0.147,
      "step": 2098
    },
    {
      "epoch": 0.08487066220546463,
      "grad_norm": 6.6080322265625,
      "learning_rate": 9.838360577491785e-05,
      "loss": 0.2138,
      "step": 2099
    },
    {
      "epoch": 0.08491109606073165,
      "grad_norm": 7.368183135986328,
      "learning_rate": 9.838199694886338e-05,
      "loss": 0.2849,
      "step": 2100
    },
    {
      "epoch": 0.08495152991599866,
      "grad_norm": 10.304824829101562,
      "learning_rate": 9.838038733572788e-05,
      "loss": 0.2754,
      "step": 2101
    },
    {
      "epoch": 0.08499196377126568,
      "grad_norm": 5.624362468719482,
      "learning_rate": 9.837877693553752e-05,
      "loss": 0.1626,
      "step": 2102
    },
    {
      "epoch": 0.0850323976265327,
      "grad_norm": 4.520270824432373,
      "learning_rate": 9.837716574831854e-05,
      "loss": 0.219,
      "step": 2103
    },
    {
      "epoch": 0.08507283148179971,
      "grad_norm": 5.025966644287109,
      "learning_rate": 9.83755537740971e-05,
      "loss": 0.0983,
      "step": 2104
    },
    {
      "epoch": 0.08511326533706673,
      "grad_norm": 7.482702255249023,
      "learning_rate": 9.837394101289945e-05,
      "loss": 0.2203,
      "step": 2105
    },
    {
      "epoch": 0.08515369919233375,
      "grad_norm": 4.165750503540039,
      "learning_rate": 9.837232746475183e-05,
      "loss": 0.1965,
      "step": 2106
    },
    {
      "epoch": 0.08519413304760076,
      "grad_norm": 7.535092353820801,
      "learning_rate": 9.837071312968049e-05,
      "loss": 0.2081,
      "step": 2107
    },
    {
      "epoch": 0.08523456690286776,
      "grad_norm": 6.105241775512695,
      "learning_rate": 9.836909800771167e-05,
      "loss": 0.2289,
      "step": 2108
    },
    {
      "epoch": 0.08527500075813478,
      "grad_norm": 6.0043044090271,
      "learning_rate": 9.836748209887165e-05,
      "loss": 0.1268,
      "step": 2109
    },
    {
      "epoch": 0.0853154346134018,
      "grad_norm": 5.218935966491699,
      "learning_rate": 9.836586540318676e-05,
      "loss": 0.1676,
      "step": 2110
    },
    {
      "epoch": 0.08535586846866881,
      "grad_norm": 4.747800827026367,
      "learning_rate": 9.836424792068323e-05,
      "loss": 0.2023,
      "step": 2111
    },
    {
      "epoch": 0.08539630232393583,
      "grad_norm": 2.5046958923339844,
      "learning_rate": 9.836262965138744e-05,
      "loss": 0.1079,
      "step": 2112
    },
    {
      "epoch": 0.08543673617920285,
      "grad_norm": 7.754920482635498,
      "learning_rate": 9.836101059532567e-05,
      "loss": 0.2388,
      "step": 2113
    },
    {
      "epoch": 0.08547717003446986,
      "grad_norm": 9.608997344970703,
      "learning_rate": 9.835939075252429e-05,
      "loss": 0.2688,
      "step": 2114
    },
    {
      "epoch": 0.08551760388973688,
      "grad_norm": 6.939508438110352,
      "learning_rate": 9.835777012300961e-05,
      "loss": 0.2353,
      "step": 2115
    },
    {
      "epoch": 0.0855580377450039,
      "grad_norm": 8.243762016296387,
      "learning_rate": 9.835614870680805e-05,
      "loss": 0.2265,
      "step": 2116
    },
    {
      "epoch": 0.08559847160027091,
      "grad_norm": 4.950127601623535,
      "learning_rate": 9.835452650394594e-05,
      "loss": 0.254,
      "step": 2117
    },
    {
      "epoch": 0.08563890545553793,
      "grad_norm": 7.1861419677734375,
      "learning_rate": 9.83529035144497e-05,
      "loss": 0.1687,
      "step": 2118
    },
    {
      "epoch": 0.08567933931080493,
      "grad_norm": 2.489009380340576,
      "learning_rate": 9.835127973834572e-05,
      "loss": 0.2136,
      "step": 2119
    },
    {
      "epoch": 0.08571977316607195,
      "grad_norm": 3.271214723587036,
      "learning_rate": 9.834965517566043e-05,
      "loss": 0.1652,
      "step": 2120
    },
    {
      "epoch": 0.08576020702133896,
      "grad_norm": 4.393527030944824,
      "learning_rate": 9.834802982642024e-05,
      "loss": 0.1565,
      "step": 2121
    },
    {
      "epoch": 0.08580064087660598,
      "grad_norm": 4.090995788574219,
      "learning_rate": 9.834640369065158e-05,
      "loss": 0.1726,
      "step": 2122
    },
    {
      "epoch": 0.085841074731873,
      "grad_norm": 5.630694389343262,
      "learning_rate": 9.834477676838095e-05,
      "loss": 0.1245,
      "step": 2123
    },
    {
      "epoch": 0.08588150858714001,
      "grad_norm": 7.705469131469727,
      "learning_rate": 9.834314905963476e-05,
      "loss": 0.1349,
      "step": 2124
    },
    {
      "epoch": 0.08592194244240703,
      "grad_norm": 3.759338140487671,
      "learning_rate": 9.834152056443952e-05,
      "loss": 0.0986,
      "step": 2125
    },
    {
      "epoch": 0.08596237629767405,
      "grad_norm": 5.309838771820068,
      "learning_rate": 9.833989128282175e-05,
      "loss": 0.1837,
      "step": 2126
    },
    {
      "epoch": 0.08600281015294106,
      "grad_norm": 9.363341331481934,
      "learning_rate": 9.83382612148079e-05,
      "loss": 0.1888,
      "step": 2127
    },
    {
      "epoch": 0.08604324400820808,
      "grad_norm": 5.140591621398926,
      "learning_rate": 9.833663036042452e-05,
      "loss": 0.1214,
      "step": 2128
    },
    {
      "epoch": 0.0860836778634751,
      "grad_norm": 4.622814655303955,
      "learning_rate": 9.833499871969814e-05,
      "loss": 0.1838,
      "step": 2129
    },
    {
      "epoch": 0.0861241117187421,
      "grad_norm": 6.063584327697754,
      "learning_rate": 9.833336629265529e-05,
      "loss": 0.3445,
      "step": 2130
    },
    {
      "epoch": 0.08616454557400911,
      "grad_norm": 3.7993083000183105,
      "learning_rate": 9.833173307932254e-05,
      "loss": 0.1353,
      "step": 2131
    },
    {
      "epoch": 0.08620497942927613,
      "grad_norm": 5.2384185791015625,
      "learning_rate": 9.833009907972645e-05,
      "loss": 0.1839,
      "step": 2132
    },
    {
      "epoch": 0.08624541328454315,
      "grad_norm": 2.9086854457855225,
      "learning_rate": 9.832846429389361e-05,
      "loss": 0.1604,
      "step": 2133
    },
    {
      "epoch": 0.08628584713981016,
      "grad_norm": 3.6259865760803223,
      "learning_rate": 9.832682872185062e-05,
      "loss": 0.1281,
      "step": 2134
    },
    {
      "epoch": 0.08632628099507718,
      "grad_norm": 9.009800910949707,
      "learning_rate": 9.832519236362406e-05,
      "loss": 0.2322,
      "step": 2135
    },
    {
      "epoch": 0.0863667148503442,
      "grad_norm": 3.8330190181732178,
      "learning_rate": 9.832355521924057e-05,
      "loss": 0.2545,
      "step": 2136
    },
    {
      "epoch": 0.08640714870561121,
      "grad_norm": 3.362452745437622,
      "learning_rate": 9.83219172887268e-05,
      "loss": 0.0815,
      "step": 2137
    },
    {
      "epoch": 0.08644758256087823,
      "grad_norm": 2.826376438140869,
      "learning_rate": 9.832027857210937e-05,
      "loss": 0.061,
      "step": 2138
    },
    {
      "epoch": 0.08648801641614524,
      "grad_norm": 4.0659918785095215,
      "learning_rate": 9.831863906941494e-05,
      "loss": 0.1098,
      "step": 2139
    },
    {
      "epoch": 0.08652845027141225,
      "grad_norm": 2.849652051925659,
      "learning_rate": 9.831699878067019e-05,
      "loss": 0.1212,
      "step": 2140
    },
    {
      "epoch": 0.08656888412667926,
      "grad_norm": 13.696837425231934,
      "learning_rate": 9.83153577059018e-05,
      "loss": 0.365,
      "step": 2141
    },
    {
      "epoch": 0.08660931798194628,
      "grad_norm": 4.096490383148193,
      "learning_rate": 9.831371584513647e-05,
      "loss": 0.1433,
      "step": 2142
    },
    {
      "epoch": 0.0866497518372133,
      "grad_norm": 3.958726644515991,
      "learning_rate": 9.831207319840092e-05,
      "loss": 0.2364,
      "step": 2143
    },
    {
      "epoch": 0.08669018569248031,
      "grad_norm": 7.25415563583374,
      "learning_rate": 9.831042976572184e-05,
      "loss": 0.177,
      "step": 2144
    },
    {
      "epoch": 0.08673061954774733,
      "grad_norm": 4.901478290557861,
      "learning_rate": 9.830878554712601e-05,
      "loss": 0.1875,
      "step": 2145
    },
    {
      "epoch": 0.08677105340301434,
      "grad_norm": 9.820860862731934,
      "learning_rate": 9.830714054264014e-05,
      "loss": 0.2231,
      "step": 2146
    },
    {
      "epoch": 0.08681148725828136,
      "grad_norm": 5.111717700958252,
      "learning_rate": 9.830549475229103e-05,
      "loss": 0.1828,
      "step": 2147
    },
    {
      "epoch": 0.08685192111354838,
      "grad_norm": 2.6316769123077393,
      "learning_rate": 9.830384817610541e-05,
      "loss": 0.2306,
      "step": 2148
    },
    {
      "epoch": 0.0868923549688154,
      "grad_norm": 6.2270965576171875,
      "learning_rate": 9.830220081411009e-05,
      "loss": 0.2074,
      "step": 2149
    },
    {
      "epoch": 0.08693278882408241,
      "grad_norm": 3.444225549697876,
      "learning_rate": 9.830055266633187e-05,
      "loss": 0.2099,
      "step": 2150
    },
    {
      "epoch": 0.08697322267934941,
      "grad_norm": 7.336159706115723,
      "learning_rate": 9.829890373279755e-05,
      "loss": 0.174,
      "step": 2151
    },
    {
      "epoch": 0.08701365653461643,
      "grad_norm": 8.895133972167969,
      "learning_rate": 9.829725401353397e-05,
      "loss": 0.213,
      "step": 2152
    },
    {
      "epoch": 0.08705409038988345,
      "grad_norm": 11.609827995300293,
      "learning_rate": 9.829560350856797e-05,
      "loss": 0.405,
      "step": 2153
    },
    {
      "epoch": 0.08709452424515046,
      "grad_norm": 10.412065505981445,
      "learning_rate": 9.829395221792639e-05,
      "loss": 0.243,
      "step": 2154
    },
    {
      "epoch": 0.08713495810041748,
      "grad_norm": 8.808917045593262,
      "learning_rate": 9.829230014163609e-05,
      "loss": 0.2615,
      "step": 2155
    },
    {
      "epoch": 0.0871753919556845,
      "grad_norm": 7.460705757141113,
      "learning_rate": 9.829064727972394e-05,
      "loss": 0.1821,
      "step": 2156
    },
    {
      "epoch": 0.08721582581095151,
      "grad_norm": 1.7824478149414062,
      "learning_rate": 9.828899363221685e-05,
      "loss": 0.1742,
      "step": 2157
    },
    {
      "epoch": 0.08725625966621853,
      "grad_norm": 6.362964630126953,
      "learning_rate": 9.828733919914172e-05,
      "loss": 0.1389,
      "step": 2158
    },
    {
      "epoch": 0.08729669352148554,
      "grad_norm": 4.3866286277771,
      "learning_rate": 9.828568398052544e-05,
      "loss": 0.1356,
      "step": 2159
    },
    {
      "epoch": 0.08733712737675256,
      "grad_norm": 6.916652202606201,
      "learning_rate": 9.828402797639496e-05,
      "loss": 0.2281,
      "step": 2160
    },
    {
      "epoch": 0.08737756123201958,
      "grad_norm": 4.915858745574951,
      "learning_rate": 9.828237118677722e-05,
      "loss": 0.1874,
      "step": 2161
    },
    {
      "epoch": 0.08741799508728658,
      "grad_norm": 5.881828784942627,
      "learning_rate": 9.828071361169917e-05,
      "loss": 0.2068,
      "step": 2162
    },
    {
      "epoch": 0.0874584289425536,
      "grad_norm": 6.93393087387085,
      "learning_rate": 9.827905525118775e-05,
      "loss": 0.2073,
      "step": 2163
    },
    {
      "epoch": 0.08749886279782061,
      "grad_norm": 5.177983283996582,
      "learning_rate": 9.827739610526997e-05,
      "loss": 0.1505,
      "step": 2164
    },
    {
      "epoch": 0.08753929665308763,
      "grad_norm": 5.685506820678711,
      "learning_rate": 9.82757361739728e-05,
      "loss": 0.1467,
      "step": 2165
    },
    {
      "epoch": 0.08757973050835464,
      "grad_norm": 5.886314392089844,
      "learning_rate": 9.827407545732328e-05,
      "loss": 0.199,
      "step": 2166
    },
    {
      "epoch": 0.08762016436362166,
      "grad_norm": 4.01646614074707,
      "learning_rate": 9.827241395534839e-05,
      "loss": 0.164,
      "step": 2167
    },
    {
      "epoch": 0.08766059821888868,
      "grad_norm": 6.278465270996094,
      "learning_rate": 9.827075166807517e-05,
      "loss": 0.1915,
      "step": 2168
    },
    {
      "epoch": 0.0877010320741557,
      "grad_norm": 3.452907085418701,
      "learning_rate": 9.826908859553067e-05,
      "loss": 0.1505,
      "step": 2169
    },
    {
      "epoch": 0.08774146592942271,
      "grad_norm": 6.76259183883667,
      "learning_rate": 9.826742473774193e-05,
      "loss": 0.2131,
      "step": 2170
    },
    {
      "epoch": 0.08778189978468973,
      "grad_norm": 5.278408527374268,
      "learning_rate": 9.826576009473602e-05,
      "loss": 0.2034,
      "step": 2171
    },
    {
      "epoch": 0.08782233363995673,
      "grad_norm": 1.8797941207885742,
      "learning_rate": 9.826409466654004e-05,
      "loss": 0.0637,
      "step": 2172
    },
    {
      "epoch": 0.08786276749522375,
      "grad_norm": 5.307410717010498,
      "learning_rate": 9.826242845318106e-05,
      "loss": 0.1738,
      "step": 2173
    },
    {
      "epoch": 0.08790320135049076,
      "grad_norm": 4.474775791168213,
      "learning_rate": 9.82607614546862e-05,
      "loss": 0.1902,
      "step": 2174
    },
    {
      "epoch": 0.08794363520575778,
      "grad_norm": 5.492799758911133,
      "learning_rate": 9.825909367108256e-05,
      "loss": 0.2255,
      "step": 2175
    },
    {
      "epoch": 0.0879840690610248,
      "grad_norm": 6.27969217300415,
      "learning_rate": 9.82574251023973e-05,
      "loss": 0.1677,
      "step": 2176
    },
    {
      "epoch": 0.08802450291629181,
      "grad_norm": 5.458862781524658,
      "learning_rate": 9.825575574865756e-05,
      "loss": 0.2177,
      "step": 2177
    },
    {
      "epoch": 0.08806493677155883,
      "grad_norm": 2.7881906032562256,
      "learning_rate": 9.825408560989047e-05,
      "loss": 0.1294,
      "step": 2178
    },
    {
      "epoch": 0.08810537062682584,
      "grad_norm": 5.399051666259766,
      "learning_rate": 9.825241468612323e-05,
      "loss": 0.1928,
      "step": 2179
    },
    {
      "epoch": 0.08814580448209286,
      "grad_norm": 6.9780778884887695,
      "learning_rate": 9.825074297738299e-05,
      "loss": 0.1956,
      "step": 2180
    },
    {
      "epoch": 0.08818623833735988,
      "grad_norm": 5.074653625488281,
      "learning_rate": 9.824907048369698e-05,
      "loss": 0.1334,
      "step": 2181
    },
    {
      "epoch": 0.08822667219262689,
      "grad_norm": 5.125280380249023,
      "learning_rate": 9.82473972050924e-05,
      "loss": 0.2299,
      "step": 2182
    },
    {
      "epoch": 0.0882671060478939,
      "grad_norm": 5.358841896057129,
      "learning_rate": 9.824572314159645e-05,
      "loss": 0.1696,
      "step": 2183
    },
    {
      "epoch": 0.08830753990316091,
      "grad_norm": 5.739036560058594,
      "learning_rate": 9.824404829323638e-05,
      "loss": 0.3155,
      "step": 2184
    },
    {
      "epoch": 0.08834797375842793,
      "grad_norm": 1.8499219417572021,
      "learning_rate": 9.824237266003945e-05,
      "loss": 0.0737,
      "step": 2185
    },
    {
      "epoch": 0.08838840761369494,
      "grad_norm": 6.4467291831970215,
      "learning_rate": 9.82406962420329e-05,
      "loss": 0.1318,
      "step": 2186
    },
    {
      "epoch": 0.08842884146896196,
      "grad_norm": 3.3310604095458984,
      "learning_rate": 9.8239019039244e-05,
      "loss": 0.1254,
      "step": 2187
    },
    {
      "epoch": 0.08846927532422898,
      "grad_norm": 2.150634765625,
      "learning_rate": 9.823734105170004e-05,
      "loss": 0.111,
      "step": 2188
    },
    {
      "epoch": 0.088509709179496,
      "grad_norm": 4.4471964836120605,
      "learning_rate": 9.823566227942832e-05,
      "loss": 0.262,
      "step": 2189
    },
    {
      "epoch": 0.08855014303476301,
      "grad_norm": 2.0157220363616943,
      "learning_rate": 9.823398272245614e-05,
      "loss": 0.0947,
      "step": 2190
    },
    {
      "epoch": 0.08859057689003003,
      "grad_norm": 9.41179370880127,
      "learning_rate": 9.823230238081085e-05,
      "loss": 0.4933,
      "step": 2191
    },
    {
      "epoch": 0.08863101074529704,
      "grad_norm": 4.385516166687012,
      "learning_rate": 9.823062125451977e-05,
      "loss": 0.1172,
      "step": 2192
    },
    {
      "epoch": 0.08867144460056406,
      "grad_norm": 2.8938488960266113,
      "learning_rate": 9.822893934361024e-05,
      "loss": 0.2403,
      "step": 2193
    },
    {
      "epoch": 0.08871187845583106,
      "grad_norm": 3.0300281047821045,
      "learning_rate": 9.822725664810962e-05,
      "loss": 0.0548,
      "step": 2194
    },
    {
      "epoch": 0.08875231231109808,
      "grad_norm": 1.6264137029647827,
      "learning_rate": 9.82255731680453e-05,
      "loss": 0.064,
      "step": 2195
    },
    {
      "epoch": 0.0887927461663651,
      "grad_norm": 2.2518630027770996,
      "learning_rate": 9.822388890344465e-05,
      "loss": 0.1196,
      "step": 2196
    },
    {
      "epoch": 0.08883318002163211,
      "grad_norm": 4.805377960205078,
      "learning_rate": 9.822220385433509e-05,
      "loss": 0.3174,
      "step": 2197
    },
    {
      "epoch": 0.08887361387689913,
      "grad_norm": 3.1552932262420654,
      "learning_rate": 9.822051802074401e-05,
      "loss": 0.2118,
      "step": 2198
    },
    {
      "epoch": 0.08891404773216614,
      "grad_norm": 2.915921688079834,
      "learning_rate": 9.821883140269886e-05,
      "loss": 0.2157,
      "step": 2199
    },
    {
      "epoch": 0.08895448158743316,
      "grad_norm": 5.489482402801514,
      "learning_rate": 9.821714400022705e-05,
      "loss": 0.1539,
      "step": 2200
    },
    {
      "epoch": 0.08899491544270018,
      "grad_norm": 5.814675331115723,
      "learning_rate": 9.821545581335606e-05,
      "loss": 0.3528,
      "step": 2201
    },
    {
      "epoch": 0.08903534929796719,
      "grad_norm": 4.942758083343506,
      "learning_rate": 9.821376684211334e-05,
      "loss": 0.3811,
      "step": 2202
    },
    {
      "epoch": 0.08907578315323421,
      "grad_norm": 5.906837463378906,
      "learning_rate": 9.821207708652636e-05,
      "loss": 0.2288,
      "step": 2203
    },
    {
      "epoch": 0.08911621700850121,
      "grad_norm": 4.528355598449707,
      "learning_rate": 9.821038654662262e-05,
      "loss": 0.1302,
      "step": 2204
    },
    {
      "epoch": 0.08915665086376823,
      "grad_norm": 5.68984842300415,
      "learning_rate": 9.82086952224296e-05,
      "loss": 0.252,
      "step": 2205
    },
    {
      "epoch": 0.08919708471903524,
      "grad_norm": 3.577126979827881,
      "learning_rate": 9.820700311397483e-05,
      "loss": 0.1572,
      "step": 2206
    },
    {
      "epoch": 0.08923751857430226,
      "grad_norm": 2.533400774002075,
      "learning_rate": 9.820531022128586e-05,
      "loss": 0.1264,
      "step": 2207
    },
    {
      "epoch": 0.08927795242956928,
      "grad_norm": 4.671881675720215,
      "learning_rate": 9.82036165443902e-05,
      "loss": 0.1817,
      "step": 2208
    },
    {
      "epoch": 0.08931838628483629,
      "grad_norm": 2.9206783771514893,
      "learning_rate": 9.82019220833154e-05,
      "loss": 0.1385,
      "step": 2209
    },
    {
      "epoch": 0.08935882014010331,
      "grad_norm": 1.8659343719482422,
      "learning_rate": 9.820022683808904e-05,
      "loss": 0.1065,
      "step": 2210
    },
    {
      "epoch": 0.08939925399537033,
      "grad_norm": 6.72808313369751,
      "learning_rate": 9.81985308087387e-05,
      "loss": 0.2347,
      "step": 2211
    },
    {
      "epoch": 0.08943968785063734,
      "grad_norm": 4.954660892486572,
      "learning_rate": 9.819683399529197e-05,
      "loss": 0.1765,
      "step": 2212
    },
    {
      "epoch": 0.08948012170590436,
      "grad_norm": 5.66634464263916,
      "learning_rate": 9.819513639777645e-05,
      "loss": 0.17,
      "step": 2213
    },
    {
      "epoch": 0.08952055556117137,
      "grad_norm": 2.5959365367889404,
      "learning_rate": 9.819343801621974e-05,
      "loss": 0.0897,
      "step": 2214
    },
    {
      "epoch": 0.08956098941643838,
      "grad_norm": 5.502926349639893,
      "learning_rate": 9.819173885064951e-05,
      "loss": 0.3099,
      "step": 2215
    },
    {
      "epoch": 0.0896014232717054,
      "grad_norm": 4.776149749755859,
      "learning_rate": 9.819003890109336e-05,
      "loss": 0.2266,
      "step": 2216
    },
    {
      "epoch": 0.08964185712697241,
      "grad_norm": 5.210961818695068,
      "learning_rate": 9.818833816757897e-05,
      "loss": 0.2625,
      "step": 2217
    },
    {
      "epoch": 0.08968229098223943,
      "grad_norm": 6.604170799255371,
      "learning_rate": 9.818663665013399e-05,
      "loss": 0.1677,
      "step": 2218
    },
    {
      "epoch": 0.08972272483750644,
      "grad_norm": 5.200939655303955,
      "learning_rate": 9.818493434878613e-05,
      "loss": 0.1489,
      "step": 2219
    },
    {
      "epoch": 0.08976315869277346,
      "grad_norm": 6.258219242095947,
      "learning_rate": 9.818323126356305e-05,
      "loss": 0.1502,
      "step": 2220
    },
    {
      "epoch": 0.08980359254804048,
      "grad_norm": 5.764971733093262,
      "learning_rate": 9.818152739449247e-05,
      "loss": 0.2215,
      "step": 2221
    },
    {
      "epoch": 0.08984402640330749,
      "grad_norm": 5.064310073852539,
      "learning_rate": 9.817982274160211e-05,
      "loss": 0.2318,
      "step": 2222
    },
    {
      "epoch": 0.08988446025857451,
      "grad_norm": 8.60485553741455,
      "learning_rate": 9.81781173049197e-05,
      "loss": 0.2433,
      "step": 2223
    },
    {
      "epoch": 0.08992489411384152,
      "grad_norm": 2.727548599243164,
      "learning_rate": 9.817641108447298e-05,
      "loss": 0.1642,
      "step": 2224
    },
    {
      "epoch": 0.08996532796910854,
      "grad_norm": 4.105924129486084,
      "learning_rate": 9.817470408028973e-05,
      "loss": 0.1444,
      "step": 2225
    },
    {
      "epoch": 0.09000576182437554,
      "grad_norm": 4.383919715881348,
      "learning_rate": 9.817299629239768e-05,
      "loss": 0.1759,
      "step": 2226
    },
    {
      "epoch": 0.09004619567964256,
      "grad_norm": 3.384050130844116,
      "learning_rate": 9.817128772082464e-05,
      "loss": 0.0857,
      "step": 2227
    },
    {
      "epoch": 0.09008662953490958,
      "grad_norm": 2.2830734252929688,
      "learning_rate": 9.816957836559839e-05,
      "loss": 0.1017,
      "step": 2228
    },
    {
      "epoch": 0.09012706339017659,
      "grad_norm": 2.948538064956665,
      "learning_rate": 9.816786822674676e-05,
      "loss": 0.151,
      "step": 2229
    },
    {
      "epoch": 0.09016749724544361,
      "grad_norm": 2.9903366565704346,
      "learning_rate": 9.816615730429754e-05,
      "loss": 0.2583,
      "step": 2230
    },
    {
      "epoch": 0.09020793110071063,
      "grad_norm": 3.8737964630126953,
      "learning_rate": 9.81644455982786e-05,
      "loss": 0.1116,
      "step": 2231
    },
    {
      "epoch": 0.09024836495597764,
      "grad_norm": 5.915012359619141,
      "learning_rate": 9.816273310871775e-05,
      "loss": 0.0925,
      "step": 2232
    },
    {
      "epoch": 0.09028879881124466,
      "grad_norm": 5.240262031555176,
      "learning_rate": 9.816101983564288e-05,
      "loss": 0.1441,
      "step": 2233
    },
    {
      "epoch": 0.09032923266651167,
      "grad_norm": 4.159955978393555,
      "learning_rate": 9.815930577908183e-05,
      "loss": 0.2605,
      "step": 2234
    },
    {
      "epoch": 0.09036966652177869,
      "grad_norm": 4.720713138580322,
      "learning_rate": 9.815759093906251e-05,
      "loss": 0.1382,
      "step": 2235
    },
    {
      "epoch": 0.0904101003770457,
      "grad_norm": 5.1785807609558105,
      "learning_rate": 9.815587531561281e-05,
      "loss": 0.1837,
      "step": 2236
    },
    {
      "epoch": 0.09045053423231271,
      "grad_norm": 6.291679859161377,
      "learning_rate": 9.815415890876063e-05,
      "loss": 0.3463,
      "step": 2237
    },
    {
      "epoch": 0.09049096808757973,
      "grad_norm": 3.1257805824279785,
      "learning_rate": 9.81524417185339e-05,
      "loss": 0.1046,
      "step": 2238
    },
    {
      "epoch": 0.09053140194284674,
      "grad_norm": 6.6786208152771,
      "learning_rate": 9.815072374496057e-05,
      "loss": 0.1548,
      "step": 2239
    },
    {
      "epoch": 0.09057183579811376,
      "grad_norm": 4.897136688232422,
      "learning_rate": 9.814900498806856e-05,
      "loss": 0.1624,
      "step": 2240
    },
    {
      "epoch": 0.09061226965338078,
      "grad_norm": 3.5868849754333496,
      "learning_rate": 9.814728544788585e-05,
      "loss": 0.221,
      "step": 2241
    },
    {
      "epoch": 0.09065270350864779,
      "grad_norm": 4.422699451446533,
      "learning_rate": 9.814556512444041e-05,
      "loss": 0.1729,
      "step": 2242
    },
    {
      "epoch": 0.09069313736391481,
      "grad_norm": 3.4839606285095215,
      "learning_rate": 9.814384401776022e-05,
      "loss": 0.1729,
      "step": 2243
    },
    {
      "epoch": 0.09073357121918182,
      "grad_norm": 6.116725444793701,
      "learning_rate": 9.814212212787328e-05,
      "loss": 0.1895,
      "step": 2244
    },
    {
      "epoch": 0.09077400507444884,
      "grad_norm": 3.250274419784546,
      "learning_rate": 9.814039945480762e-05,
      "loss": 0.1047,
      "step": 2245
    },
    {
      "epoch": 0.09081443892971586,
      "grad_norm": 2.428373098373413,
      "learning_rate": 9.813867599859124e-05,
      "loss": 0.1586,
      "step": 2246
    },
    {
      "epoch": 0.09085487278498286,
      "grad_norm": 9.389092445373535,
      "learning_rate": 9.813695175925219e-05,
      "loss": 0.2336,
      "step": 2247
    },
    {
      "epoch": 0.09089530664024988,
      "grad_norm": 7.255312442779541,
      "learning_rate": 9.813522673681852e-05,
      "loss": 0.1713,
      "step": 2248
    },
    {
      "epoch": 0.09093574049551689,
      "grad_norm": 3.897879123687744,
      "learning_rate": 9.813350093131828e-05,
      "loss": 0.1543,
      "step": 2249
    },
    {
      "epoch": 0.09097617435078391,
      "grad_norm": 7.267826557159424,
      "learning_rate": 9.813177434277955e-05,
      "loss": 0.1413,
      "step": 2250
    },
    {
      "epoch": 0.09101660820605093,
      "grad_norm": 2.8419721126556396,
      "learning_rate": 9.813004697123045e-05,
      "loss": 0.1406,
      "step": 2251
    },
    {
      "epoch": 0.09105704206131794,
      "grad_norm": 9.892241477966309,
      "learning_rate": 9.812831881669904e-05,
      "loss": 0.2846,
      "step": 2252
    },
    {
      "epoch": 0.09109747591658496,
      "grad_norm": 4.605840682983398,
      "learning_rate": 9.812658987921346e-05,
      "loss": 0.2651,
      "step": 2253
    },
    {
      "epoch": 0.09113790977185197,
      "grad_norm": 11.260722160339355,
      "learning_rate": 9.81248601588018e-05,
      "loss": 0.2861,
      "step": 2254
    },
    {
      "epoch": 0.09117834362711899,
      "grad_norm": 6.907670497894287,
      "learning_rate": 9.812312965549224e-05,
      "loss": 0.3225,
      "step": 2255
    },
    {
      "epoch": 0.09121877748238601,
      "grad_norm": 8.962262153625488,
      "learning_rate": 9.812139836931292e-05,
      "loss": 0.2525,
      "step": 2256
    },
    {
      "epoch": 0.09125921133765302,
      "grad_norm": 6.3889594078063965,
      "learning_rate": 9.811966630029199e-05,
      "loss": 0.3241,
      "step": 2257
    },
    {
      "epoch": 0.09129964519292003,
      "grad_norm": 5.194851875305176,
      "learning_rate": 9.811793344845766e-05,
      "loss": 0.2486,
      "step": 2258
    },
    {
      "epoch": 0.09134007904818704,
      "grad_norm": 4.658890247344971,
      "learning_rate": 9.811619981383808e-05,
      "loss": 0.1358,
      "step": 2259
    },
    {
      "epoch": 0.09138051290345406,
      "grad_norm": 4.40917444229126,
      "learning_rate": 9.811446539646147e-05,
      "loss": 0.1986,
      "step": 2260
    },
    {
      "epoch": 0.09142094675872108,
      "grad_norm": 3.4145700931549072,
      "learning_rate": 9.811273019635606e-05,
      "loss": 0.1297,
      "step": 2261
    },
    {
      "epoch": 0.09146138061398809,
      "grad_norm": 5.548608779907227,
      "learning_rate": 9.811099421355005e-05,
      "loss": 0.1841,
      "step": 2262
    },
    {
      "epoch": 0.09150181446925511,
      "grad_norm": 5.969812870025635,
      "learning_rate": 9.81092574480717e-05,
      "loss": 0.1879,
      "step": 2263
    },
    {
      "epoch": 0.09154224832452212,
      "grad_norm": 3.1076273918151855,
      "learning_rate": 9.810751989994926e-05,
      "loss": 0.2241,
      "step": 2264
    },
    {
      "epoch": 0.09158268217978914,
      "grad_norm": 10.572524070739746,
      "learning_rate": 9.8105781569211e-05,
      "loss": 0.2509,
      "step": 2265
    },
    {
      "epoch": 0.09162311603505616,
      "grad_norm": 7.3493242263793945,
      "learning_rate": 9.810404245588518e-05,
      "loss": 0.2702,
      "step": 2266
    },
    {
      "epoch": 0.09166354989032317,
      "grad_norm": 3.0788419246673584,
      "learning_rate": 9.810230256000013e-05,
      "loss": 0.1239,
      "step": 2267
    },
    {
      "epoch": 0.09170398374559018,
      "grad_norm": 2.265183925628662,
      "learning_rate": 9.810056188158412e-05,
      "loss": 0.1548,
      "step": 2268
    },
    {
      "epoch": 0.09174441760085719,
      "grad_norm": 11.08302116394043,
      "learning_rate": 9.80988204206655e-05,
      "loss": 0.24,
      "step": 2269
    },
    {
      "epoch": 0.09178485145612421,
      "grad_norm": 13.470440864562988,
      "learning_rate": 9.809707817727256e-05,
      "loss": 0.2711,
      "step": 2270
    },
    {
      "epoch": 0.09182528531139122,
      "grad_norm": 2.089656352996826,
      "learning_rate": 9.809533515143365e-05,
      "loss": 0.1104,
      "step": 2271
    },
    {
      "epoch": 0.09186571916665824,
      "grad_norm": 5.489585876464844,
      "learning_rate": 9.809359134317717e-05,
      "loss": 0.1055,
      "step": 2272
    },
    {
      "epoch": 0.09190615302192526,
      "grad_norm": 3.6772634983062744,
      "learning_rate": 9.809184675253145e-05,
      "loss": 0.1474,
      "step": 2273
    },
    {
      "epoch": 0.09194658687719227,
      "grad_norm": 3.185612201690674,
      "learning_rate": 9.809010137952486e-05,
      "loss": 0.1982,
      "step": 2274
    },
    {
      "epoch": 0.09198702073245929,
      "grad_norm": 3.679205894470215,
      "learning_rate": 9.808835522418583e-05,
      "loss": 0.2316,
      "step": 2275
    },
    {
      "epoch": 0.0920274545877263,
      "grad_norm": 5.975610733032227,
      "learning_rate": 9.808660828654274e-05,
      "loss": 0.1062,
      "step": 2276
    },
    {
      "epoch": 0.09206788844299332,
      "grad_norm": 6.427095890045166,
      "learning_rate": 9.808486056662402e-05,
      "loss": 0.2033,
      "step": 2277
    },
    {
      "epoch": 0.09210832229826034,
      "grad_norm": 4.552356243133545,
      "learning_rate": 9.80831120644581e-05,
      "loss": 0.1275,
      "step": 2278
    },
    {
      "epoch": 0.09214875615352734,
      "grad_norm": 3.961167335510254,
      "learning_rate": 9.808136278007342e-05,
      "loss": 0.3201,
      "step": 2279
    },
    {
      "epoch": 0.09218919000879436,
      "grad_norm": 4.479267120361328,
      "learning_rate": 9.807961271349845e-05,
      "loss": 0.2011,
      "step": 2280
    },
    {
      "epoch": 0.09222962386406137,
      "grad_norm": 4.2484893798828125,
      "learning_rate": 9.807786186476165e-05,
      "loss": 0.2427,
      "step": 2281
    },
    {
      "epoch": 0.09227005771932839,
      "grad_norm": 4.434413909912109,
      "learning_rate": 9.807611023389149e-05,
      "loss": 0.3435,
      "step": 2282
    },
    {
      "epoch": 0.09231049157459541,
      "grad_norm": 5.6522111892700195,
      "learning_rate": 9.80743578209165e-05,
      "loss": 0.2296,
      "step": 2283
    },
    {
      "epoch": 0.09235092542986242,
      "grad_norm": 6.11889123916626,
      "learning_rate": 9.807260462586515e-05,
      "loss": 0.3065,
      "step": 2284
    },
    {
      "epoch": 0.09239135928512944,
      "grad_norm": 6.927918910980225,
      "learning_rate": 9.8070850648766e-05,
      "loss": 0.1586,
      "step": 2285
    },
    {
      "epoch": 0.09243179314039646,
      "grad_norm": 5.574503421783447,
      "learning_rate": 9.806909588964757e-05,
      "loss": 0.2084,
      "step": 2286
    },
    {
      "epoch": 0.09247222699566347,
      "grad_norm": 5.07774543762207,
      "learning_rate": 9.806734034853838e-05,
      "loss": 0.1233,
      "step": 2287
    },
    {
      "epoch": 0.09251266085093049,
      "grad_norm": 3.0053672790527344,
      "learning_rate": 9.806558402546702e-05,
      "loss": 0.1388,
      "step": 2288
    },
    {
      "epoch": 0.0925530947061975,
      "grad_norm": 7.3748273849487305,
      "learning_rate": 9.806382692046205e-05,
      "loss": 0.1838,
      "step": 2289
    },
    {
      "epoch": 0.09259352856146451,
      "grad_norm": 9.24937629699707,
      "learning_rate": 9.806206903355207e-05,
      "loss": 0.2564,
      "step": 2290
    },
    {
      "epoch": 0.09263396241673152,
      "grad_norm": 4.831010341644287,
      "learning_rate": 9.806031036476565e-05,
      "loss": 0.1918,
      "step": 2291
    },
    {
      "epoch": 0.09267439627199854,
      "grad_norm": 7.247109413146973,
      "learning_rate": 9.80585509141314e-05,
      "loss": 0.1526,
      "step": 2292
    },
    {
      "epoch": 0.09271483012726556,
      "grad_norm": 5.867323398590088,
      "learning_rate": 9.805679068167797e-05,
      "loss": 0.1606,
      "step": 2293
    },
    {
      "epoch": 0.09275526398253257,
      "grad_norm": 6.42877197265625,
      "learning_rate": 9.805502966743398e-05,
      "loss": 0.262,
      "step": 2294
    },
    {
      "epoch": 0.09279569783779959,
      "grad_norm": 6.985595226287842,
      "learning_rate": 9.805326787142807e-05,
      "loss": 0.24,
      "step": 2295
    },
    {
      "epoch": 0.0928361316930666,
      "grad_norm": 3.854750871658325,
      "learning_rate": 9.805150529368894e-05,
      "loss": 0.1979,
      "step": 2296
    },
    {
      "epoch": 0.09287656554833362,
      "grad_norm": 5.416162967681885,
      "learning_rate": 9.804974193424521e-05,
      "loss": 0.244,
      "step": 2297
    },
    {
      "epoch": 0.09291699940360064,
      "grad_norm": 9.629919052124023,
      "learning_rate": 9.80479777931256e-05,
      "loss": 0.2549,
      "step": 2298
    },
    {
      "epoch": 0.09295743325886766,
      "grad_norm": 5.016849517822266,
      "learning_rate": 9.80462128703588e-05,
      "loss": 0.1112,
      "step": 2299
    },
    {
      "epoch": 0.09299786711413467,
      "grad_norm": 7.681748867034912,
      "learning_rate": 9.804444716597353e-05,
      "loss": 0.2611,
      "step": 2300
    },
    {
      "epoch": 0.09303830096940167,
      "grad_norm": 4.031108856201172,
      "learning_rate": 9.80426806799985e-05,
      "loss": 0.1459,
      "step": 2301
    },
    {
      "epoch": 0.09307873482466869,
      "grad_norm": 3.7044501304626465,
      "learning_rate": 9.804091341246246e-05,
      "loss": 0.1416,
      "step": 2302
    },
    {
      "epoch": 0.09311916867993571,
      "grad_norm": 5.128602504730225,
      "learning_rate": 9.803914536339415e-05,
      "loss": 0.2431,
      "step": 2303
    },
    {
      "epoch": 0.09315960253520272,
      "grad_norm": 6.475131034851074,
      "learning_rate": 9.803737653282235e-05,
      "loss": 0.1361,
      "step": 2304
    },
    {
      "epoch": 0.09320003639046974,
      "grad_norm": 3.5635321140289307,
      "learning_rate": 9.803560692077581e-05,
      "loss": 0.2319,
      "step": 2305
    },
    {
      "epoch": 0.09324047024573676,
      "grad_norm": 3.0276970863342285,
      "learning_rate": 9.803383652728332e-05,
      "loss": 0.1658,
      "step": 2306
    },
    {
      "epoch": 0.09328090410100377,
      "grad_norm": 6.259897232055664,
      "learning_rate": 9.80320653523737e-05,
      "loss": 0.2293,
      "step": 2307
    },
    {
      "epoch": 0.09332133795627079,
      "grad_norm": 2.9486749172210693,
      "learning_rate": 9.803029339607577e-05,
      "loss": 0.1483,
      "step": 2308
    },
    {
      "epoch": 0.0933617718115378,
      "grad_norm": 2.9913809299468994,
      "learning_rate": 9.802852065841832e-05,
      "loss": 0.1336,
      "step": 2309
    },
    {
      "epoch": 0.09340220566680482,
      "grad_norm": 7.980592250823975,
      "learning_rate": 9.802674713943023e-05,
      "loss": 0.3284,
      "step": 2310
    },
    {
      "epoch": 0.09344263952207182,
      "grad_norm": 6.209901809692383,
      "learning_rate": 9.802497283914032e-05,
      "loss": 0.2272,
      "step": 2311
    },
    {
      "epoch": 0.09348307337733884,
      "grad_norm": 6.8691606521606445,
      "learning_rate": 9.802319775757746e-05,
      "loss": 0.2789,
      "step": 2312
    },
    {
      "epoch": 0.09352350723260586,
      "grad_norm": 3.4447829723358154,
      "learning_rate": 9.802142189477055e-05,
      "loss": 0.2127,
      "step": 2313
    },
    {
      "epoch": 0.09356394108787287,
      "grad_norm": 2.679886817932129,
      "learning_rate": 9.801964525074846e-05,
      "loss": 0.1106,
      "step": 2314
    },
    {
      "epoch": 0.09360437494313989,
      "grad_norm": 11.390369415283203,
      "learning_rate": 9.80178678255401e-05,
      "loss": 0.2558,
      "step": 2315
    },
    {
      "epoch": 0.0936448087984069,
      "grad_norm": 4.282725811004639,
      "learning_rate": 9.801608961917436e-05,
      "loss": 0.0932,
      "step": 2316
    },
    {
      "epoch": 0.09368524265367392,
      "grad_norm": 4.97322940826416,
      "learning_rate": 9.801431063168022e-05,
      "loss": 0.1618,
      "step": 2317
    },
    {
      "epoch": 0.09372567650894094,
      "grad_norm": 3.6021881103515625,
      "learning_rate": 9.801253086308657e-05,
      "loss": 0.2275,
      "step": 2318
    },
    {
      "epoch": 0.09376611036420796,
      "grad_norm": 3.369175434112549,
      "learning_rate": 9.80107503134224e-05,
      "loss": 0.0691,
      "step": 2319
    },
    {
      "epoch": 0.09380654421947497,
      "grad_norm": 4.46785831451416,
      "learning_rate": 9.800896898271663e-05,
      "loss": 0.2762,
      "step": 2320
    },
    {
      "epoch": 0.09384697807474199,
      "grad_norm": 7.003039836883545,
      "learning_rate": 9.800718687099829e-05,
      "loss": 0.2147,
      "step": 2321
    },
    {
      "epoch": 0.09388741193000899,
      "grad_norm": 9.663853645324707,
      "learning_rate": 9.800540397829635e-05,
      "loss": 0.3023,
      "step": 2322
    },
    {
      "epoch": 0.093927845785276,
      "grad_norm": 4.49720573425293,
      "learning_rate": 9.800362030463981e-05,
      "loss": 0.2633,
      "step": 2323
    },
    {
      "epoch": 0.09396827964054302,
      "grad_norm": 4.352441787719727,
      "learning_rate": 9.800183585005769e-05,
      "loss": 0.213,
      "step": 2324
    },
    {
      "epoch": 0.09400871349581004,
      "grad_norm": 5.2932281494140625,
      "learning_rate": 9.800005061457903e-05,
      "loss": 0.1802,
      "step": 2325
    },
    {
      "epoch": 0.09404914735107706,
      "grad_norm": 2.619900703430176,
      "learning_rate": 9.799826459823284e-05,
      "loss": 0.1651,
      "step": 2326
    },
    {
      "epoch": 0.09408958120634407,
      "grad_norm": 3.619932174682617,
      "learning_rate": 9.799647780104821e-05,
      "loss": 0.2416,
      "step": 2327
    },
    {
      "epoch": 0.09413001506161109,
      "grad_norm": 2.854600429534912,
      "learning_rate": 9.79946902230542e-05,
      "loss": 0.142,
      "step": 2328
    },
    {
      "epoch": 0.0941704489168781,
      "grad_norm": 6.940002918243408,
      "learning_rate": 9.799290186427989e-05,
      "loss": 0.2661,
      "step": 2329
    },
    {
      "epoch": 0.09421088277214512,
      "grad_norm": 7.716872692108154,
      "learning_rate": 9.799111272475437e-05,
      "loss": 0.2366,
      "step": 2330
    },
    {
      "epoch": 0.09425131662741214,
      "grad_norm": 5.361137390136719,
      "learning_rate": 9.798932280450673e-05,
      "loss": 0.1887,
      "step": 2331
    },
    {
      "epoch": 0.09429175048267915,
      "grad_norm": 5.568938732147217,
      "learning_rate": 9.798753210356612e-05,
      "loss": 0.2432,
      "step": 2332
    },
    {
      "epoch": 0.09433218433794616,
      "grad_norm": 9.440380096435547,
      "learning_rate": 9.798574062196165e-05,
      "loss": 0.3208,
      "step": 2333
    },
    {
      "epoch": 0.09437261819321317,
      "grad_norm": 2.5683531761169434,
      "learning_rate": 9.798394835972246e-05,
      "loss": 0.1044,
      "step": 2334
    },
    {
      "epoch": 0.09441305204848019,
      "grad_norm": 3.219914197921753,
      "learning_rate": 9.798215531687771e-05,
      "loss": 0.2715,
      "step": 2335
    },
    {
      "epoch": 0.0944534859037472,
      "grad_norm": 4.68519401550293,
      "learning_rate": 9.79803614934566e-05,
      "loss": 0.2324,
      "step": 2336
    },
    {
      "epoch": 0.09449391975901422,
      "grad_norm": 4.057682514190674,
      "learning_rate": 9.797856688948825e-05,
      "loss": 0.2082,
      "step": 2337
    },
    {
      "epoch": 0.09453435361428124,
      "grad_norm": 7.075843334197998,
      "learning_rate": 9.797677150500193e-05,
      "loss": 0.201,
      "step": 2338
    },
    {
      "epoch": 0.09457478746954825,
      "grad_norm": 6.316398620605469,
      "learning_rate": 9.797497534002678e-05,
      "loss": 0.1786,
      "step": 2339
    },
    {
      "epoch": 0.09461522132481527,
      "grad_norm": 2.8940532207489014,
      "learning_rate": 9.797317839459206e-05,
      "loss": 0.1701,
      "step": 2340
    },
    {
      "epoch": 0.09465565518008229,
      "grad_norm": 4.366870403289795,
      "learning_rate": 9.7971380668727e-05,
      "loss": 0.1744,
      "step": 2341
    },
    {
      "epoch": 0.0946960890353493,
      "grad_norm": 6.161758899688721,
      "learning_rate": 9.796958216246083e-05,
      "loss": 0.0882,
      "step": 2342
    },
    {
      "epoch": 0.0947365228906163,
      "grad_norm": 6.820497989654541,
      "learning_rate": 9.796778287582281e-05,
      "loss": 0.18,
      "step": 2343
    },
    {
      "epoch": 0.09477695674588332,
      "grad_norm": 4.593965530395508,
      "learning_rate": 9.796598280884222e-05,
      "loss": 0.18,
      "step": 2344
    },
    {
      "epoch": 0.09481739060115034,
      "grad_norm": 4.700291156768799,
      "learning_rate": 9.796418196154834e-05,
      "loss": 0.1334,
      "step": 2345
    },
    {
      "epoch": 0.09485782445641736,
      "grad_norm": 6.834756851196289,
      "learning_rate": 9.796238033397048e-05,
      "loss": 0.264,
      "step": 2346
    },
    {
      "epoch": 0.09489825831168437,
      "grad_norm": 12.748082160949707,
      "learning_rate": 9.796057792613791e-05,
      "loss": 0.2344,
      "step": 2347
    },
    {
      "epoch": 0.09493869216695139,
      "grad_norm": 5.65842866897583,
      "learning_rate": 9.795877473807999e-05,
      "loss": 0.1911,
      "step": 2348
    },
    {
      "epoch": 0.0949791260222184,
      "grad_norm": 4.4694719314575195,
      "learning_rate": 9.795697076982604e-05,
      "loss": 0.1223,
      "step": 2349
    },
    {
      "epoch": 0.09501955987748542,
      "grad_norm": 4.0597734451293945,
      "learning_rate": 9.79551660214054e-05,
      "loss": 0.2015,
      "step": 2350
    },
    {
      "epoch": 0.09505999373275244,
      "grad_norm": 7.8599419593811035,
      "learning_rate": 9.795336049284745e-05,
      "loss": 0.3264,
      "step": 2351
    },
    {
      "epoch": 0.09510042758801945,
      "grad_norm": 7.018628120422363,
      "learning_rate": 9.795155418418155e-05,
      "loss": 0.2164,
      "step": 2352
    },
    {
      "epoch": 0.09514086144328647,
      "grad_norm": 11.056584358215332,
      "learning_rate": 9.794974709543708e-05,
      "loss": 0.35,
      "step": 2353
    },
    {
      "epoch": 0.09518129529855347,
      "grad_norm": 3.0478007793426514,
      "learning_rate": 9.794793922664344e-05,
      "loss": 0.1912,
      "step": 2354
    },
    {
      "epoch": 0.09522172915382049,
      "grad_norm": 9.538176536560059,
      "learning_rate": 9.794613057783006e-05,
      "loss": 0.2636,
      "step": 2355
    },
    {
      "epoch": 0.0952621630090875,
      "grad_norm": 2.7587785720825195,
      "learning_rate": 9.794432114902631e-05,
      "loss": 0.2339,
      "step": 2356
    },
    {
      "epoch": 0.09530259686435452,
      "grad_norm": 6.431471347808838,
      "learning_rate": 9.79425109402617e-05,
      "loss": 0.2687,
      "step": 2357
    },
    {
      "epoch": 0.09534303071962154,
      "grad_norm": 2.70796275138855,
      "learning_rate": 9.794069995156564e-05,
      "loss": 0.2169,
      "step": 2358
    },
    {
      "epoch": 0.09538346457488855,
      "grad_norm": 2.348987579345703,
      "learning_rate": 9.793888818296759e-05,
      "loss": 0.1636,
      "step": 2359
    },
    {
      "epoch": 0.09542389843015557,
      "grad_norm": 5.423366069793701,
      "learning_rate": 9.7937075634497e-05,
      "loss": 0.2116,
      "step": 2360
    },
    {
      "epoch": 0.09546433228542259,
      "grad_norm": 6.75554895401001,
      "learning_rate": 9.793526230618342e-05,
      "loss": 0.1585,
      "step": 2361
    },
    {
      "epoch": 0.0955047661406896,
      "grad_norm": 4.1601786613464355,
      "learning_rate": 9.793344819805629e-05,
      "loss": 0.1993,
      "step": 2362
    },
    {
      "epoch": 0.09554519999595662,
      "grad_norm": 5.453439235687256,
      "learning_rate": 9.793163331014514e-05,
      "loss": 0.1417,
      "step": 2363
    },
    {
      "epoch": 0.09558563385122364,
      "grad_norm": 9.135724067687988,
      "learning_rate": 9.792981764247951e-05,
      "loss": 0.1983,
      "step": 2364
    },
    {
      "epoch": 0.09562606770649064,
      "grad_norm": 6.387965679168701,
      "learning_rate": 9.792800119508894e-05,
      "loss": 0.2236,
      "step": 2365
    },
    {
      "epoch": 0.09566650156175766,
      "grad_norm": 5.228516578674316,
      "learning_rate": 9.792618396800295e-05,
      "loss": 0.1353,
      "step": 2366
    },
    {
      "epoch": 0.09570693541702467,
      "grad_norm": 8.783320426940918,
      "learning_rate": 9.792436596125112e-05,
      "loss": 0.1833,
      "step": 2367
    },
    {
      "epoch": 0.09574736927229169,
      "grad_norm": 4.6242499351501465,
      "learning_rate": 9.792254717486302e-05,
      "loss": 0.179,
      "step": 2368
    },
    {
      "epoch": 0.0957878031275587,
      "grad_norm": 6.260232448577881,
      "learning_rate": 9.792072760886825e-05,
      "loss": 0.2655,
      "step": 2369
    },
    {
      "epoch": 0.09582823698282572,
      "grad_norm": 5.022913455963135,
      "learning_rate": 9.791890726329641e-05,
      "loss": 0.2284,
      "step": 2370
    },
    {
      "epoch": 0.09586867083809274,
      "grad_norm": 2.8345248699188232,
      "learning_rate": 9.791708613817711e-05,
      "loss": 0.2472,
      "step": 2371
    },
    {
      "epoch": 0.09590910469335975,
      "grad_norm": 4.589327812194824,
      "learning_rate": 9.791526423353996e-05,
      "loss": 0.1778,
      "step": 2372
    },
    {
      "epoch": 0.09594953854862677,
      "grad_norm": 5.834810733795166,
      "learning_rate": 9.791344154941462e-05,
      "loss": 0.2944,
      "step": 2373
    },
    {
      "epoch": 0.09598997240389379,
      "grad_norm": 7.875876426696777,
      "learning_rate": 9.791161808583073e-05,
      "loss": 0.2651,
      "step": 2374
    },
    {
      "epoch": 0.09603040625916079,
      "grad_norm": 5.008292198181152,
      "learning_rate": 9.790979384281797e-05,
      "loss": 0.1309,
      "step": 2375
    },
    {
      "epoch": 0.0960708401144278,
      "grad_norm": 3.4788336753845215,
      "learning_rate": 9.790796882040598e-05,
      "loss": 0.2338,
      "step": 2376
    },
    {
      "epoch": 0.09611127396969482,
      "grad_norm": 4.632124900817871,
      "learning_rate": 9.790614301862451e-05,
      "loss": 0.2173,
      "step": 2377
    },
    {
      "epoch": 0.09615170782496184,
      "grad_norm": 6.004955291748047,
      "learning_rate": 9.79043164375032e-05,
      "loss": 0.2094,
      "step": 2378
    },
    {
      "epoch": 0.09619214168022885,
      "grad_norm": 3.2258377075195312,
      "learning_rate": 9.79024890770718e-05,
      "loss": 0.1867,
      "step": 2379
    },
    {
      "epoch": 0.09623257553549587,
      "grad_norm": 5.448134422302246,
      "learning_rate": 9.790066093736003e-05,
      "loss": 0.2152,
      "step": 2380
    },
    {
      "epoch": 0.09627300939076289,
      "grad_norm": 3.776277542114258,
      "learning_rate": 9.789883201839762e-05,
      "loss": 0.1363,
      "step": 2381
    },
    {
      "epoch": 0.0963134432460299,
      "grad_norm": 3.4171829223632812,
      "learning_rate": 9.789700232021435e-05,
      "loss": 0.1202,
      "step": 2382
    },
    {
      "epoch": 0.09635387710129692,
      "grad_norm": 4.5001139640808105,
      "learning_rate": 9.789517184283995e-05,
      "loss": 0.2128,
      "step": 2383
    },
    {
      "epoch": 0.09639431095656394,
      "grad_norm": 7.376933574676514,
      "learning_rate": 9.789334058630423e-05,
      "loss": 0.1918,
      "step": 2384
    },
    {
      "epoch": 0.09643474481183095,
      "grad_norm": 12.173112869262695,
      "learning_rate": 9.789150855063696e-05,
      "loss": 0.2342,
      "step": 2385
    },
    {
      "epoch": 0.09647517866709795,
      "grad_norm": 7.604058742523193,
      "learning_rate": 9.788967573586796e-05,
      "loss": 0.1715,
      "step": 2386
    },
    {
      "epoch": 0.09651561252236497,
      "grad_norm": 3.3801162242889404,
      "learning_rate": 9.788784214202702e-05,
      "loss": 0.101,
      "step": 2387
    },
    {
      "epoch": 0.09655604637763199,
      "grad_norm": 3.534852981567383,
      "learning_rate": 9.7886007769144e-05,
      "loss": 0.1955,
      "step": 2388
    },
    {
      "epoch": 0.096596480232899,
      "grad_norm": 2.4402477741241455,
      "learning_rate": 9.788417261724874e-05,
      "loss": 0.13,
      "step": 2389
    },
    {
      "epoch": 0.09663691408816602,
      "grad_norm": 2.2972757816314697,
      "learning_rate": 9.788233668637105e-05,
      "loss": 0.1139,
      "step": 2390
    },
    {
      "epoch": 0.09667734794343304,
      "grad_norm": 7.445667266845703,
      "learning_rate": 9.788049997654086e-05,
      "loss": 0.2261,
      "step": 2391
    },
    {
      "epoch": 0.09671778179870005,
      "grad_norm": 5.345921993255615,
      "learning_rate": 9.7878662487788e-05,
      "loss": 0.1429,
      "step": 2392
    },
    {
      "epoch": 0.09675821565396707,
      "grad_norm": 6.710997581481934,
      "learning_rate": 9.78768242201424e-05,
      "loss": 0.2283,
      "step": 2393
    },
    {
      "epoch": 0.09679864950923409,
      "grad_norm": 4.346287250518799,
      "learning_rate": 9.787498517363393e-05,
      "loss": 0.1457,
      "step": 2394
    },
    {
      "epoch": 0.0968390833645011,
      "grad_norm": 6.163570880889893,
      "learning_rate": 9.787314534829254e-05,
      "loss": 0.2182,
      "step": 2395
    },
    {
      "epoch": 0.09687951721976812,
      "grad_norm": 5.140466213226318,
      "learning_rate": 9.787130474414814e-05,
      "loss": 0.2425,
      "step": 2396
    },
    {
      "epoch": 0.09691995107503512,
      "grad_norm": 2.088500499725342,
      "learning_rate": 9.786946336123066e-05,
      "loss": 0.0817,
      "step": 2397
    },
    {
      "epoch": 0.09696038493030214,
      "grad_norm": 7.57895040512085,
      "learning_rate": 9.786762119957009e-05,
      "loss": 0.2599,
      "step": 2398
    },
    {
      "epoch": 0.09700081878556915,
      "grad_norm": 9.20680046081543,
      "learning_rate": 9.786577825919639e-05,
      "loss": 0.2617,
      "step": 2399
    },
    {
      "epoch": 0.09704125264083617,
      "grad_norm": 3.2030398845672607,
      "learning_rate": 9.786393454013951e-05,
      "loss": 0.1953,
      "step": 2400
    },
    {
      "epoch": 0.09708168649610319,
      "grad_norm": 5.331843852996826,
      "learning_rate": 9.786209004242949e-05,
      "loss": 0.2938,
      "step": 2401
    },
    {
      "epoch": 0.0971221203513702,
      "grad_norm": 8.140605926513672,
      "learning_rate": 9.78602447660963e-05,
      "loss": 0.1807,
      "step": 2402
    },
    {
      "epoch": 0.09716255420663722,
      "grad_norm": 7.53940486907959,
      "learning_rate": 9.785839871116998e-05,
      "loss": 0.1917,
      "step": 2403
    },
    {
      "epoch": 0.09720298806190424,
      "grad_norm": 5.732079982757568,
      "learning_rate": 9.785655187768055e-05,
      "loss": 0.1968,
      "step": 2404
    },
    {
      "epoch": 0.09724342191717125,
      "grad_norm": 7.798459053039551,
      "learning_rate": 9.785470426565805e-05,
      "loss": 0.2388,
      "step": 2405
    },
    {
      "epoch": 0.09728385577243827,
      "grad_norm": 6.750057697296143,
      "learning_rate": 9.785285587513255e-05,
      "loss": 0.1345,
      "step": 2406
    },
    {
      "epoch": 0.09732428962770527,
      "grad_norm": 3.0654959678649902,
      "learning_rate": 9.785100670613412e-05,
      "loss": 0.1532,
      "step": 2407
    },
    {
      "epoch": 0.09736472348297229,
      "grad_norm": 7.730163097381592,
      "learning_rate": 9.784915675869284e-05,
      "loss": 0.2342,
      "step": 2408
    },
    {
      "epoch": 0.0974051573382393,
      "grad_norm": 2.1438019275665283,
      "learning_rate": 9.784730603283879e-05,
      "loss": 0.0762,
      "step": 2409
    },
    {
      "epoch": 0.09744559119350632,
      "grad_norm": 5.887228012084961,
      "learning_rate": 9.78454545286021e-05,
      "loss": 0.2943,
      "step": 2410
    },
    {
      "epoch": 0.09748602504877334,
      "grad_norm": 4.239229679107666,
      "learning_rate": 9.784360224601287e-05,
      "loss": 0.1639,
      "step": 2411
    },
    {
      "epoch": 0.09752645890404035,
      "grad_norm": 5.6172075271606445,
      "learning_rate": 9.784174918510126e-05,
      "loss": 0.2887,
      "step": 2412
    },
    {
      "epoch": 0.09756689275930737,
      "grad_norm": 3.6733312606811523,
      "learning_rate": 9.78398953458974e-05,
      "loss": 0.1413,
      "step": 2413
    },
    {
      "epoch": 0.09760732661457439,
      "grad_norm": 2.701106071472168,
      "learning_rate": 9.783804072843143e-05,
      "loss": 0.0637,
      "step": 2414
    },
    {
      "epoch": 0.0976477604698414,
      "grad_norm": 3.5051467418670654,
      "learning_rate": 9.783618533273355e-05,
      "loss": 0.1592,
      "step": 2415
    },
    {
      "epoch": 0.09768819432510842,
      "grad_norm": 2.417534828186035,
      "learning_rate": 9.783432915883392e-05,
      "loss": 0.1798,
      "step": 2416
    },
    {
      "epoch": 0.09772862818037543,
      "grad_norm": 3.666066884994507,
      "learning_rate": 9.783247220676277e-05,
      "loss": 0.1354,
      "step": 2417
    },
    {
      "epoch": 0.09776906203564244,
      "grad_norm": 3.6225175857543945,
      "learning_rate": 9.783061447655028e-05,
      "loss": 0.1361,
      "step": 2418
    },
    {
      "epoch": 0.09780949589090945,
      "grad_norm": 5.939593315124512,
      "learning_rate": 9.782875596822667e-05,
      "loss": 0.3096,
      "step": 2419
    },
    {
      "epoch": 0.09784992974617647,
      "grad_norm": 9.711254119873047,
      "learning_rate": 9.782689668182219e-05,
      "loss": 0.2653,
      "step": 2420
    },
    {
      "epoch": 0.09789036360144349,
      "grad_norm": 3.847031354904175,
      "learning_rate": 9.782503661736708e-05,
      "loss": 0.1406,
      "step": 2421
    },
    {
      "epoch": 0.0979307974567105,
      "grad_norm": 4.211605548858643,
      "learning_rate": 9.78231757748916e-05,
      "loss": 0.1231,
      "step": 2422
    },
    {
      "epoch": 0.09797123131197752,
      "grad_norm": 5.243746280670166,
      "learning_rate": 9.782131415442604e-05,
      "loss": 0.2451,
      "step": 2423
    },
    {
      "epoch": 0.09801166516724454,
      "grad_norm": 7.009279727935791,
      "learning_rate": 9.781945175600064e-05,
      "loss": 0.1939,
      "step": 2424
    },
    {
      "epoch": 0.09805209902251155,
      "grad_norm": 4.188933849334717,
      "learning_rate": 9.781758857964573e-05,
      "loss": 0.1123,
      "step": 2425
    },
    {
      "epoch": 0.09809253287777857,
      "grad_norm": 6.537064552307129,
      "learning_rate": 9.781572462539163e-05,
      "loss": 0.1959,
      "step": 2426
    },
    {
      "epoch": 0.09813296673304558,
      "grad_norm": 9.561412811279297,
      "learning_rate": 9.781385989326863e-05,
      "loss": 0.2633,
      "step": 2427
    },
    {
      "epoch": 0.0981734005883126,
      "grad_norm": 5.648425579071045,
      "learning_rate": 9.78119943833071e-05,
      "loss": 0.1442,
      "step": 2428
    },
    {
      "epoch": 0.0982138344435796,
      "grad_norm": 11.195969581604004,
      "learning_rate": 9.781012809553736e-05,
      "loss": 0.327,
      "step": 2429
    },
    {
      "epoch": 0.09825426829884662,
      "grad_norm": 8.821499824523926,
      "learning_rate": 9.78082610299898e-05,
      "loss": 0.3025,
      "step": 2430
    },
    {
      "epoch": 0.09829470215411364,
      "grad_norm": 2.235635995864868,
      "learning_rate": 9.780639318669477e-05,
      "loss": 0.1974,
      "step": 2431
    },
    {
      "epoch": 0.09833513600938065,
      "grad_norm": 4.393800735473633,
      "learning_rate": 9.780452456568265e-05,
      "loss": 0.1963,
      "step": 2432
    },
    {
      "epoch": 0.09837556986464767,
      "grad_norm": 3.7783164978027344,
      "learning_rate": 9.780265516698385e-05,
      "loss": 0.1642,
      "step": 2433
    },
    {
      "epoch": 0.09841600371991469,
      "grad_norm": 5.590644836425781,
      "learning_rate": 9.78007849906288e-05,
      "loss": 0.268,
      "step": 2434
    },
    {
      "epoch": 0.0984564375751817,
      "grad_norm": 2.904095411300659,
      "learning_rate": 9.779891403664791e-05,
      "loss": 0.1476,
      "step": 2435
    },
    {
      "epoch": 0.09849687143044872,
      "grad_norm": 7.576403617858887,
      "learning_rate": 9.779704230507161e-05,
      "loss": 0.1813,
      "step": 2436
    },
    {
      "epoch": 0.09853730528571573,
      "grad_norm": 4.0357513427734375,
      "learning_rate": 9.779516979593034e-05,
      "loss": 0.1725,
      "step": 2437
    },
    {
      "epoch": 0.09857773914098275,
      "grad_norm": 5.825844764709473,
      "learning_rate": 9.779329650925459e-05,
      "loss": 0.1834,
      "step": 2438
    },
    {
      "epoch": 0.09861817299624975,
      "grad_norm": 9.193845748901367,
      "learning_rate": 9.779142244507481e-05,
      "loss": 0.2482,
      "step": 2439
    },
    {
      "epoch": 0.09865860685151677,
      "grad_norm": 6.3326096534729,
      "learning_rate": 9.778954760342149e-05,
      "loss": 0.156,
      "step": 2440
    },
    {
      "epoch": 0.09869904070678379,
      "grad_norm": 7.504809856414795,
      "learning_rate": 9.778767198432516e-05,
      "loss": 0.259,
      "step": 2441
    },
    {
      "epoch": 0.0987394745620508,
      "grad_norm": 4.722307205200195,
      "learning_rate": 9.77857955878163e-05,
      "loss": 0.1556,
      "step": 2442
    },
    {
      "epoch": 0.09877990841731782,
      "grad_norm": 6.758153915405273,
      "learning_rate": 9.778391841392544e-05,
      "loss": 0.27,
      "step": 2443
    },
    {
      "epoch": 0.09882034227258484,
      "grad_norm": 4.701530933380127,
      "learning_rate": 9.778204046268314e-05,
      "loss": 0.1771,
      "step": 2444
    },
    {
      "epoch": 0.09886077612785185,
      "grad_norm": 8.15428638458252,
      "learning_rate": 9.778016173411991e-05,
      "loss": 0.2203,
      "step": 2445
    },
    {
      "epoch": 0.09890120998311887,
      "grad_norm": 4.283844947814941,
      "learning_rate": 9.777828222826635e-05,
      "loss": 0.093,
      "step": 2446
    },
    {
      "epoch": 0.09894164383838588,
      "grad_norm": 4.030569076538086,
      "learning_rate": 9.777640194515303e-05,
      "loss": 0.1767,
      "step": 2447
    },
    {
      "epoch": 0.0989820776936529,
      "grad_norm": 8.366825103759766,
      "learning_rate": 9.777452088481052e-05,
      "loss": 0.2,
      "step": 2448
    },
    {
      "epoch": 0.09902251154891992,
      "grad_norm": 3.536785364151001,
      "learning_rate": 9.777263904726943e-05,
      "loss": 0.1883,
      "step": 2449
    },
    {
      "epoch": 0.09906294540418692,
      "grad_norm": 10.687562942504883,
      "learning_rate": 9.77707564325604e-05,
      "loss": 0.202,
      "step": 2450
    },
    {
      "epoch": 0.09910337925945394,
      "grad_norm": 5.098746299743652,
      "learning_rate": 9.776887304071401e-05,
      "loss": 0.2371,
      "step": 2451
    },
    {
      "epoch": 0.09914381311472095,
      "grad_norm": 8.345916748046875,
      "learning_rate": 9.776698887176094e-05,
      "loss": 0.2531,
      "step": 2452
    },
    {
      "epoch": 0.09918424696998797,
      "grad_norm": 4.950525283813477,
      "learning_rate": 9.776510392573181e-05,
      "loss": 0.2183,
      "step": 2453
    },
    {
      "epoch": 0.09922468082525499,
      "grad_norm": 2.885133981704712,
      "learning_rate": 9.77632182026573e-05,
      "loss": 0.1548,
      "step": 2454
    },
    {
      "epoch": 0.099265114680522,
      "grad_norm": 6.458444595336914,
      "learning_rate": 9.77613317025681e-05,
      "loss": 0.3088,
      "step": 2455
    },
    {
      "epoch": 0.09930554853578902,
      "grad_norm": 3.442085027694702,
      "learning_rate": 9.775944442549487e-05,
      "loss": 0.1963,
      "step": 2456
    },
    {
      "epoch": 0.09934598239105603,
      "grad_norm": 3.849655866622925,
      "learning_rate": 9.775755637146833e-05,
      "loss": 0.2634,
      "step": 2457
    },
    {
      "epoch": 0.09938641624632305,
      "grad_norm": 5.94687557220459,
      "learning_rate": 9.77556675405192e-05,
      "loss": 0.2145,
      "step": 2458
    },
    {
      "epoch": 0.09942685010159007,
      "grad_norm": 4.754313945770264,
      "learning_rate": 9.775377793267819e-05,
      "loss": 0.1409,
      "step": 2459
    },
    {
      "epoch": 0.09946728395685708,
      "grad_norm": 6.588354587554932,
      "learning_rate": 9.775188754797607e-05,
      "loss": 0.1796,
      "step": 2460
    },
    {
      "epoch": 0.09950771781212409,
      "grad_norm": 1.8200455904006958,
      "learning_rate": 9.774999638644356e-05,
      "loss": 0.1508,
      "step": 2461
    },
    {
      "epoch": 0.0995481516673911,
      "grad_norm": 9.260015487670898,
      "learning_rate": 9.774810444811143e-05,
      "loss": 0.2466,
      "step": 2462
    },
    {
      "epoch": 0.09958858552265812,
      "grad_norm": 4.322933197021484,
      "learning_rate": 9.774621173301049e-05,
      "loss": 0.1357,
      "step": 2463
    },
    {
      "epoch": 0.09962901937792513,
      "grad_norm": 3.3649911880493164,
      "learning_rate": 9.774431824117148e-05,
      "loss": 0.1475,
      "step": 2464
    },
    {
      "epoch": 0.09966945323319215,
      "grad_norm": 4.089843273162842,
      "learning_rate": 9.774242397262525e-05,
      "loss": 0.083,
      "step": 2465
    },
    {
      "epoch": 0.09970988708845917,
      "grad_norm": 3.2392704486846924,
      "learning_rate": 9.774052892740259e-05,
      "loss": 0.2142,
      "step": 2466
    },
    {
      "epoch": 0.09975032094372618,
      "grad_norm": 5.873871803283691,
      "learning_rate": 9.773863310553435e-05,
      "loss": 0.1257,
      "step": 2467
    },
    {
      "epoch": 0.0997907547989932,
      "grad_norm": 5.071615219116211,
      "learning_rate": 9.773673650705134e-05,
      "loss": 0.2606,
      "step": 2468
    },
    {
      "epoch": 0.09983118865426022,
      "grad_norm": 3.3344192504882812,
      "learning_rate": 9.773483913198445e-05,
      "loss": 0.1159,
      "step": 2469
    },
    {
      "epoch": 0.09987162250952723,
      "grad_norm": 5.9349212646484375,
      "learning_rate": 9.77329409803645e-05,
      "loss": 0.1306,
      "step": 2470
    },
    {
      "epoch": 0.09991205636479424,
      "grad_norm": 4.591471195220947,
      "learning_rate": 9.773104205222243e-05,
      "loss": 0.1707,
      "step": 2471
    },
    {
      "epoch": 0.09995249022006125,
      "grad_norm": 7.9459404945373535,
      "learning_rate": 9.772914234758908e-05,
      "loss": 0.1776,
      "step": 2472
    },
    {
      "epoch": 0.09999292407532827,
      "grad_norm": 3.891667127609253,
      "learning_rate": 9.772724186649537e-05,
      "loss": 0.1022,
      "step": 2473
    },
    {
      "epoch": 0.10003335793059528,
      "grad_norm": 3.172915458679199,
      "learning_rate": 9.772534060897222e-05,
      "loss": 0.1029,
      "step": 2474
    },
    {
      "epoch": 0.1000737917858623,
      "grad_norm": 8.28514289855957,
      "learning_rate": 9.772343857505057e-05,
      "loss": 0.1409,
      "step": 2475
    },
    {
      "epoch": 0.10011422564112932,
      "grad_norm": 3.2993650436401367,
      "learning_rate": 9.772153576476137e-05,
      "loss": 0.1828,
      "step": 2476
    },
    {
      "epoch": 0.10015465949639633,
      "grad_norm": 7.087630271911621,
      "learning_rate": 9.771963217813553e-05,
      "loss": 0.1395,
      "step": 2477
    },
    {
      "epoch": 0.10019509335166335,
      "grad_norm": 2.8749966621398926,
      "learning_rate": 9.771772781520405e-05,
      "loss": 0.1573,
      "step": 2478
    },
    {
      "epoch": 0.10023552720693037,
      "grad_norm": 4.505865097045898,
      "learning_rate": 9.771582267599793e-05,
      "loss": 0.1744,
      "step": 2479
    },
    {
      "epoch": 0.10027596106219738,
      "grad_norm": 5.275548458099365,
      "learning_rate": 9.771391676054811e-05,
      "loss": 0.2281,
      "step": 2480
    },
    {
      "epoch": 0.1003163949174644,
      "grad_norm": 6.282968044281006,
      "learning_rate": 9.771201006888564e-05,
      "loss": 0.1555,
      "step": 2481
    },
    {
      "epoch": 0.1003568287727314,
      "grad_norm": 4.892156600952148,
      "learning_rate": 9.771010260104151e-05,
      "loss": 0.1152,
      "step": 2482
    },
    {
      "epoch": 0.10039726262799842,
      "grad_norm": 6.0804443359375,
      "learning_rate": 9.770819435704677e-05,
      "loss": 0.1923,
      "step": 2483
    },
    {
      "epoch": 0.10043769648326543,
      "grad_norm": 4.544283866882324,
      "learning_rate": 9.770628533693246e-05,
      "loss": 0.1425,
      "step": 2484
    },
    {
      "epoch": 0.10047813033853245,
      "grad_norm": 6.655029773712158,
      "learning_rate": 9.770437554072962e-05,
      "loss": 0.1918,
      "step": 2485
    },
    {
      "epoch": 0.10051856419379947,
      "grad_norm": 5.840301513671875,
      "learning_rate": 9.770246496846936e-05,
      "loss": 0.1625,
      "step": 2486
    },
    {
      "epoch": 0.10055899804906648,
      "grad_norm": 9.146269798278809,
      "learning_rate": 9.770055362018271e-05,
      "loss": 0.3865,
      "step": 2487
    },
    {
      "epoch": 0.1005994319043335,
      "grad_norm": 2.1974895000457764,
      "learning_rate": 9.769864149590078e-05,
      "loss": 0.0968,
      "step": 2488
    },
    {
      "epoch": 0.10063986575960052,
      "grad_norm": 3.824338436126709,
      "learning_rate": 9.76967285956547e-05,
      "loss": 0.136,
      "step": 2489
    },
    {
      "epoch": 0.10068029961486753,
      "grad_norm": 3.1779847145080566,
      "learning_rate": 9.769481491947556e-05,
      "loss": 0.1138,
      "step": 2490
    },
    {
      "epoch": 0.10072073347013455,
      "grad_norm": 2.4270894527435303,
      "learning_rate": 9.769290046739452e-05,
      "loss": 0.0955,
      "step": 2491
    },
    {
      "epoch": 0.10076116732540157,
      "grad_norm": 3.3970696926116943,
      "learning_rate": 9.769098523944269e-05,
      "loss": 0.108,
      "step": 2492
    },
    {
      "epoch": 0.10080160118066857,
      "grad_norm": 3.540525436401367,
      "learning_rate": 9.768906923565126e-05,
      "loss": 0.1301,
      "step": 2493
    },
    {
      "epoch": 0.10084203503593558,
      "grad_norm": 3.485556125640869,
      "learning_rate": 9.768715245605138e-05,
      "loss": 0.128,
      "step": 2494
    },
    {
      "epoch": 0.1008824688912026,
      "grad_norm": 7.347784996032715,
      "learning_rate": 9.768523490067424e-05,
      "loss": 0.1604,
      "step": 2495
    },
    {
      "epoch": 0.10092290274646962,
      "grad_norm": 4.8804240226745605,
      "learning_rate": 9.768331656955103e-05,
      "loss": 0.1478,
      "step": 2496
    },
    {
      "epoch": 0.10096333660173663,
      "grad_norm": 4.804251670837402,
      "learning_rate": 9.768139746271296e-05,
      "loss": 0.1321,
      "step": 2497
    },
    {
      "epoch": 0.10100377045700365,
      "grad_norm": 5.65533971786499,
      "learning_rate": 9.767947758019125e-05,
      "loss": 0.1959,
      "step": 2498
    },
    {
      "epoch": 0.10104420431227067,
      "grad_norm": 4.901950359344482,
      "learning_rate": 9.767755692201714e-05,
      "loss": 0.1663,
      "step": 2499
    },
    {
      "epoch": 0.10108463816753768,
      "grad_norm": 3.760631561279297,
      "learning_rate": 9.767563548822185e-05,
      "loss": 0.1908,
      "step": 2500
    },
    {
      "epoch": 0.1011250720228047,
      "grad_norm": 8.324453353881836,
      "learning_rate": 9.767371327883668e-05,
      "loss": 0.2475,
      "step": 2501
    },
    {
      "epoch": 0.10116550587807172,
      "grad_norm": 5.723111629486084,
      "learning_rate": 9.767179029389286e-05,
      "loss": 0.2451,
      "step": 2502
    },
    {
      "epoch": 0.10120593973333872,
      "grad_norm": 11.699164390563965,
      "learning_rate": 9.76698665334217e-05,
      "loss": 0.4002,
      "step": 2503
    },
    {
      "epoch": 0.10124637358860573,
      "grad_norm": 2.534505844116211,
      "learning_rate": 9.766794199745449e-05,
      "loss": 0.1095,
      "step": 2504
    },
    {
      "epoch": 0.10128680744387275,
      "grad_norm": 7.547207355499268,
      "learning_rate": 9.766601668602251e-05,
      "loss": 0.1754,
      "step": 2505
    },
    {
      "epoch": 0.10132724129913977,
      "grad_norm": 11.498230934143066,
      "learning_rate": 9.766409059915712e-05,
      "loss": 0.2364,
      "step": 2506
    },
    {
      "epoch": 0.10136767515440678,
      "grad_norm": 4.574555397033691,
      "learning_rate": 9.766216373688965e-05,
      "loss": 0.3948,
      "step": 2507
    },
    {
      "epoch": 0.1014081090096738,
      "grad_norm": 6.651365280151367,
      "learning_rate": 9.76602360992514e-05,
      "loss": 0.1504,
      "step": 2508
    },
    {
      "epoch": 0.10144854286494082,
      "grad_norm": 7.506748199462891,
      "learning_rate": 9.76583076862738e-05,
      "loss": 0.1802,
      "step": 2509
    },
    {
      "epoch": 0.10148897672020783,
      "grad_norm": 4.7307844161987305,
      "learning_rate": 9.765637849798819e-05,
      "loss": 0.1929,
      "step": 2510
    },
    {
      "epoch": 0.10152941057547485,
      "grad_norm": 5.459943771362305,
      "learning_rate": 9.765444853442593e-05,
      "loss": 0.1234,
      "step": 2511
    },
    {
      "epoch": 0.10156984443074187,
      "grad_norm": 3.9879961013793945,
      "learning_rate": 9.765251779561844e-05,
      "loss": 0.1324,
      "step": 2512
    },
    {
      "epoch": 0.10161027828600888,
      "grad_norm": 6.194891452789307,
      "learning_rate": 9.765058628159714e-05,
      "loss": 0.2509,
      "step": 2513
    },
    {
      "epoch": 0.10165071214127588,
      "grad_norm": 6.467249393463135,
      "learning_rate": 9.764865399239343e-05,
      "loss": 0.1748,
      "step": 2514
    },
    {
      "epoch": 0.1016911459965429,
      "grad_norm": 12.116089820861816,
      "learning_rate": 9.764672092803875e-05,
      "loss": 0.3745,
      "step": 2515
    },
    {
      "epoch": 0.10173157985180992,
      "grad_norm": 3.2278056144714355,
      "learning_rate": 9.764478708856453e-05,
      "loss": 0.2162,
      "step": 2516
    },
    {
      "epoch": 0.10177201370707693,
      "grad_norm": 7.449737071990967,
      "learning_rate": 9.764285247400228e-05,
      "loss": 0.2414,
      "step": 2517
    },
    {
      "epoch": 0.10181244756234395,
      "grad_norm": 5.789047718048096,
      "learning_rate": 9.764091708438343e-05,
      "loss": 0.2006,
      "step": 2518
    },
    {
      "epoch": 0.10185288141761097,
      "grad_norm": 8.225090980529785,
      "learning_rate": 9.763898091973948e-05,
      "loss": 0.1984,
      "step": 2519
    },
    {
      "epoch": 0.10189331527287798,
      "grad_norm": 5.858513832092285,
      "learning_rate": 9.763704398010192e-05,
      "loss": 0.212,
      "step": 2520
    },
    {
      "epoch": 0.101933749128145,
      "grad_norm": 5.700589656829834,
      "learning_rate": 9.763510626550228e-05,
      "loss": 0.1647,
      "step": 2521
    },
    {
      "epoch": 0.10197418298341202,
      "grad_norm": 6.944118499755859,
      "learning_rate": 9.763316777597205e-05,
      "loss": 0.1529,
      "step": 2522
    },
    {
      "epoch": 0.10201461683867903,
      "grad_norm": 6.806957721710205,
      "learning_rate": 9.763122851154279e-05,
      "loss": 0.2315,
      "step": 2523
    },
    {
      "epoch": 0.10205505069394605,
      "grad_norm": 4.808497905731201,
      "learning_rate": 9.762928847224603e-05,
      "loss": 0.1869,
      "step": 2524
    },
    {
      "epoch": 0.10209548454921305,
      "grad_norm": 3.917487144470215,
      "learning_rate": 9.762734765811336e-05,
      "loss": 0.139,
      "step": 2525
    },
    {
      "epoch": 0.10213591840448007,
      "grad_norm": 4.561970233917236,
      "learning_rate": 9.762540606917635e-05,
      "loss": 0.2009,
      "step": 2526
    },
    {
      "epoch": 0.10217635225974708,
      "grad_norm": 4.424883842468262,
      "learning_rate": 9.762346370546653e-05,
      "loss": 0.2007,
      "step": 2527
    },
    {
      "epoch": 0.1022167861150141,
      "grad_norm": 3.584731101989746,
      "learning_rate": 9.762152056701557e-05,
      "loss": 0.2781,
      "step": 2528
    },
    {
      "epoch": 0.10225721997028112,
      "grad_norm": 8.954307556152344,
      "learning_rate": 9.761957665385504e-05,
      "loss": 0.2325,
      "step": 2529
    },
    {
      "epoch": 0.10229765382554813,
      "grad_norm": 6.497585773468018,
      "learning_rate": 9.761763196601657e-05,
      "loss": 0.2666,
      "step": 2530
    },
    {
      "epoch": 0.10233808768081515,
      "grad_norm": 2.843245506286621,
      "learning_rate": 9.761568650353182e-05,
      "loss": 0.1886,
      "step": 2531
    },
    {
      "epoch": 0.10237852153608216,
      "grad_norm": 4.022202968597412,
      "learning_rate": 9.76137402664324e-05,
      "loss": 0.1199,
      "step": 2532
    },
    {
      "epoch": 0.10241895539134918,
      "grad_norm": 3.432631731033325,
      "learning_rate": 9.761179325475e-05,
      "loss": 0.1677,
      "step": 2533
    },
    {
      "epoch": 0.1024593892466162,
      "grad_norm": 7.130154132843018,
      "learning_rate": 9.760984546851628e-05,
      "loss": 0.1963,
      "step": 2534
    },
    {
      "epoch": 0.1024998231018832,
      "grad_norm": 8.010909080505371,
      "learning_rate": 9.760789690776293e-05,
      "loss": 0.1553,
      "step": 2535
    },
    {
      "epoch": 0.10254025695715022,
      "grad_norm": 3.1638104915618896,
      "learning_rate": 9.760594757252167e-05,
      "loss": 0.3099,
      "step": 2536
    },
    {
      "epoch": 0.10258069081241723,
      "grad_norm": 3.687932252883911,
      "learning_rate": 9.760399746282418e-05,
      "loss": 0.1665,
      "step": 2537
    },
    {
      "epoch": 0.10262112466768425,
      "grad_norm": 6.876625061035156,
      "learning_rate": 9.760204657870218e-05,
      "loss": 0.1454,
      "step": 2538
    },
    {
      "epoch": 0.10266155852295127,
      "grad_norm": 6.954944133758545,
      "learning_rate": 9.760009492018745e-05,
      "loss": 0.216,
      "step": 2539
    },
    {
      "epoch": 0.10270199237821828,
      "grad_norm": 4.816905498504639,
      "learning_rate": 9.75981424873117e-05,
      "loss": 0.1664,
      "step": 2540
    },
    {
      "epoch": 0.1027424262334853,
      "grad_norm": 6.038915157318115,
      "learning_rate": 9.759618928010672e-05,
      "loss": 0.2699,
      "step": 2541
    },
    {
      "epoch": 0.10278286008875231,
      "grad_norm": 7.990200996398926,
      "learning_rate": 9.759423529860425e-05,
      "loss": 0.1707,
      "step": 2542
    },
    {
      "epoch": 0.10282329394401933,
      "grad_norm": 4.619919776916504,
      "learning_rate": 9.759228054283612e-05,
      "loss": 0.1139,
      "step": 2543
    },
    {
      "epoch": 0.10286372779928635,
      "grad_norm": 5.433123588562012,
      "learning_rate": 9.75903250128341e-05,
      "loss": 0.1466,
      "step": 2544
    },
    {
      "epoch": 0.10290416165455336,
      "grad_norm": 15.167848587036133,
      "learning_rate": 9.758836870863002e-05,
      "loss": 0.3061,
      "step": 2545
    },
    {
      "epoch": 0.10294459550982037,
      "grad_norm": 3.3776862621307373,
      "learning_rate": 9.758641163025569e-05,
      "loss": 0.1995,
      "step": 2546
    },
    {
      "epoch": 0.10298502936508738,
      "grad_norm": 3.598172903060913,
      "learning_rate": 9.758445377774297e-05,
      "loss": 0.0684,
      "step": 2547
    },
    {
      "epoch": 0.1030254632203544,
      "grad_norm": 9.661309242248535,
      "learning_rate": 9.758249515112368e-05,
      "loss": 0.203,
      "step": 2548
    },
    {
      "epoch": 0.10306589707562142,
      "grad_norm": 9.811541557312012,
      "learning_rate": 9.75805357504297e-05,
      "loss": 0.2533,
      "step": 2549
    },
    {
      "epoch": 0.10310633093088843,
      "grad_norm": 8.600652694702148,
      "learning_rate": 9.75785755756929e-05,
      "loss": 0.2372,
      "step": 2550
    },
    {
      "epoch": 0.10314676478615545,
      "grad_norm": 7.904280662536621,
      "learning_rate": 9.757661462694517e-05,
      "loss": 0.2045,
      "step": 2551
    },
    {
      "epoch": 0.10318719864142246,
      "grad_norm": 3.6131582260131836,
      "learning_rate": 9.757465290421843e-05,
      "loss": 0.1504,
      "step": 2552
    },
    {
      "epoch": 0.10322763249668948,
      "grad_norm": 8.17551040649414,
      "learning_rate": 9.757269040754456e-05,
      "loss": 0.1741,
      "step": 2553
    },
    {
      "epoch": 0.1032680663519565,
      "grad_norm": 3.899264097213745,
      "learning_rate": 9.757072713695552e-05,
      "loss": 0.1362,
      "step": 2554
    },
    {
      "epoch": 0.10330850020722351,
      "grad_norm": 5.639105796813965,
      "learning_rate": 9.756876309248322e-05,
      "loss": 0.2148,
      "step": 2555
    },
    {
      "epoch": 0.10334893406249053,
      "grad_norm": 3.2982025146484375,
      "learning_rate": 9.756679827415962e-05,
      "loss": 0.173,
      "step": 2556
    },
    {
      "epoch": 0.10338936791775753,
      "grad_norm": 3.4742422103881836,
      "learning_rate": 9.756483268201668e-05,
      "loss": 0.209,
      "step": 2557
    },
    {
      "epoch": 0.10342980177302455,
      "grad_norm": 5.409919261932373,
      "learning_rate": 9.75628663160864e-05,
      "loss": 0.117,
      "step": 2558
    },
    {
      "epoch": 0.10347023562829157,
      "grad_norm": 12.965363502502441,
      "learning_rate": 9.756089917640074e-05,
      "loss": 0.3244,
      "step": 2559
    },
    {
      "epoch": 0.10351066948355858,
      "grad_norm": 3.46207332611084,
      "learning_rate": 9.755893126299172e-05,
      "loss": 0.1689,
      "step": 2560
    },
    {
      "epoch": 0.1035511033388256,
      "grad_norm": 2.05094575881958,
      "learning_rate": 9.755696257589134e-05,
      "loss": 0.1317,
      "step": 2561
    },
    {
      "epoch": 0.10359153719409261,
      "grad_norm": 10.600028038024902,
      "learning_rate": 9.755499311513163e-05,
      "loss": 0.3098,
      "step": 2562
    },
    {
      "epoch": 0.10363197104935963,
      "grad_norm": 3.8559842109680176,
      "learning_rate": 9.755302288074465e-05,
      "loss": 0.2773,
      "step": 2563
    },
    {
      "epoch": 0.10367240490462665,
      "grad_norm": 5.687427520751953,
      "learning_rate": 9.755105187276243e-05,
      "loss": 0.1922,
      "step": 2564
    },
    {
      "epoch": 0.10371283875989366,
      "grad_norm": 3.554979085922241,
      "learning_rate": 9.754908009121704e-05,
      "loss": 0.142,
      "step": 2565
    },
    {
      "epoch": 0.10375327261516068,
      "grad_norm": 3.9039525985717773,
      "learning_rate": 9.754710753614053e-05,
      "loss": 0.1101,
      "step": 2566
    },
    {
      "epoch": 0.10379370647042768,
      "grad_norm": 5.692020893096924,
      "learning_rate": 9.754513420756505e-05,
      "loss": 0.1942,
      "step": 2567
    },
    {
      "epoch": 0.1038341403256947,
      "grad_norm": 4.613198757171631,
      "learning_rate": 9.754316010552266e-05,
      "loss": 0.2561,
      "step": 2568
    },
    {
      "epoch": 0.10387457418096172,
      "grad_norm": 2.0203208923339844,
      "learning_rate": 9.754118523004547e-05,
      "loss": 0.1248,
      "step": 2569
    },
    {
      "epoch": 0.10391500803622873,
      "grad_norm": 4.539797782897949,
      "learning_rate": 9.753920958116563e-05,
      "loss": 0.1351,
      "step": 2570
    },
    {
      "epoch": 0.10395544189149575,
      "grad_norm": 6.297520160675049,
      "learning_rate": 9.753723315891528e-05,
      "loss": 0.1446,
      "step": 2571
    },
    {
      "epoch": 0.10399587574676276,
      "grad_norm": 8.990057945251465,
      "learning_rate": 9.753525596332655e-05,
      "loss": 0.3219,
      "step": 2572
    },
    {
      "epoch": 0.10403630960202978,
      "grad_norm": 4.516120433807373,
      "learning_rate": 9.753327799443162e-05,
      "loss": 0.3151,
      "step": 2573
    },
    {
      "epoch": 0.1040767434572968,
      "grad_norm": 1.7801768779754639,
      "learning_rate": 9.753129925226267e-05,
      "loss": 0.1212,
      "step": 2574
    },
    {
      "epoch": 0.10411717731256381,
      "grad_norm": 4.906913757324219,
      "learning_rate": 9.752931973685188e-05,
      "loss": 0.2197,
      "step": 2575
    },
    {
      "epoch": 0.10415761116783083,
      "grad_norm": 2.8843557834625244,
      "learning_rate": 9.752733944823148e-05,
      "loss": 0.1664,
      "step": 2576
    },
    {
      "epoch": 0.10419804502309785,
      "grad_norm": 4.998475551605225,
      "learning_rate": 9.752535838643365e-05,
      "loss": 0.2623,
      "step": 2577
    },
    {
      "epoch": 0.10423847887836485,
      "grad_norm": 7.933319568634033,
      "learning_rate": 9.752337655149063e-05,
      "loss": 0.1384,
      "step": 2578
    },
    {
      "epoch": 0.10427891273363186,
      "grad_norm": 2.314054489135742,
      "learning_rate": 9.752139394343465e-05,
      "loss": 0.1032,
      "step": 2579
    },
    {
      "epoch": 0.10431934658889888,
      "grad_norm": 4.564206600189209,
      "learning_rate": 9.751941056229798e-05,
      "loss": 0.1448,
      "step": 2580
    },
    {
      "epoch": 0.1043597804441659,
      "grad_norm": 3.409414291381836,
      "learning_rate": 9.75174264081129e-05,
      "loss": 0.2387,
      "step": 2581
    },
    {
      "epoch": 0.10440021429943291,
      "grad_norm": 8.89101505279541,
      "learning_rate": 9.751544148091166e-05,
      "loss": 0.2414,
      "step": 2582
    },
    {
      "epoch": 0.10444064815469993,
      "grad_norm": 8.945465087890625,
      "learning_rate": 9.751345578072657e-05,
      "loss": 0.2245,
      "step": 2583
    },
    {
      "epoch": 0.10448108200996695,
      "grad_norm": 6.268547534942627,
      "learning_rate": 9.751146930758992e-05,
      "loss": 0.1258,
      "step": 2584
    },
    {
      "epoch": 0.10452151586523396,
      "grad_norm": 6.661411285400391,
      "learning_rate": 9.750948206153403e-05,
      "loss": 0.1935,
      "step": 2585
    },
    {
      "epoch": 0.10456194972050098,
      "grad_norm": 2.6590983867645264,
      "learning_rate": 9.750749404259122e-05,
      "loss": 0.1255,
      "step": 2586
    },
    {
      "epoch": 0.104602383575768,
      "grad_norm": 2.3793978691101074,
      "learning_rate": 9.750550525079385e-05,
      "loss": 0.1851,
      "step": 2587
    },
    {
      "epoch": 0.10464281743103501,
      "grad_norm": 3.544565200805664,
      "learning_rate": 9.750351568617427e-05,
      "loss": 0.2373,
      "step": 2588
    },
    {
      "epoch": 0.10468325128630201,
      "grad_norm": 4.042886734008789,
      "learning_rate": 9.750152534876483e-05,
      "loss": 0.1816,
      "step": 2589
    },
    {
      "epoch": 0.10472368514156903,
      "grad_norm": 4.757035732269287,
      "learning_rate": 9.749953423859792e-05,
      "loss": 0.1838,
      "step": 2590
    },
    {
      "epoch": 0.10476411899683605,
      "grad_norm": 5.477441310882568,
      "learning_rate": 9.749754235570594e-05,
      "loss": 0.1754,
      "step": 2591
    },
    {
      "epoch": 0.10480455285210306,
      "grad_norm": 2.837709665298462,
      "learning_rate": 9.749554970012128e-05,
      "loss": 0.1629,
      "step": 2592
    },
    {
      "epoch": 0.10484498670737008,
      "grad_norm": 8.560015678405762,
      "learning_rate": 9.749355627187636e-05,
      "loss": 0.2376,
      "step": 2593
    },
    {
      "epoch": 0.1048854205626371,
      "grad_norm": 6.735232353210449,
      "learning_rate": 9.749156207100361e-05,
      "loss": 0.1789,
      "step": 2594
    },
    {
      "epoch": 0.10492585441790411,
      "grad_norm": 5.727558135986328,
      "learning_rate": 9.748956709753548e-05,
      "loss": 0.1638,
      "step": 2595
    },
    {
      "epoch": 0.10496628827317113,
      "grad_norm": 2.2542691230773926,
      "learning_rate": 9.748757135150441e-05,
      "loss": 0.0651,
      "step": 2596
    },
    {
      "epoch": 0.10500672212843815,
      "grad_norm": 7.2391743659973145,
      "learning_rate": 9.748557483294289e-05,
      "loss": 0.249,
      "step": 2597
    },
    {
      "epoch": 0.10504715598370516,
      "grad_norm": 4.413931846618652,
      "learning_rate": 9.748357754188338e-05,
      "loss": 0.1303,
      "step": 2598
    },
    {
      "epoch": 0.10508758983897216,
      "grad_norm": 3.348846912384033,
      "learning_rate": 9.748157947835836e-05,
      "loss": 0.0541,
      "step": 2599
    },
    {
      "epoch": 0.10512802369423918,
      "grad_norm": 4.107471942901611,
      "learning_rate": 9.747958064240036e-05,
      "loss": 0.1738,
      "step": 2600
    },
    {
      "epoch": 0.1051684575495062,
      "grad_norm": 5.994881629943848,
      "learning_rate": 9.747758103404187e-05,
      "loss": 0.2161,
      "step": 2601
    },
    {
      "epoch": 0.10520889140477321,
      "grad_norm": 9.327596664428711,
      "learning_rate": 9.747558065331547e-05,
      "loss": 0.2078,
      "step": 2602
    },
    {
      "epoch": 0.10524932526004023,
      "grad_norm": 4.450102806091309,
      "learning_rate": 9.747357950025364e-05,
      "loss": 0.1162,
      "step": 2603
    },
    {
      "epoch": 0.10528975911530725,
      "grad_norm": 9.443557739257812,
      "learning_rate": 9.747157757488899e-05,
      "loss": 0.3205,
      "step": 2604
    },
    {
      "epoch": 0.10533019297057426,
      "grad_norm": 9.321355819702148,
      "learning_rate": 9.746957487725405e-05,
      "loss": 0.3228,
      "step": 2605
    },
    {
      "epoch": 0.10537062682584128,
      "grad_norm": 4.556854724884033,
      "learning_rate": 9.74675714073814e-05,
      "loss": 0.1375,
      "step": 2606
    },
    {
      "epoch": 0.1054110606811083,
      "grad_norm": 6.583793640136719,
      "learning_rate": 9.746556716530366e-05,
      "loss": 0.2026,
      "step": 2607
    },
    {
      "epoch": 0.10545149453637531,
      "grad_norm": 8.31285285949707,
      "learning_rate": 9.746356215105341e-05,
      "loss": 0.3478,
      "step": 2608
    },
    {
      "epoch": 0.10549192839164233,
      "grad_norm": 7.38174295425415,
      "learning_rate": 9.746155636466329e-05,
      "loss": 0.2631,
      "step": 2609
    },
    {
      "epoch": 0.10553236224690933,
      "grad_norm": 11.29359245300293,
      "learning_rate": 9.74595498061659e-05,
      "loss": 0.2973,
      "step": 2610
    },
    {
      "epoch": 0.10557279610217635,
      "grad_norm": 4.096506118774414,
      "learning_rate": 9.74575424755939e-05,
      "loss": 0.1724,
      "step": 2611
    },
    {
      "epoch": 0.10561322995744336,
      "grad_norm": 3.680166244506836,
      "learning_rate": 9.745553437297996e-05,
      "loss": 0.2394,
      "step": 2612
    },
    {
      "epoch": 0.10565366381271038,
      "grad_norm": 5.235030651092529,
      "learning_rate": 9.745352549835673e-05,
      "loss": 0.1306,
      "step": 2613
    },
    {
      "epoch": 0.1056940976679774,
      "grad_norm": 9.476926803588867,
      "learning_rate": 9.745151585175689e-05,
      "loss": 0.3702,
      "step": 2614
    },
    {
      "epoch": 0.10573453152324441,
      "grad_norm": 2.8062472343444824,
      "learning_rate": 9.744950543321315e-05,
      "loss": 0.2559,
      "step": 2615
    },
    {
      "epoch": 0.10577496537851143,
      "grad_norm": 4.972790718078613,
      "learning_rate": 9.744749424275819e-05,
      "loss": 0.1325,
      "step": 2616
    },
    {
      "epoch": 0.10581539923377845,
      "grad_norm": 6.9773430824279785,
      "learning_rate": 9.744548228042474e-05,
      "loss": 0.1538,
      "step": 2617
    },
    {
      "epoch": 0.10585583308904546,
      "grad_norm": 3.244921922683716,
      "learning_rate": 9.744346954624554e-05,
      "loss": 0.1735,
      "step": 2618
    },
    {
      "epoch": 0.10589626694431248,
      "grad_norm": 4.883279800415039,
      "learning_rate": 9.744145604025332e-05,
      "loss": 0.3768,
      "step": 2619
    },
    {
      "epoch": 0.1059367007995795,
      "grad_norm": 3.5831642150878906,
      "learning_rate": 9.743944176248083e-05,
      "loss": 0.1056,
      "step": 2620
    },
    {
      "epoch": 0.1059771346548465,
      "grad_norm": 2.673980474472046,
      "learning_rate": 9.743742671296086e-05,
      "loss": 0.0757,
      "step": 2621
    },
    {
      "epoch": 0.10601756851011351,
      "grad_norm": 4.598702907562256,
      "learning_rate": 9.743541089172619e-05,
      "loss": 0.184,
      "step": 2622
    },
    {
      "epoch": 0.10605800236538053,
      "grad_norm": 6.9495439529418945,
      "learning_rate": 9.743339429880959e-05,
      "loss": 0.1981,
      "step": 2623
    },
    {
      "epoch": 0.10609843622064755,
      "grad_norm": 6.059831619262695,
      "learning_rate": 9.743137693424388e-05,
      "loss": 0.1613,
      "step": 2624
    },
    {
      "epoch": 0.10613887007591456,
      "grad_norm": 7.759637832641602,
      "learning_rate": 9.742935879806188e-05,
      "loss": 0.2149,
      "step": 2625
    },
    {
      "epoch": 0.10617930393118158,
      "grad_norm": 5.176877975463867,
      "learning_rate": 9.742733989029642e-05,
      "loss": 0.1673,
      "step": 2626
    },
    {
      "epoch": 0.1062197377864486,
      "grad_norm": 3.517822742462158,
      "learning_rate": 9.742532021098034e-05,
      "loss": 0.1188,
      "step": 2627
    },
    {
      "epoch": 0.10626017164171561,
      "grad_norm": 7.5783162117004395,
      "learning_rate": 9.74232997601465e-05,
      "loss": 0.3091,
      "step": 2628
    },
    {
      "epoch": 0.10630060549698263,
      "grad_norm": 3.8475637435913086,
      "learning_rate": 9.742127853782776e-05,
      "loss": 0.228,
      "step": 2629
    },
    {
      "epoch": 0.10634103935224964,
      "grad_norm": 5.954957008361816,
      "learning_rate": 9.741925654405703e-05,
      "loss": 0.2003,
      "step": 2630
    },
    {
      "epoch": 0.10638147320751665,
      "grad_norm": 3.682616710662842,
      "learning_rate": 9.741723377886717e-05,
      "loss": 0.1523,
      "step": 2631
    },
    {
      "epoch": 0.10642190706278366,
      "grad_norm": 7.642892360687256,
      "learning_rate": 9.74152102422911e-05,
      "loss": 0.1892,
      "step": 2632
    },
    {
      "epoch": 0.10646234091805068,
      "grad_norm": 6.745631694793701,
      "learning_rate": 9.741318593436174e-05,
      "loss": 0.2081,
      "step": 2633
    },
    {
      "epoch": 0.1065027747733177,
      "grad_norm": 5.4782280921936035,
      "learning_rate": 9.741116085511202e-05,
      "loss": 0.2708,
      "step": 2634
    },
    {
      "epoch": 0.10654320862858471,
      "grad_norm": 6.153581619262695,
      "learning_rate": 9.740913500457489e-05,
      "loss": 0.1855,
      "step": 2635
    },
    {
      "epoch": 0.10658364248385173,
      "grad_norm": 3.1392714977264404,
      "learning_rate": 9.740710838278328e-05,
      "loss": 0.1564,
      "step": 2636
    },
    {
      "epoch": 0.10662407633911875,
      "grad_norm": 2.9353301525115967,
      "learning_rate": 9.740508098977021e-05,
      "loss": 0.1249,
      "step": 2637
    },
    {
      "epoch": 0.10666451019438576,
      "grad_norm": 2.9880435466766357,
      "learning_rate": 9.740305282556862e-05,
      "loss": 0.2419,
      "step": 2638
    },
    {
      "epoch": 0.10670494404965278,
      "grad_norm": 1.8715635538101196,
      "learning_rate": 9.74010238902115e-05,
      "loss": 0.2404,
      "step": 2639
    },
    {
      "epoch": 0.1067453779049198,
      "grad_norm": 4.032928943634033,
      "learning_rate": 9.739899418373191e-05,
      "loss": 0.2471,
      "step": 2640
    },
    {
      "epoch": 0.10678581176018681,
      "grad_norm": 8.548482894897461,
      "learning_rate": 9.73969637061628e-05,
      "loss": 0.3243,
      "step": 2641
    },
    {
      "epoch": 0.10682624561545381,
      "grad_norm": 8.329529762268066,
      "learning_rate": 9.739493245753725e-05,
      "loss": 0.2439,
      "step": 2642
    },
    {
      "epoch": 0.10686667947072083,
      "grad_norm": 3.4948480129241943,
      "learning_rate": 9.739290043788828e-05,
      "loss": 0.1048,
      "step": 2643
    },
    {
      "epoch": 0.10690711332598785,
      "grad_norm": 7.83454704284668,
      "learning_rate": 9.739086764724896e-05,
      "loss": 0.2579,
      "step": 2644
    },
    {
      "epoch": 0.10694754718125486,
      "grad_norm": 5.200916290283203,
      "learning_rate": 9.738883408565236e-05,
      "loss": 0.1851,
      "step": 2645
    },
    {
      "epoch": 0.10698798103652188,
      "grad_norm": 5.6606597900390625,
      "learning_rate": 9.738679975313154e-05,
      "loss": 0.1372,
      "step": 2646
    },
    {
      "epoch": 0.1070284148917889,
      "grad_norm": 4.007655620574951,
      "learning_rate": 9.738476464971963e-05,
      "loss": 0.2171,
      "step": 2647
    },
    {
      "epoch": 0.10706884874705591,
      "grad_norm": 2.8583180904388428,
      "learning_rate": 9.738272877544972e-05,
      "loss": 0.073,
      "step": 2648
    },
    {
      "epoch": 0.10710928260232293,
      "grad_norm": 5.274252414703369,
      "learning_rate": 9.738069213035492e-05,
      "loss": 0.1734,
      "step": 2649
    },
    {
      "epoch": 0.10714971645758994,
      "grad_norm": 5.548259258270264,
      "learning_rate": 9.737865471446836e-05,
      "loss": 0.1858,
      "step": 2650
    },
    {
      "epoch": 0.10719015031285696,
      "grad_norm": 4.957038879394531,
      "learning_rate": 9.737661652782322e-05,
      "loss": 0.2133,
      "step": 2651
    },
    {
      "epoch": 0.10723058416812398,
      "grad_norm": 2.9975955486297607,
      "learning_rate": 9.737457757045263e-05,
      "loss": 0.1835,
      "step": 2652
    },
    {
      "epoch": 0.10727101802339098,
      "grad_norm": 3.912949800491333,
      "learning_rate": 9.737253784238975e-05,
      "loss": 0.12,
      "step": 2653
    },
    {
      "epoch": 0.107311451878658,
      "grad_norm": 4.06726598739624,
      "learning_rate": 9.737049734366779e-05,
      "loss": 0.191,
      "step": 2654
    },
    {
      "epoch": 0.10735188573392501,
      "grad_norm": 3.494539976119995,
      "learning_rate": 9.736845607431992e-05,
      "loss": 0.1338,
      "step": 2655
    },
    {
      "epoch": 0.10739231958919203,
      "grad_norm": 3.1799476146698,
      "learning_rate": 9.736641403437936e-05,
      "loss": 0.1771,
      "step": 2656
    },
    {
      "epoch": 0.10743275344445904,
      "grad_norm": 2.293144464492798,
      "learning_rate": 9.736437122387933e-05,
      "loss": 0.1267,
      "step": 2657
    },
    {
      "epoch": 0.10747318729972606,
      "grad_norm": 6.9902544021606445,
      "learning_rate": 9.736232764285305e-05,
      "loss": 0.183,
      "step": 2658
    },
    {
      "epoch": 0.10751362115499308,
      "grad_norm": 9.211111068725586,
      "learning_rate": 9.73602832913338e-05,
      "loss": 0.1976,
      "step": 2659
    },
    {
      "epoch": 0.1075540550102601,
      "grad_norm": 2.938033103942871,
      "learning_rate": 9.735823816935479e-05,
      "loss": 0.1057,
      "step": 2660
    },
    {
      "epoch": 0.10759448886552711,
      "grad_norm": 2.7003142833709717,
      "learning_rate": 9.735619227694932e-05,
      "loss": 0.1381,
      "step": 2661
    },
    {
      "epoch": 0.10763492272079413,
      "grad_norm": 1.4364328384399414,
      "learning_rate": 9.735414561415068e-05,
      "loss": 0.0424,
      "step": 2662
    },
    {
      "epoch": 0.10767535657606113,
      "grad_norm": 7.795121192932129,
      "learning_rate": 9.735209818099213e-05,
      "loss": 0.3413,
      "step": 2663
    },
    {
      "epoch": 0.10771579043132815,
      "grad_norm": 2.788208484649658,
      "learning_rate": 9.735004997750702e-05,
      "loss": 0.3104,
      "step": 2664
    },
    {
      "epoch": 0.10775622428659516,
      "grad_norm": 5.234107494354248,
      "learning_rate": 9.734800100372864e-05,
      "loss": 0.1955,
      "step": 2665
    },
    {
      "epoch": 0.10779665814186218,
      "grad_norm": 9.522753715515137,
      "learning_rate": 9.734595125969033e-05,
      "loss": 0.307,
      "step": 2666
    },
    {
      "epoch": 0.1078370919971292,
      "grad_norm": 8.522855758666992,
      "learning_rate": 9.734390074542544e-05,
      "loss": 0.2987,
      "step": 2667
    },
    {
      "epoch": 0.10787752585239621,
      "grad_norm": 9.226008415222168,
      "learning_rate": 9.734184946096732e-05,
      "loss": 0.164,
      "step": 2668
    },
    {
      "epoch": 0.10791795970766323,
      "grad_norm": 4.560689449310303,
      "learning_rate": 9.733979740634935e-05,
      "loss": 0.3146,
      "step": 2669
    },
    {
      "epoch": 0.10795839356293024,
      "grad_norm": 4.631193161010742,
      "learning_rate": 9.733774458160492e-05,
      "loss": 0.1172,
      "step": 2670
    },
    {
      "epoch": 0.10799882741819726,
      "grad_norm": 10.885786056518555,
      "learning_rate": 9.73356909867674e-05,
      "loss": 0.476,
      "step": 2671
    },
    {
      "epoch": 0.10803926127346428,
      "grad_norm": 6.305382251739502,
      "learning_rate": 9.733363662187022e-05,
      "loss": 0.1872,
      "step": 2672
    },
    {
      "epoch": 0.10807969512873129,
      "grad_norm": 4.7866363525390625,
      "learning_rate": 9.733158148694679e-05,
      "loss": 0.2623,
      "step": 2673
    },
    {
      "epoch": 0.1081201289839983,
      "grad_norm": 3.0159430503845215,
      "learning_rate": 9.732952558203055e-05,
      "loss": 0.1493,
      "step": 2674
    },
    {
      "epoch": 0.10816056283926531,
      "grad_norm": 6.093318462371826,
      "learning_rate": 9.732746890715494e-05,
      "loss": 0.1673,
      "step": 2675
    },
    {
      "epoch": 0.10820099669453233,
      "grad_norm": 2.653470993041992,
      "learning_rate": 9.732541146235341e-05,
      "loss": 0.221,
      "step": 2676
    },
    {
      "epoch": 0.10824143054979934,
      "grad_norm": 4.70076322555542,
      "learning_rate": 9.732335324765944e-05,
      "loss": 0.1217,
      "step": 2677
    },
    {
      "epoch": 0.10828186440506636,
      "grad_norm": 4.704703330993652,
      "learning_rate": 9.732129426310653e-05,
      "loss": 0.1468,
      "step": 2678
    },
    {
      "epoch": 0.10832229826033338,
      "grad_norm": 4.209667205810547,
      "learning_rate": 9.731923450872816e-05,
      "loss": 0.1468,
      "step": 2679
    },
    {
      "epoch": 0.1083627321156004,
      "grad_norm": 6.749548435211182,
      "learning_rate": 9.731717398455782e-05,
      "loss": 0.1884,
      "step": 2680
    },
    {
      "epoch": 0.10840316597086741,
      "grad_norm": 4.8933916091918945,
      "learning_rate": 9.731511269062906e-05,
      "loss": 0.1953,
      "step": 2681
    },
    {
      "epoch": 0.10844359982613443,
      "grad_norm": 2.1450068950653076,
      "learning_rate": 9.73130506269754e-05,
      "loss": 0.1302,
      "step": 2682
    },
    {
      "epoch": 0.10848403368140144,
      "grad_norm": 10.049890518188477,
      "learning_rate": 9.731098779363037e-05,
      "loss": 0.2393,
      "step": 2683
    },
    {
      "epoch": 0.10852446753666846,
      "grad_norm": 2.8999364376068115,
      "learning_rate": 9.730892419062758e-05,
      "loss": 0.1349,
      "step": 2684
    },
    {
      "epoch": 0.10856490139193546,
      "grad_norm": 4.442298412322998,
      "learning_rate": 9.730685981800054e-05,
      "loss": 0.1293,
      "step": 2685
    },
    {
      "epoch": 0.10860533524720248,
      "grad_norm": 7.971304416656494,
      "learning_rate": 9.730479467578286e-05,
      "loss": 0.3336,
      "step": 2686
    },
    {
      "epoch": 0.1086457691024695,
      "grad_norm": 7.373893737792969,
      "learning_rate": 9.730272876400815e-05,
      "loss": 0.199,
      "step": 2687
    },
    {
      "epoch": 0.10868620295773651,
      "grad_norm": 3.977097749710083,
      "learning_rate": 9.730066208270998e-05,
      "loss": 0.0826,
      "step": 2688
    },
    {
      "epoch": 0.10872663681300353,
      "grad_norm": 7.866024971008301,
      "learning_rate": 9.729859463192202e-05,
      "loss": 0.2753,
      "step": 2689
    },
    {
      "epoch": 0.10876707066827054,
      "grad_norm": 3.4729580879211426,
      "learning_rate": 9.729652641167786e-05,
      "loss": 0.2569,
      "step": 2690
    },
    {
      "epoch": 0.10880750452353756,
      "grad_norm": 4.893687725067139,
      "learning_rate": 9.729445742201118e-05,
      "loss": 0.1736,
      "step": 2691
    },
    {
      "epoch": 0.10884793837880458,
      "grad_norm": 5.954108715057373,
      "learning_rate": 9.729238766295561e-05,
      "loss": 0.2383,
      "step": 2692
    },
    {
      "epoch": 0.10888837223407159,
      "grad_norm": 5.850533962249756,
      "learning_rate": 9.729031713454483e-05,
      "loss": 0.1203,
      "step": 2693
    },
    {
      "epoch": 0.10892880608933861,
      "grad_norm": 1.6005723476409912,
      "learning_rate": 9.728824583681253e-05,
      "loss": 0.0717,
      "step": 2694
    },
    {
      "epoch": 0.10896923994460561,
      "grad_norm": 3.6024787425994873,
      "learning_rate": 9.728617376979241e-05,
      "loss": 0.1976,
      "step": 2695
    },
    {
      "epoch": 0.10900967379987263,
      "grad_norm": 6.509448528289795,
      "learning_rate": 9.728410093351816e-05,
      "loss": 0.188,
      "step": 2696
    },
    {
      "epoch": 0.10905010765513964,
      "grad_norm": 3.901845932006836,
      "learning_rate": 9.728202732802353e-05,
      "loss": 0.1169,
      "step": 2697
    },
    {
      "epoch": 0.10909054151040666,
      "grad_norm": 1.925940990447998,
      "learning_rate": 9.727995295334223e-05,
      "loss": 0.1377,
      "step": 2698
    },
    {
      "epoch": 0.10913097536567368,
      "grad_norm": 1.1993727684020996,
      "learning_rate": 9.727787780950801e-05,
      "loss": 0.0928,
      "step": 2699
    },
    {
      "epoch": 0.1091714092209407,
      "grad_norm": 5.564543724060059,
      "learning_rate": 9.727580189655462e-05,
      "loss": 0.1535,
      "step": 2700
    },
    {
      "epoch": 0.10921184307620771,
      "grad_norm": 9.033365249633789,
      "learning_rate": 9.727372521451586e-05,
      "loss": 0.1942,
      "step": 2701
    },
    {
      "epoch": 0.10925227693147473,
      "grad_norm": 8.384749412536621,
      "learning_rate": 9.727164776342548e-05,
      "loss": 0.2544,
      "step": 2702
    },
    {
      "epoch": 0.10929271078674174,
      "grad_norm": 11.48187255859375,
      "learning_rate": 9.72695695433173e-05,
      "loss": 0.2571,
      "step": 2703
    },
    {
      "epoch": 0.10933314464200876,
      "grad_norm": 6.480170726776123,
      "learning_rate": 9.72674905542251e-05,
      "loss": 0.1475,
      "step": 2704
    },
    {
      "epoch": 0.10937357849727578,
      "grad_norm": 2.7404513359069824,
      "learning_rate": 9.726541079618274e-05,
      "loss": 0.2311,
      "step": 2705
    },
    {
      "epoch": 0.10941401235254278,
      "grad_norm": 5.194509983062744,
      "learning_rate": 9.726333026922403e-05,
      "loss": 0.1699,
      "step": 2706
    },
    {
      "epoch": 0.1094544462078098,
      "grad_norm": 6.627828121185303,
      "learning_rate": 9.726124897338282e-05,
      "loss": 0.1202,
      "step": 2707
    },
    {
      "epoch": 0.10949488006307681,
      "grad_norm": 3.3623600006103516,
      "learning_rate": 9.725916690869297e-05,
      "loss": 0.1436,
      "step": 2708
    },
    {
      "epoch": 0.10953531391834383,
      "grad_norm": 3.284147262573242,
      "learning_rate": 9.725708407518836e-05,
      "loss": 0.1123,
      "step": 2709
    },
    {
      "epoch": 0.10957574777361084,
      "grad_norm": 5.178160190582275,
      "learning_rate": 9.725500047290286e-05,
      "loss": 0.1836,
      "step": 2710
    },
    {
      "epoch": 0.10961618162887786,
      "grad_norm": 2.245945453643799,
      "learning_rate": 9.725291610187034e-05,
      "loss": 0.2059,
      "step": 2711
    },
    {
      "epoch": 0.10965661548414488,
      "grad_norm": 7.060356616973877,
      "learning_rate": 9.725083096212479e-05,
      "loss": 0.2369,
      "step": 2712
    },
    {
      "epoch": 0.10969704933941189,
      "grad_norm": 6.570603370666504,
      "learning_rate": 9.724874505370005e-05,
      "loss": 0.1301,
      "step": 2713
    },
    {
      "epoch": 0.10973748319467891,
      "grad_norm": 3.98063063621521,
      "learning_rate": 9.724665837663008e-05,
      "loss": 0.1218,
      "step": 2714
    },
    {
      "epoch": 0.10977791704994593,
      "grad_norm": 9.319286346435547,
      "learning_rate": 9.724457093094885e-05,
      "loss": 0.2142,
      "step": 2715
    },
    {
      "epoch": 0.10981835090521294,
      "grad_norm": 11.230998992919922,
      "learning_rate": 9.724248271669027e-05,
      "loss": 0.2342,
      "step": 2716
    },
    {
      "epoch": 0.10985878476047994,
      "grad_norm": 6.4659576416015625,
      "learning_rate": 9.724039373388837e-05,
      "loss": 0.2117,
      "step": 2717
    },
    {
      "epoch": 0.10989921861574696,
      "grad_norm": 7.619095325469971,
      "learning_rate": 9.72383039825771e-05,
      "loss": 0.182,
      "step": 2718
    },
    {
      "epoch": 0.10993965247101398,
      "grad_norm": 3.9667162895202637,
      "learning_rate": 9.723621346279044e-05,
      "loss": 0.1588,
      "step": 2719
    },
    {
      "epoch": 0.10998008632628099,
      "grad_norm": 2.387579917907715,
      "learning_rate": 9.723412217456243e-05,
      "loss": 0.0712,
      "step": 2720
    },
    {
      "epoch": 0.11002052018154801,
      "grad_norm": 4.160400390625,
      "learning_rate": 9.723203011792708e-05,
      "loss": 0.203,
      "step": 2721
    },
    {
      "epoch": 0.11006095403681503,
      "grad_norm": 7.204240798950195,
      "learning_rate": 9.722993729291841e-05,
      "loss": 0.135,
      "step": 2722
    },
    {
      "epoch": 0.11010138789208204,
      "grad_norm": 5.673966407775879,
      "learning_rate": 9.72278436995705e-05,
      "loss": 0.1558,
      "step": 2723
    },
    {
      "epoch": 0.11014182174734906,
      "grad_norm": 6.922881603240967,
      "learning_rate": 9.722574933791738e-05,
      "loss": 0.1889,
      "step": 2724
    },
    {
      "epoch": 0.11018225560261607,
      "grad_norm": 6.871776580810547,
      "learning_rate": 9.722365420799312e-05,
      "loss": 0.2179,
      "step": 2725
    },
    {
      "epoch": 0.11022268945788309,
      "grad_norm": 4.513554573059082,
      "learning_rate": 9.722155830983182e-05,
      "loss": 0.1149,
      "step": 2726
    },
    {
      "epoch": 0.1102631233131501,
      "grad_norm": 6.205270290374756,
      "learning_rate": 9.721946164346757e-05,
      "loss": 0.1274,
      "step": 2727
    },
    {
      "epoch": 0.11030355716841711,
      "grad_norm": 4.183711528778076,
      "learning_rate": 9.721736420893448e-05,
      "loss": 0.1242,
      "step": 2728
    },
    {
      "epoch": 0.11034399102368413,
      "grad_norm": 4.3171539306640625,
      "learning_rate": 9.721526600626667e-05,
      "loss": 0.1162,
      "step": 2729
    },
    {
      "epoch": 0.11038442487895114,
      "grad_norm": 4.711130142211914,
      "learning_rate": 9.721316703549826e-05,
      "loss": 0.1429,
      "step": 2730
    },
    {
      "epoch": 0.11042485873421816,
      "grad_norm": 3.8932604789733887,
      "learning_rate": 9.721106729666343e-05,
      "loss": 0.1219,
      "step": 2731
    },
    {
      "epoch": 0.11046529258948518,
      "grad_norm": 3.7617011070251465,
      "learning_rate": 9.72089667897963e-05,
      "loss": 0.2082,
      "step": 2732
    },
    {
      "epoch": 0.11050572644475219,
      "grad_norm": 4.332525730133057,
      "learning_rate": 9.720686551493105e-05,
      "loss": 0.2002,
      "step": 2733
    },
    {
      "epoch": 0.11054616030001921,
      "grad_norm": 3.4756953716278076,
      "learning_rate": 9.72047634721019e-05,
      "loss": 0.2166,
      "step": 2734
    },
    {
      "epoch": 0.11058659415528622,
      "grad_norm": 4.317763328552246,
      "learning_rate": 9.7202660661343e-05,
      "loss": 0.2381,
      "step": 2735
    },
    {
      "epoch": 0.11062702801055324,
      "grad_norm": 5.27736234664917,
      "learning_rate": 9.720055708268858e-05,
      "loss": 0.0793,
      "step": 2736
    },
    {
      "epoch": 0.11066746186582026,
      "grad_norm": 5.588968276977539,
      "learning_rate": 9.719845273617288e-05,
      "loss": 0.2335,
      "step": 2737
    },
    {
      "epoch": 0.11070789572108726,
      "grad_norm": 5.556321620941162,
      "learning_rate": 9.719634762183008e-05,
      "loss": 0.1966,
      "step": 2738
    },
    {
      "epoch": 0.11074832957635428,
      "grad_norm": 10.09548568725586,
      "learning_rate": 9.719424173969447e-05,
      "loss": 0.2576,
      "step": 2739
    },
    {
      "epoch": 0.11078876343162129,
      "grad_norm": 4.945663928985596,
      "learning_rate": 9.71921350898003e-05,
      "loss": 0.3158,
      "step": 2740
    },
    {
      "epoch": 0.11082919728688831,
      "grad_norm": 3.6579713821411133,
      "learning_rate": 9.719002767218184e-05,
      "loss": 0.1436,
      "step": 2741
    },
    {
      "epoch": 0.11086963114215533,
      "grad_norm": 5.560998439788818,
      "learning_rate": 9.718791948687336e-05,
      "loss": 0.292,
      "step": 2742
    },
    {
      "epoch": 0.11091006499742234,
      "grad_norm": 9.81323528289795,
      "learning_rate": 9.718581053390918e-05,
      "loss": 0.1936,
      "step": 2743
    },
    {
      "epoch": 0.11095049885268936,
      "grad_norm": 3.0580177307128906,
      "learning_rate": 9.71837008133236e-05,
      "loss": 0.125,
      "step": 2744
    },
    {
      "epoch": 0.11099093270795637,
      "grad_norm": 6.840806484222412,
      "learning_rate": 9.718159032515092e-05,
      "loss": 0.2768,
      "step": 2745
    },
    {
      "epoch": 0.11103136656322339,
      "grad_norm": 7.835115909576416,
      "learning_rate": 9.71794790694255e-05,
      "loss": 0.2581,
      "step": 2746
    },
    {
      "epoch": 0.11107180041849041,
      "grad_norm": 3.199481725692749,
      "learning_rate": 9.717736704618168e-05,
      "loss": 0.2213,
      "step": 2747
    },
    {
      "epoch": 0.11111223427375742,
      "grad_norm": 9.323312759399414,
      "learning_rate": 9.71752542554538e-05,
      "loss": 0.2492,
      "step": 2748
    },
    {
      "epoch": 0.11115266812902443,
      "grad_norm": 4.180640697479248,
      "learning_rate": 9.717314069727626e-05,
      "loss": 0.1817,
      "step": 2749
    },
    {
      "epoch": 0.11119310198429144,
      "grad_norm": 5.289682865142822,
      "learning_rate": 9.717102637168342e-05,
      "loss": 0.2364,
      "step": 2750
    },
    {
      "epoch": 0.11123353583955846,
      "grad_norm": 5.480874061584473,
      "learning_rate": 9.71689112787097e-05,
      "loss": 0.2005,
      "step": 2751
    },
    {
      "epoch": 0.11127396969482548,
      "grad_norm": 6.438024520874023,
      "learning_rate": 9.716679541838949e-05,
      "loss": 0.2478,
      "step": 2752
    },
    {
      "epoch": 0.11131440355009249,
      "grad_norm": 2.5672767162323,
      "learning_rate": 9.71646787907572e-05,
      "loss": 0.203,
      "step": 2753
    },
    {
      "epoch": 0.11135483740535951,
      "grad_norm": 4.372616767883301,
      "learning_rate": 9.716256139584729e-05,
      "loss": 0.2053,
      "step": 2754
    },
    {
      "epoch": 0.11139527126062652,
      "grad_norm": 7.948934555053711,
      "learning_rate": 9.71604432336942e-05,
      "loss": 0.1504,
      "step": 2755
    },
    {
      "epoch": 0.11143570511589354,
      "grad_norm": 7.318721294403076,
      "learning_rate": 9.715832430433237e-05,
      "loss": 0.2799,
      "step": 2756
    },
    {
      "epoch": 0.11147613897116056,
      "grad_norm": 2.2729272842407227,
      "learning_rate": 9.715620460779628e-05,
      "loss": 0.121,
      "step": 2757
    },
    {
      "epoch": 0.11151657282642757,
      "grad_norm": 6.40389347076416,
      "learning_rate": 9.715408414412044e-05,
      "loss": 0.251,
      "step": 2758
    },
    {
      "epoch": 0.11155700668169458,
      "grad_norm": 8.632926940917969,
      "learning_rate": 9.71519629133393e-05,
      "loss": 0.2993,
      "step": 2759
    },
    {
      "epoch": 0.11159744053696159,
      "grad_norm": 12.319863319396973,
      "learning_rate": 9.71498409154874e-05,
      "loss": 0.3096,
      "step": 2760
    },
    {
      "epoch": 0.11163787439222861,
      "grad_norm": 7.936651706695557,
      "learning_rate": 9.714771815059926e-05,
      "loss": 0.2166,
      "step": 2761
    },
    {
      "epoch": 0.11167830824749563,
      "grad_norm": 8.032295227050781,
      "learning_rate": 9.71455946187094e-05,
      "loss": 0.2454,
      "step": 2762
    },
    {
      "epoch": 0.11171874210276264,
      "grad_norm": 1.930954098701477,
      "learning_rate": 9.714347031985236e-05,
      "loss": 0.0956,
      "step": 2763
    },
    {
      "epoch": 0.11175917595802966,
      "grad_norm": 3.1181681156158447,
      "learning_rate": 9.714134525406272e-05,
      "loss": 0.1198,
      "step": 2764
    },
    {
      "epoch": 0.11179960981329667,
      "grad_norm": 6.893107891082764,
      "learning_rate": 9.713921942137504e-05,
      "loss": 0.1825,
      "step": 2765
    },
    {
      "epoch": 0.11184004366856369,
      "grad_norm": 6.599243640899658,
      "learning_rate": 9.713709282182389e-05,
      "loss": 0.2384,
      "step": 2766
    },
    {
      "epoch": 0.11188047752383071,
      "grad_norm": 3.781618118286133,
      "learning_rate": 9.71349654554439e-05,
      "loss": 0.1734,
      "step": 2767
    },
    {
      "epoch": 0.11192091137909772,
      "grad_norm": 5.28074312210083,
      "learning_rate": 9.713283732226963e-05,
      "loss": 0.1842,
      "step": 2768
    },
    {
      "epoch": 0.11196134523436474,
      "grad_norm": 2.2957258224487305,
      "learning_rate": 9.713070842233575e-05,
      "loss": 0.1333,
      "step": 2769
    },
    {
      "epoch": 0.11200177908963174,
      "grad_norm": 13.93614673614502,
      "learning_rate": 9.712857875567686e-05,
      "loss": 0.3176,
      "step": 2770
    },
    {
      "epoch": 0.11204221294489876,
      "grad_norm": 4.601606845855713,
      "learning_rate": 9.712644832232762e-05,
      "loss": 0.2109,
      "step": 2771
    },
    {
      "epoch": 0.11208264680016577,
      "grad_norm": 5.562735080718994,
      "learning_rate": 9.712431712232268e-05,
      "loss": 0.183,
      "step": 2772
    },
    {
      "epoch": 0.11212308065543279,
      "grad_norm": 5.232434272766113,
      "learning_rate": 9.71221851556967e-05,
      "loss": 0.1423,
      "step": 2773
    },
    {
      "epoch": 0.11216351451069981,
      "grad_norm": 2.3837761878967285,
      "learning_rate": 9.712005242248441e-05,
      "loss": 0.1261,
      "step": 2774
    },
    {
      "epoch": 0.11220394836596682,
      "grad_norm": 2.0290541648864746,
      "learning_rate": 9.711791892272045e-05,
      "loss": 0.0847,
      "step": 2775
    },
    {
      "epoch": 0.11224438222123384,
      "grad_norm": 4.864945411682129,
      "learning_rate": 9.711578465643956e-05,
      "loss": 0.1764,
      "step": 2776
    },
    {
      "epoch": 0.11228481607650086,
      "grad_norm": 3.0482337474823,
      "learning_rate": 9.711364962367644e-05,
      "loss": 0.1628,
      "step": 2777
    },
    {
      "epoch": 0.11232524993176787,
      "grad_norm": 4.731161117553711,
      "learning_rate": 9.711151382446584e-05,
      "loss": 0.2279,
      "step": 2778
    },
    {
      "epoch": 0.11236568378703489,
      "grad_norm": 6.966867923736572,
      "learning_rate": 9.71093772588425e-05,
      "loss": 0.2333,
      "step": 2779
    },
    {
      "epoch": 0.1124061176423019,
      "grad_norm": 4.990914821624756,
      "learning_rate": 9.710723992684119e-05,
      "loss": 0.1166,
      "step": 2780
    },
    {
      "epoch": 0.11244655149756891,
      "grad_norm": 4.783514499664307,
      "learning_rate": 9.710510182849664e-05,
      "loss": 0.1727,
      "step": 2781
    },
    {
      "epoch": 0.11248698535283592,
      "grad_norm": 3.762657403945923,
      "learning_rate": 9.710296296384368e-05,
      "loss": 0.1835,
      "step": 2782
    },
    {
      "epoch": 0.11252741920810294,
      "grad_norm": 2.1442465782165527,
      "learning_rate": 9.710082333291708e-05,
      "loss": 0.0687,
      "step": 2783
    },
    {
      "epoch": 0.11256785306336996,
      "grad_norm": 4.6128082275390625,
      "learning_rate": 9.709868293575164e-05,
      "loss": 0.2047,
      "step": 2784
    },
    {
      "epoch": 0.11260828691863697,
      "grad_norm": 8.868513107299805,
      "learning_rate": 9.709654177238221e-05,
      "loss": 0.2216,
      "step": 2785
    },
    {
      "epoch": 0.11264872077390399,
      "grad_norm": 4.794582366943359,
      "learning_rate": 9.70943998428436e-05,
      "loss": 0.1769,
      "step": 2786
    },
    {
      "epoch": 0.112689154629171,
      "grad_norm": 6.929192066192627,
      "learning_rate": 9.709225714717065e-05,
      "loss": 0.2603,
      "step": 2787
    },
    {
      "epoch": 0.11272958848443802,
      "grad_norm": 7.446916103363037,
      "learning_rate": 9.709011368539825e-05,
      "loss": 0.124,
      "step": 2788
    },
    {
      "epoch": 0.11277002233970504,
      "grad_norm": 4.700041770935059,
      "learning_rate": 9.708796945756123e-05,
      "loss": 0.1363,
      "step": 2789
    },
    {
      "epoch": 0.11281045619497206,
      "grad_norm": 3.09904408454895,
      "learning_rate": 9.70858244636945e-05,
      "loss": 0.0924,
      "step": 2790
    },
    {
      "epoch": 0.11285089005023906,
      "grad_norm": 5.390627384185791,
      "learning_rate": 9.708367870383293e-05,
      "loss": 0.1501,
      "step": 2791
    },
    {
      "epoch": 0.11289132390550607,
      "grad_norm": 8.436245918273926,
      "learning_rate": 9.708153217801146e-05,
      "loss": 0.2869,
      "step": 2792
    },
    {
      "epoch": 0.11293175776077309,
      "grad_norm": 5.075014114379883,
      "learning_rate": 9.707938488626497e-05,
      "loss": 0.2997,
      "step": 2793
    },
    {
      "epoch": 0.11297219161604011,
      "grad_norm": 5.2177734375,
      "learning_rate": 9.707723682862844e-05,
      "loss": 0.1808,
      "step": 2794
    },
    {
      "epoch": 0.11301262547130712,
      "grad_norm": 3.8890151977539062,
      "learning_rate": 9.707508800513679e-05,
      "loss": 0.0964,
      "step": 2795
    },
    {
      "epoch": 0.11305305932657414,
      "grad_norm": 7.150461196899414,
      "learning_rate": 9.707293841582494e-05,
      "loss": 0.2491,
      "step": 2796
    },
    {
      "epoch": 0.11309349318184116,
      "grad_norm": 2.5301554203033447,
      "learning_rate": 9.707078806072794e-05,
      "loss": 0.1432,
      "step": 2797
    },
    {
      "epoch": 0.11313392703710817,
      "grad_norm": 6.122782230377197,
      "learning_rate": 9.70686369398807e-05,
      "loss": 0.1928,
      "step": 2798
    },
    {
      "epoch": 0.11317436089237519,
      "grad_norm": 3.736940860748291,
      "learning_rate": 9.706648505331824e-05,
      "loss": 0.1805,
      "step": 2799
    },
    {
      "epoch": 0.1132147947476422,
      "grad_norm": 9.061598777770996,
      "learning_rate": 9.706433240107558e-05,
      "loss": 0.2938,
      "step": 2800
    },
    {
      "epoch": 0.11325522860290922,
      "grad_norm": 5.604365348815918,
      "learning_rate": 9.706217898318774e-05,
      "loss": 0.1786,
      "step": 2801
    },
    {
      "epoch": 0.11329566245817622,
      "grad_norm": 5.7084808349609375,
      "learning_rate": 9.706002479968972e-05,
      "loss": 0.2238,
      "step": 2802
    },
    {
      "epoch": 0.11333609631344324,
      "grad_norm": 6.061769485473633,
      "learning_rate": 9.70578698506166e-05,
      "loss": 0.3304,
      "step": 2803
    },
    {
      "epoch": 0.11337653016871026,
      "grad_norm": 6.470050811767578,
      "learning_rate": 9.705571413600342e-05,
      "loss": 0.1748,
      "step": 2804
    },
    {
      "epoch": 0.11341696402397727,
      "grad_norm": 4.864773750305176,
      "learning_rate": 9.705355765588524e-05,
      "loss": 0.2596,
      "step": 2805
    },
    {
      "epoch": 0.11345739787924429,
      "grad_norm": 4.01631498336792,
      "learning_rate": 9.705140041029717e-05,
      "loss": 0.123,
      "step": 2806
    },
    {
      "epoch": 0.1134978317345113,
      "grad_norm": 4.96744441986084,
      "learning_rate": 9.704924239927429e-05,
      "loss": 0.2615,
      "step": 2807
    },
    {
      "epoch": 0.11353826558977832,
      "grad_norm": 7.059342861175537,
      "learning_rate": 9.704708362285169e-05,
      "loss": 0.248,
      "step": 2808
    },
    {
      "epoch": 0.11357869944504534,
      "grad_norm": 11.899310111999512,
      "learning_rate": 9.704492408106449e-05,
      "loss": 0.3957,
      "step": 2809
    },
    {
      "epoch": 0.11361913330031236,
      "grad_norm": 9.38026237487793,
      "learning_rate": 9.704276377394785e-05,
      "loss": 0.2114,
      "step": 2810
    },
    {
      "epoch": 0.11365956715557937,
      "grad_norm": 4.253027439117432,
      "learning_rate": 9.704060270153692e-05,
      "loss": 0.1947,
      "step": 2811
    },
    {
      "epoch": 0.11370000101084639,
      "grad_norm": 2.6575069427490234,
      "learning_rate": 9.70384408638668e-05,
      "loss": 0.1814,
      "step": 2812
    },
    {
      "epoch": 0.11374043486611339,
      "grad_norm": 5.6372904777526855,
      "learning_rate": 9.703627826097272e-05,
      "loss": 0.2942,
      "step": 2813
    },
    {
      "epoch": 0.11378086872138041,
      "grad_norm": 4.492222785949707,
      "learning_rate": 9.703411489288983e-05,
      "loss": 0.1833,
      "step": 2814
    },
    {
      "epoch": 0.11382130257664742,
      "grad_norm": 3.756523609161377,
      "learning_rate": 9.703195075965333e-05,
      "loss": 0.1731,
      "step": 2815
    },
    {
      "epoch": 0.11386173643191444,
      "grad_norm": 4.3988776206970215,
      "learning_rate": 9.702978586129841e-05,
      "loss": 0.2562,
      "step": 2816
    },
    {
      "epoch": 0.11390217028718146,
      "grad_norm": 4.0481085777282715,
      "learning_rate": 9.702762019786032e-05,
      "loss": 0.2208,
      "step": 2817
    },
    {
      "epoch": 0.11394260414244847,
      "grad_norm": 4.251232147216797,
      "learning_rate": 9.702545376937426e-05,
      "loss": 0.2023,
      "step": 2818
    },
    {
      "epoch": 0.11398303799771549,
      "grad_norm": 11.273815155029297,
      "learning_rate": 9.70232865758755e-05,
      "loss": 0.2524,
      "step": 2819
    },
    {
      "epoch": 0.1140234718529825,
      "grad_norm": 2.6662657260894775,
      "learning_rate": 9.702111861739929e-05,
      "loss": 0.1827,
      "step": 2820
    },
    {
      "epoch": 0.11406390570824952,
      "grad_norm": 5.680376052856445,
      "learning_rate": 9.701894989398088e-05,
      "loss": 0.1964,
      "step": 2821
    },
    {
      "epoch": 0.11410433956351654,
      "grad_norm": 4.24598503112793,
      "learning_rate": 9.701678040565557e-05,
      "loss": 0.1934,
      "step": 2822
    },
    {
      "epoch": 0.11414477341878354,
      "grad_norm": 1.675412893295288,
      "learning_rate": 9.701461015245866e-05,
      "loss": 0.1314,
      "step": 2823
    },
    {
      "epoch": 0.11418520727405056,
      "grad_norm": 6.086452960968018,
      "learning_rate": 9.701243913442541e-05,
      "loss": 0.2253,
      "step": 2824
    },
    {
      "epoch": 0.11422564112931757,
      "grad_norm": 3.9554800987243652,
      "learning_rate": 9.70102673515912e-05,
      "loss": 0.159,
      "step": 2825
    },
    {
      "epoch": 0.11426607498458459,
      "grad_norm": 5.287421226501465,
      "learning_rate": 9.700809480399132e-05,
      "loss": 0.1553,
      "step": 2826
    },
    {
      "epoch": 0.1143065088398516,
      "grad_norm": 3.096757650375366,
      "learning_rate": 9.700592149166113e-05,
      "loss": 0.1299,
      "step": 2827
    },
    {
      "epoch": 0.11434694269511862,
      "grad_norm": 4.448558807373047,
      "learning_rate": 9.700374741463598e-05,
      "loss": 0.1541,
      "step": 2828
    },
    {
      "epoch": 0.11438737655038564,
      "grad_norm": 2.8007421493530273,
      "learning_rate": 9.700157257295124e-05,
      "loss": 0.1252,
      "step": 2829
    },
    {
      "epoch": 0.11442781040565266,
      "grad_norm": 7.094184398651123,
      "learning_rate": 9.69993969666423e-05,
      "loss": 0.2429,
      "step": 2830
    },
    {
      "epoch": 0.11446824426091967,
      "grad_norm": 5.922878742218018,
      "learning_rate": 9.699722059574452e-05,
      "loss": 0.2679,
      "step": 2831
    },
    {
      "epoch": 0.11450867811618669,
      "grad_norm": 2.5166139602661133,
      "learning_rate": 9.699504346029335e-05,
      "loss": 0.1654,
      "step": 2832
    },
    {
      "epoch": 0.1145491119714537,
      "grad_norm": 4.236229419708252,
      "learning_rate": 9.699286556032418e-05,
      "loss": 0.2288,
      "step": 2833
    },
    {
      "epoch": 0.1145895458267207,
      "grad_norm": 4.291086673736572,
      "learning_rate": 9.699068689587243e-05,
      "loss": 0.2011,
      "step": 2834
    },
    {
      "epoch": 0.11462997968198772,
      "grad_norm": 1.7348905801773071,
      "learning_rate": 9.698850746697358e-05,
      "loss": 0.1787,
      "step": 2835
    },
    {
      "epoch": 0.11467041353725474,
      "grad_norm": 2.2019577026367188,
      "learning_rate": 9.698632727366304e-05,
      "loss": 0.0998,
      "step": 2836
    },
    {
      "epoch": 0.11471084739252176,
      "grad_norm": 4.81380558013916,
      "learning_rate": 9.698414631597632e-05,
      "loss": 0.2333,
      "step": 2837
    },
    {
      "epoch": 0.11475128124778877,
      "grad_norm": 7.548548221588135,
      "learning_rate": 9.698196459394888e-05,
      "loss": 0.1661,
      "step": 2838
    },
    {
      "epoch": 0.11479171510305579,
      "grad_norm": 4.569876670837402,
      "learning_rate": 9.69797821076162e-05,
      "loss": 0.1994,
      "step": 2839
    },
    {
      "epoch": 0.1148321489583228,
      "grad_norm": 4.292208671569824,
      "learning_rate": 9.697759885701381e-05,
      "loss": 0.2227,
      "step": 2840
    },
    {
      "epoch": 0.11487258281358982,
      "grad_norm": 3.5582735538482666,
      "learning_rate": 9.69754148421772e-05,
      "loss": 0.1225,
      "step": 2841
    },
    {
      "epoch": 0.11491301666885684,
      "grad_norm": 10.757134437561035,
      "learning_rate": 9.697323006314194e-05,
      "loss": 0.3325,
      "step": 2842
    },
    {
      "epoch": 0.11495345052412385,
      "grad_norm": 4.589961528778076,
      "learning_rate": 9.697104451994352e-05,
      "loss": 0.1452,
      "step": 2843
    },
    {
      "epoch": 0.11499388437939087,
      "grad_norm": 5.3293256759643555,
      "learning_rate": 9.696885821261755e-05,
      "loss": 0.3409,
      "step": 2844
    },
    {
      "epoch": 0.11503431823465787,
      "grad_norm": 4.313346862792969,
      "learning_rate": 9.696667114119955e-05,
      "loss": 0.168,
      "step": 2845
    },
    {
      "epoch": 0.11507475208992489,
      "grad_norm": 8.4745512008667,
      "learning_rate": 9.696448330572511e-05,
      "loss": 0.2417,
      "step": 2846
    },
    {
      "epoch": 0.1151151859451919,
      "grad_norm": 3.573390007019043,
      "learning_rate": 9.696229470622984e-05,
      "loss": 0.1945,
      "step": 2847
    },
    {
      "epoch": 0.11515561980045892,
      "grad_norm": 3.2278902530670166,
      "learning_rate": 9.696010534274934e-05,
      "loss": 0.2087,
      "step": 2848
    },
    {
      "epoch": 0.11519605365572594,
      "grad_norm": 6.376348495483398,
      "learning_rate": 9.695791521531922e-05,
      "loss": 0.1889,
      "step": 2849
    },
    {
      "epoch": 0.11523648751099295,
      "grad_norm": 17.30323600769043,
      "learning_rate": 9.69557243239751e-05,
      "loss": 0.5533,
      "step": 2850
    },
    {
      "epoch": 0.11527692136625997,
      "grad_norm": 8.09468936920166,
      "learning_rate": 9.695353266875264e-05,
      "loss": 0.2648,
      "step": 2851
    },
    {
      "epoch": 0.11531735522152699,
      "grad_norm": 9.455105781555176,
      "learning_rate": 9.695134024968749e-05,
      "loss": 0.2743,
      "step": 2852
    },
    {
      "epoch": 0.115357789076794,
      "grad_norm": 7.2366766929626465,
      "learning_rate": 9.694914706681529e-05,
      "loss": 0.188,
      "step": 2853
    },
    {
      "epoch": 0.11539822293206102,
      "grad_norm": 4.267734050750732,
      "learning_rate": 9.694695312017175e-05,
      "loss": 0.2088,
      "step": 2854
    },
    {
      "epoch": 0.11543865678732802,
      "grad_norm": 3.5485422611236572,
      "learning_rate": 9.694475840979255e-05,
      "loss": 0.2015,
      "step": 2855
    },
    {
      "epoch": 0.11547909064259504,
      "grad_norm": 2.5305027961730957,
      "learning_rate": 9.694256293571342e-05,
      "loss": 0.1543,
      "step": 2856
    },
    {
      "epoch": 0.11551952449786206,
      "grad_norm": 8.649937629699707,
      "learning_rate": 9.694036669797002e-05,
      "loss": 0.2721,
      "step": 2857
    },
    {
      "epoch": 0.11555995835312907,
      "grad_norm": 3.4112722873687744,
      "learning_rate": 9.693816969659812e-05,
      "loss": 0.2146,
      "step": 2858
    },
    {
      "epoch": 0.11560039220839609,
      "grad_norm": 2.542811155319214,
      "learning_rate": 9.693597193163346e-05,
      "loss": 0.1256,
      "step": 2859
    },
    {
      "epoch": 0.1156408260636631,
      "grad_norm": 10.791812896728516,
      "learning_rate": 9.693377340311178e-05,
      "loss": 0.3058,
      "step": 2860
    },
    {
      "epoch": 0.11568125991893012,
      "grad_norm": 7.143285274505615,
      "learning_rate": 9.693157411106886e-05,
      "loss": 0.1771,
      "step": 2861
    },
    {
      "epoch": 0.11572169377419714,
      "grad_norm": 6.040772914886475,
      "learning_rate": 9.692937405554044e-05,
      "loss": 0.2373,
      "step": 2862
    },
    {
      "epoch": 0.11576212762946415,
      "grad_norm": 12.410866737365723,
      "learning_rate": 9.692717323656236e-05,
      "loss": 0.2924,
      "step": 2863
    },
    {
      "epoch": 0.11580256148473117,
      "grad_norm": 4.204496383666992,
      "learning_rate": 9.692497165417041e-05,
      "loss": 0.2891,
      "step": 2864
    },
    {
      "epoch": 0.11584299533999819,
      "grad_norm": 8.65272331237793,
      "learning_rate": 9.692276930840037e-05,
      "loss": 0.2046,
      "step": 2865
    },
    {
      "epoch": 0.11588342919526519,
      "grad_norm": 10.338696479797363,
      "learning_rate": 9.692056619928811e-05,
      "loss": 0.285,
      "step": 2866
    },
    {
      "epoch": 0.1159238630505322,
      "grad_norm": 5.936294078826904,
      "learning_rate": 9.691836232686946e-05,
      "loss": 0.1837,
      "step": 2867
    },
    {
      "epoch": 0.11596429690579922,
      "grad_norm": 3.968024730682373,
      "learning_rate": 9.691615769118027e-05,
      "loss": 0.2064,
      "step": 2868
    },
    {
      "epoch": 0.11600473076106624,
      "grad_norm": 6.816439151763916,
      "learning_rate": 9.691395229225639e-05,
      "loss": 0.173,
      "step": 2869
    },
    {
      "epoch": 0.11604516461633325,
      "grad_norm": 6.925784111022949,
      "learning_rate": 9.691174613013372e-05,
      "loss": 0.1999,
      "step": 2870
    },
    {
      "epoch": 0.11608559847160027,
      "grad_norm": 6.165559768676758,
      "learning_rate": 9.690953920484815e-05,
      "loss": 0.2788,
      "step": 2871
    },
    {
      "epoch": 0.11612603232686729,
      "grad_norm": 4.14732551574707,
      "learning_rate": 9.690733151643556e-05,
      "loss": 0.2287,
      "step": 2872
    },
    {
      "epoch": 0.1161664661821343,
      "grad_norm": 7.598495960235596,
      "learning_rate": 9.690512306493188e-05,
      "loss": 0.2153,
      "step": 2873
    },
    {
      "epoch": 0.11620690003740132,
      "grad_norm": 2.3728749752044678,
      "learning_rate": 9.690291385037303e-05,
      "loss": 0.094,
      "step": 2874
    },
    {
      "epoch": 0.11624733389266834,
      "grad_norm": 5.555934429168701,
      "learning_rate": 9.690070387279497e-05,
      "loss": 0.1197,
      "step": 2875
    },
    {
      "epoch": 0.11628776774793535,
      "grad_norm": 5.817972660064697,
      "learning_rate": 9.689849313223362e-05,
      "loss": 0.1979,
      "step": 2876
    },
    {
      "epoch": 0.11632820160320236,
      "grad_norm": 3.825354814529419,
      "learning_rate": 9.689628162872497e-05,
      "loss": 0.1742,
      "step": 2877
    },
    {
      "epoch": 0.11636863545846937,
      "grad_norm": 4.773843288421631,
      "learning_rate": 9.6894069362305e-05,
      "loss": 0.2321,
      "step": 2878
    },
    {
      "epoch": 0.11640906931373639,
      "grad_norm": 7.5234479904174805,
      "learning_rate": 9.689185633300967e-05,
      "loss": 0.2166,
      "step": 2879
    },
    {
      "epoch": 0.1164495031690034,
      "grad_norm": 5.8592400550842285,
      "learning_rate": 9.688964254087501e-05,
      "loss": 0.1752,
      "step": 2880
    },
    {
      "epoch": 0.11648993702427042,
      "grad_norm": 3.7419512271881104,
      "learning_rate": 9.688742798593702e-05,
      "loss": 0.1721,
      "step": 2881
    },
    {
      "epoch": 0.11653037087953744,
      "grad_norm": 1.8167724609375,
      "learning_rate": 9.688521266823174e-05,
      "loss": 0.0862,
      "step": 2882
    },
    {
      "epoch": 0.11657080473480445,
      "grad_norm": 15.352806091308594,
      "learning_rate": 9.688299658779518e-05,
      "loss": 0.3135,
      "step": 2883
    },
    {
      "epoch": 0.11661123859007147,
      "grad_norm": 7.225453853607178,
      "learning_rate": 9.688077974466342e-05,
      "loss": 0.2418,
      "step": 2884
    },
    {
      "epoch": 0.11665167244533849,
      "grad_norm": 3.5432960987091064,
      "learning_rate": 9.687856213887252e-05,
      "loss": 0.1118,
      "step": 2885
    },
    {
      "epoch": 0.1166921063006055,
      "grad_norm": 8.04720687866211,
      "learning_rate": 9.687634377045854e-05,
      "loss": 0.2901,
      "step": 2886
    },
    {
      "epoch": 0.11673254015587252,
      "grad_norm": 4.017889976501465,
      "learning_rate": 9.687412463945758e-05,
      "loss": 0.1241,
      "step": 2887
    },
    {
      "epoch": 0.11677297401113952,
      "grad_norm": 8.064705848693848,
      "learning_rate": 9.687190474590575e-05,
      "loss": 0.1945,
      "step": 2888
    },
    {
      "epoch": 0.11681340786640654,
      "grad_norm": 3.754807949066162,
      "learning_rate": 9.686968408983913e-05,
      "loss": 0.118,
      "step": 2889
    },
    {
      "epoch": 0.11685384172167355,
      "grad_norm": 2.702610731124878,
      "learning_rate": 9.68674626712939e-05,
      "loss": 0.1241,
      "step": 2890
    },
    {
      "epoch": 0.11689427557694057,
      "grad_norm": 5.889130592346191,
      "learning_rate": 9.686524049030614e-05,
      "loss": 0.3585,
      "step": 2891
    },
    {
      "epoch": 0.11693470943220759,
      "grad_norm": 3.7998948097229004,
      "learning_rate": 9.686301754691204e-05,
      "loss": 0.1747,
      "step": 2892
    },
    {
      "epoch": 0.1169751432874746,
      "grad_norm": 2.8380517959594727,
      "learning_rate": 9.686079384114774e-05,
      "loss": 0.0758,
      "step": 2893
    },
    {
      "epoch": 0.11701557714274162,
      "grad_norm": 3.963920831680298,
      "learning_rate": 9.685856937304944e-05,
      "loss": 0.1456,
      "step": 2894
    },
    {
      "epoch": 0.11705601099800864,
      "grad_norm": 1.5102020502090454,
      "learning_rate": 9.68563441426533e-05,
      "loss": 0.0429,
      "step": 2895
    },
    {
      "epoch": 0.11709644485327565,
      "grad_norm": 6.050605773925781,
      "learning_rate": 9.685411814999554e-05,
      "loss": 0.2432,
      "step": 2896
    },
    {
      "epoch": 0.11713687870854267,
      "grad_norm": 5.8350629806518555,
      "learning_rate": 9.685189139511237e-05,
      "loss": 0.1828,
      "step": 2897
    },
    {
      "epoch": 0.11717731256380967,
      "grad_norm": 3.5678701400756836,
      "learning_rate": 9.684966387804e-05,
      "loss": 0.2427,
      "step": 2898
    },
    {
      "epoch": 0.11721774641907669,
      "grad_norm": 7.5123467445373535,
      "learning_rate": 9.684743559881471e-05,
      "loss": 0.2762,
      "step": 2899
    },
    {
      "epoch": 0.1172581802743437,
      "grad_norm": 6.001222610473633,
      "learning_rate": 9.684520655747269e-05,
      "loss": 0.1943,
      "step": 2900
    },
    {
      "epoch": 0.11729861412961072,
      "grad_norm": 7.042908668518066,
      "learning_rate": 9.684297675405023e-05,
      "loss": 0.3361,
      "step": 2901
    },
    {
      "epoch": 0.11733904798487774,
      "grad_norm": 8.585587501525879,
      "learning_rate": 9.684074618858361e-05,
      "loss": 0.2291,
      "step": 2902
    },
    {
      "epoch": 0.11737948184014475,
      "grad_norm": 5.050895690917969,
      "learning_rate": 9.68385148611091e-05,
      "loss": 0.1206,
      "step": 2903
    },
    {
      "epoch": 0.11741991569541177,
      "grad_norm": 5.075092792510986,
      "learning_rate": 9.683628277166304e-05,
      "loss": 0.0917,
      "step": 2904
    },
    {
      "epoch": 0.11746034955067879,
      "grad_norm": 5.4044084548950195,
      "learning_rate": 9.683404992028168e-05,
      "loss": 0.1249,
      "step": 2905
    },
    {
      "epoch": 0.1175007834059458,
      "grad_norm": 4.345137119293213,
      "learning_rate": 9.683181630700141e-05,
      "loss": 0.167,
      "step": 2906
    },
    {
      "epoch": 0.11754121726121282,
      "grad_norm": 10.543241500854492,
      "learning_rate": 9.682958193185851e-05,
      "loss": 0.331,
      "step": 2907
    },
    {
      "epoch": 0.11758165111647984,
      "grad_norm": 5.086560249328613,
      "learning_rate": 9.682734679488937e-05,
      "loss": 0.1541,
      "step": 2908
    },
    {
      "epoch": 0.11762208497174684,
      "grad_norm": 4.951198577880859,
      "learning_rate": 9.682511089613033e-05,
      "loss": 0.1004,
      "step": 2909
    },
    {
      "epoch": 0.11766251882701385,
      "grad_norm": 3.595529079437256,
      "learning_rate": 9.682287423561775e-05,
      "loss": 0.0661,
      "step": 2910
    },
    {
      "epoch": 0.11770295268228087,
      "grad_norm": 4.575246810913086,
      "learning_rate": 9.682063681338807e-05,
      "loss": 0.0704,
      "step": 2911
    },
    {
      "epoch": 0.11774338653754789,
      "grad_norm": 6.341940402984619,
      "learning_rate": 9.681839862947762e-05,
      "loss": 0.2537,
      "step": 2912
    },
    {
      "epoch": 0.1177838203928149,
      "grad_norm": 9.490710258483887,
      "learning_rate": 9.681615968392285e-05,
      "loss": 0.2607,
      "step": 2913
    },
    {
      "epoch": 0.11782425424808192,
      "grad_norm": 2.703104019165039,
      "learning_rate": 9.681391997676019e-05,
      "loss": 0.126,
      "step": 2914
    },
    {
      "epoch": 0.11786468810334894,
      "grad_norm": 7.211833953857422,
      "learning_rate": 9.681167950802605e-05,
      "loss": 0.1605,
      "step": 2915
    },
    {
      "epoch": 0.11790512195861595,
      "grad_norm": 11.522160530090332,
      "learning_rate": 9.680943827775689e-05,
      "loss": 0.3022,
      "step": 2916
    },
    {
      "epoch": 0.11794555581388297,
      "grad_norm": 6.5127105712890625,
      "learning_rate": 9.680719628598918e-05,
      "loss": 0.1515,
      "step": 2917
    },
    {
      "epoch": 0.11798598966914998,
      "grad_norm": 8.002151489257812,
      "learning_rate": 9.680495353275937e-05,
      "loss": 0.2211,
      "step": 2918
    },
    {
      "epoch": 0.118026423524417,
      "grad_norm": 5.835089206695557,
      "learning_rate": 9.680271001810398e-05,
      "loss": 0.2736,
      "step": 2919
    },
    {
      "epoch": 0.118066857379684,
      "grad_norm": 3.0196926593780518,
      "learning_rate": 9.680046574205947e-05,
      "loss": 0.0866,
      "step": 2920
    },
    {
      "epoch": 0.11810729123495102,
      "grad_norm": 4.441501140594482,
      "learning_rate": 9.679822070466235e-05,
      "loss": 0.1361,
      "step": 2921
    },
    {
      "epoch": 0.11814772509021804,
      "grad_norm": 4.595466613769531,
      "learning_rate": 9.679597490594919e-05,
      "loss": 0.1426,
      "step": 2922
    },
    {
      "epoch": 0.11818815894548505,
      "grad_norm": 5.34171724319458,
      "learning_rate": 9.679372834595648e-05,
      "loss": 0.1522,
      "step": 2923
    },
    {
      "epoch": 0.11822859280075207,
      "grad_norm": 2.730515956878662,
      "learning_rate": 9.67914810247208e-05,
      "loss": 0.0997,
      "step": 2924
    },
    {
      "epoch": 0.11826902665601909,
      "grad_norm": 4.846869468688965,
      "learning_rate": 9.678923294227866e-05,
      "loss": 0.1346,
      "step": 2925
    },
    {
      "epoch": 0.1183094605112861,
      "grad_norm": 3.9147000312805176,
      "learning_rate": 9.678698409866668e-05,
      "loss": 0.1102,
      "step": 2926
    },
    {
      "epoch": 0.11834989436655312,
      "grad_norm": 5.644834995269775,
      "learning_rate": 9.678473449392143e-05,
      "loss": 0.1487,
      "step": 2927
    },
    {
      "epoch": 0.11839032822182013,
      "grad_norm": 11.185388565063477,
      "learning_rate": 9.678248412807949e-05,
      "loss": 0.3721,
      "step": 2928
    },
    {
      "epoch": 0.11843076207708715,
      "grad_norm": 6.348679542541504,
      "learning_rate": 9.67802330011775e-05,
      "loss": 0.1995,
      "step": 2929
    },
    {
      "epoch": 0.11847119593235415,
      "grad_norm": 7.819919586181641,
      "learning_rate": 9.677798111325204e-05,
      "loss": 0.1301,
      "step": 2930
    },
    {
      "epoch": 0.11851162978762117,
      "grad_norm": 2.2132420539855957,
      "learning_rate": 9.677572846433979e-05,
      "loss": 0.1122,
      "step": 2931
    },
    {
      "epoch": 0.11855206364288819,
      "grad_norm": 5.594497203826904,
      "learning_rate": 9.677347505447735e-05,
      "loss": 0.1793,
      "step": 2932
    },
    {
      "epoch": 0.1185924974981552,
      "grad_norm": 2.5468409061431885,
      "learning_rate": 9.677122088370142e-05,
      "loss": 0.1043,
      "step": 2933
    },
    {
      "epoch": 0.11863293135342222,
      "grad_norm": 2.5158779621124268,
      "learning_rate": 9.676896595204867e-05,
      "loss": 0.0716,
      "step": 2934
    },
    {
      "epoch": 0.11867336520868924,
      "grad_norm": 7.755125522613525,
      "learning_rate": 9.676671025955575e-05,
      "loss": 0.1982,
      "step": 2935
    },
    {
      "epoch": 0.11871379906395625,
      "grad_norm": 7.16972541809082,
      "learning_rate": 9.676445380625936e-05,
      "loss": 0.2671,
      "step": 2936
    },
    {
      "epoch": 0.11875423291922327,
      "grad_norm": 4.511007785797119,
      "learning_rate": 9.676219659219624e-05,
      "loss": 0.1064,
      "step": 2937
    },
    {
      "epoch": 0.11879466677449028,
      "grad_norm": 7.883511066436768,
      "learning_rate": 9.67599386174031e-05,
      "loss": 0.223,
      "step": 2938
    },
    {
      "epoch": 0.1188351006297573,
      "grad_norm": 7.464617729187012,
      "learning_rate": 9.675767988191663e-05,
      "loss": 0.1993,
      "step": 2939
    },
    {
      "epoch": 0.11887553448502432,
      "grad_norm": 7.7478556632995605,
      "learning_rate": 9.675542038577363e-05,
      "loss": 0.2791,
      "step": 2940
    },
    {
      "epoch": 0.11891596834029132,
      "grad_norm": 2.9234964847564697,
      "learning_rate": 9.675316012901083e-05,
      "loss": 0.0939,
      "step": 2941
    },
    {
      "epoch": 0.11895640219555834,
      "grad_norm": 4.106435298919678,
      "learning_rate": 9.675089911166501e-05,
      "loss": 0.1419,
      "step": 2942
    },
    {
      "epoch": 0.11899683605082535,
      "grad_norm": 4.083134174346924,
      "learning_rate": 9.674863733377295e-05,
      "loss": 0.1615,
      "step": 2943
    },
    {
      "epoch": 0.11903726990609237,
      "grad_norm": 3.784355878829956,
      "learning_rate": 9.674637479537145e-05,
      "loss": 0.1165,
      "step": 2944
    },
    {
      "epoch": 0.11907770376135939,
      "grad_norm": 3.947577714920044,
      "learning_rate": 9.67441114964973e-05,
      "loss": 0.1572,
      "step": 2945
    },
    {
      "epoch": 0.1191181376166264,
      "grad_norm": 4.7058424949646,
      "learning_rate": 9.674184743718735e-05,
      "loss": 0.3749,
      "step": 2946
    },
    {
      "epoch": 0.11915857147189342,
      "grad_norm": 4.219198703765869,
      "learning_rate": 9.67395826174784e-05,
      "loss": 0.1209,
      "step": 2947
    },
    {
      "epoch": 0.11919900532716043,
      "grad_norm": 5.43949031829834,
      "learning_rate": 9.673731703740729e-05,
      "loss": 0.1161,
      "step": 2948
    },
    {
      "epoch": 0.11923943918242745,
      "grad_norm": 3.8300564289093018,
      "learning_rate": 9.673505069701092e-05,
      "loss": 0.1312,
      "step": 2949
    },
    {
      "epoch": 0.11927987303769447,
      "grad_norm": 9.595552444458008,
      "learning_rate": 9.673278359632611e-05,
      "loss": 0.1875,
      "step": 2950
    },
    {
      "epoch": 0.11932030689296148,
      "grad_norm": 5.195373058319092,
      "learning_rate": 9.673051573538978e-05,
      "loss": 0.1812,
      "step": 2951
    },
    {
      "epoch": 0.11936074074822849,
      "grad_norm": 5.826850414276123,
      "learning_rate": 9.672824711423881e-05,
      "loss": 0.2981,
      "step": 2952
    },
    {
      "epoch": 0.1194011746034955,
      "grad_norm": 5.422791004180908,
      "learning_rate": 9.672597773291009e-05,
      "loss": 0.1378,
      "step": 2953
    },
    {
      "epoch": 0.11944160845876252,
      "grad_norm": 2.1189658641815186,
      "learning_rate": 9.672370759144056e-05,
      "loss": 0.075,
      "step": 2954
    },
    {
      "epoch": 0.11948204231402954,
      "grad_norm": 3.689981698989868,
      "learning_rate": 9.672143668986714e-05,
      "loss": 0.222,
      "step": 2955
    },
    {
      "epoch": 0.11952247616929655,
      "grad_norm": 6.844985008239746,
      "learning_rate": 9.671916502822678e-05,
      "loss": 0.1343,
      "step": 2956
    },
    {
      "epoch": 0.11956291002456357,
      "grad_norm": 2.57948899269104,
      "learning_rate": 9.671689260655642e-05,
      "loss": 0.1418,
      "step": 2957
    },
    {
      "epoch": 0.11960334387983058,
      "grad_norm": 5.1418867111206055,
      "learning_rate": 9.671461942489303e-05,
      "loss": 0.187,
      "step": 2958
    },
    {
      "epoch": 0.1196437777350976,
      "grad_norm": 4.860428333282471,
      "learning_rate": 9.671234548327364e-05,
      "loss": 0.2048,
      "step": 2959
    },
    {
      "epoch": 0.11968421159036462,
      "grad_norm": 3.503718376159668,
      "learning_rate": 9.671007078173517e-05,
      "loss": 0.2176,
      "step": 2960
    },
    {
      "epoch": 0.11972464544563163,
      "grad_norm": 6.7525458335876465,
      "learning_rate": 9.670779532031466e-05,
      "loss": 0.1666,
      "step": 2961
    },
    {
      "epoch": 0.11976507930089864,
      "grad_norm": 4.184514045715332,
      "learning_rate": 9.670551909904912e-05,
      "loss": 0.1141,
      "step": 2962
    },
    {
      "epoch": 0.11980551315616565,
      "grad_norm": 5.12261438369751,
      "learning_rate": 9.67032421179756e-05,
      "loss": 0.2025,
      "step": 2963
    },
    {
      "epoch": 0.11984594701143267,
      "grad_norm": 4.050202369689941,
      "learning_rate": 9.670096437713112e-05,
      "loss": 0.1337,
      "step": 2964
    },
    {
      "epoch": 0.11988638086669968,
      "grad_norm": 6.612905979156494,
      "learning_rate": 9.669868587655274e-05,
      "loss": 0.2901,
      "step": 2965
    },
    {
      "epoch": 0.1199268147219667,
      "grad_norm": 7.494313716888428,
      "learning_rate": 9.669640661627752e-05,
      "loss": 0.2689,
      "step": 2966
    },
    {
      "epoch": 0.11996724857723372,
      "grad_norm": 3.799276351928711,
      "learning_rate": 9.669412659634256e-05,
      "loss": 0.161,
      "step": 2967
    },
    {
      "epoch": 0.12000768243250073,
      "grad_norm": 3.0851922035217285,
      "learning_rate": 9.669184581678492e-05,
      "loss": 0.131,
      "step": 2968
    },
    {
      "epoch": 0.12004811628776775,
      "grad_norm": 6.839437007904053,
      "learning_rate": 9.668956427764172e-05,
      "loss": 0.3246,
      "step": 2969
    },
    {
      "epoch": 0.12008855014303477,
      "grad_norm": 3.828230142593384,
      "learning_rate": 9.668728197895011e-05,
      "loss": 0.2178,
      "step": 2970
    },
    {
      "epoch": 0.12012898399830178,
      "grad_norm": 5.820709228515625,
      "learning_rate": 9.668499892074716e-05,
      "loss": 0.1257,
      "step": 2971
    },
    {
      "epoch": 0.1201694178535688,
      "grad_norm": 8.27147102355957,
      "learning_rate": 9.668271510307006e-05,
      "loss": 0.2794,
      "step": 2972
    },
    {
      "epoch": 0.1202098517088358,
      "grad_norm": 4.4981608390808105,
      "learning_rate": 9.668043052595591e-05,
      "loss": 0.1387,
      "step": 2973
    },
    {
      "epoch": 0.12025028556410282,
      "grad_norm": 5.739572048187256,
      "learning_rate": 9.667814518944192e-05,
      "loss": 0.1338,
      "step": 2974
    },
    {
      "epoch": 0.12029071941936983,
      "grad_norm": 2.2831265926361084,
      "learning_rate": 9.667585909356524e-05,
      "loss": 0.1188,
      "step": 2975
    },
    {
      "epoch": 0.12033115327463685,
      "grad_norm": 8.788087844848633,
      "learning_rate": 9.667357223836311e-05,
      "loss": 0.1895,
      "step": 2976
    },
    {
      "epoch": 0.12037158712990387,
      "grad_norm": 5.529927730560303,
      "learning_rate": 9.667128462387267e-05,
      "loss": 0.2347,
      "step": 2977
    },
    {
      "epoch": 0.12041202098517088,
      "grad_norm": 7.444605350494385,
      "learning_rate": 9.666899625013117e-05,
      "loss": 0.2303,
      "step": 2978
    },
    {
      "epoch": 0.1204524548404379,
      "grad_norm": 2.706998109817505,
      "learning_rate": 9.666670711717582e-05,
      "loss": 0.2456,
      "step": 2979
    },
    {
      "epoch": 0.12049288869570492,
      "grad_norm": 3.930973529815674,
      "learning_rate": 9.66644172250439e-05,
      "loss": 0.1255,
      "step": 2980
    },
    {
      "epoch": 0.12053332255097193,
      "grad_norm": 2.2668471336364746,
      "learning_rate": 9.66621265737726e-05,
      "loss": 0.0546,
      "step": 2981
    },
    {
      "epoch": 0.12057375640623895,
      "grad_norm": 6.2577900886535645,
      "learning_rate": 9.665983516339921e-05,
      "loss": 0.2068,
      "step": 2982
    },
    {
      "epoch": 0.12061419026150597,
      "grad_norm": 6.115801811218262,
      "learning_rate": 9.665754299396104e-05,
      "loss": 0.1935,
      "step": 2983
    },
    {
      "epoch": 0.12065462411677297,
      "grad_norm": 7.309426307678223,
      "learning_rate": 9.665525006549534e-05,
      "loss": 0.2154,
      "step": 2984
    },
    {
      "epoch": 0.12069505797203998,
      "grad_norm": 8.554681777954102,
      "learning_rate": 9.665295637803942e-05,
      "loss": 0.1984,
      "step": 2985
    },
    {
      "epoch": 0.120735491827307,
      "grad_norm": 6.619512557983398,
      "learning_rate": 9.665066193163059e-05,
      "loss": 0.2085,
      "step": 2986
    },
    {
      "epoch": 0.12077592568257402,
      "grad_norm": 8.011435508728027,
      "learning_rate": 9.66483667263062e-05,
      "loss": 0.1813,
      "step": 2987
    },
    {
      "epoch": 0.12081635953784103,
      "grad_norm": 3.2516934871673584,
      "learning_rate": 9.664607076210357e-05,
      "loss": 0.1095,
      "step": 2988
    },
    {
      "epoch": 0.12085679339310805,
      "grad_norm": 4.392043590545654,
      "learning_rate": 9.664377403906004e-05,
      "loss": 0.1706,
      "step": 2989
    },
    {
      "epoch": 0.12089722724837507,
      "grad_norm": 6.85225248336792,
      "learning_rate": 9.6641476557213e-05,
      "loss": 0.3218,
      "step": 2990
    },
    {
      "epoch": 0.12093766110364208,
      "grad_norm": 6.397796630859375,
      "learning_rate": 9.663917831659982e-05,
      "loss": 0.2325,
      "step": 2991
    },
    {
      "epoch": 0.1209780949589091,
      "grad_norm": 4.243820667266846,
      "learning_rate": 9.663687931725784e-05,
      "loss": 0.1134,
      "step": 2992
    },
    {
      "epoch": 0.12101852881417612,
      "grad_norm": 4.1751708984375,
      "learning_rate": 9.663457955922454e-05,
      "loss": 0.1695,
      "step": 2993
    },
    {
      "epoch": 0.12105896266944312,
      "grad_norm": 8.421492576599121,
      "learning_rate": 9.663227904253727e-05,
      "loss": 0.2128,
      "step": 2994
    },
    {
      "epoch": 0.12109939652471013,
      "grad_norm": 3.88852858543396,
      "learning_rate": 9.662997776723348e-05,
      "loss": 0.0923,
      "step": 2995
    },
    {
      "epoch": 0.12113983037997715,
      "grad_norm": 5.185141563415527,
      "learning_rate": 9.66276757333506e-05,
      "loss": 0.0745,
      "step": 2996
    },
    {
      "epoch": 0.12118026423524417,
      "grad_norm": 3.653033971786499,
      "learning_rate": 9.662537294092611e-05,
      "loss": 0.1154,
      "step": 2997
    },
    {
      "epoch": 0.12122069809051118,
      "grad_norm": 2.7818996906280518,
      "learning_rate": 9.662306938999744e-05,
      "loss": 0.1131,
      "step": 2998
    },
    {
      "epoch": 0.1212611319457782,
      "grad_norm": 5.6376471519470215,
      "learning_rate": 9.662076508060206e-05,
      "loss": 0.2322,
      "step": 2999
    },
    {
      "epoch": 0.12130156580104522,
      "grad_norm": 3.066875696182251,
      "learning_rate": 9.661846001277745e-05,
      "loss": 0.1278,
      "step": 3000
    },
    {
      "epoch": 0.12134199965631223,
      "grad_norm": 4.2620978355407715,
      "learning_rate": 9.661615418656115e-05,
      "loss": 0.2307,
      "step": 3001
    },
    {
      "epoch": 0.12138243351157925,
      "grad_norm": 8.69017505645752,
      "learning_rate": 9.661384760199063e-05,
      "loss": 0.1725,
      "step": 3002
    },
    {
      "epoch": 0.12142286736684627,
      "grad_norm": 2.538160562515259,
      "learning_rate": 9.661154025910345e-05,
      "loss": 0.0573,
      "step": 3003
    },
    {
      "epoch": 0.12146330122211328,
      "grad_norm": 5.251650810241699,
      "learning_rate": 9.660923215793712e-05,
      "loss": 0.3084,
      "step": 3004
    },
    {
      "epoch": 0.12150373507738028,
      "grad_norm": 3.497436285018921,
      "learning_rate": 9.66069232985292e-05,
      "loss": 0.1702,
      "step": 3005
    },
    {
      "epoch": 0.1215441689326473,
      "grad_norm": 2.5164244174957275,
      "learning_rate": 9.660461368091724e-05,
      "loss": 0.101,
      "step": 3006
    },
    {
      "epoch": 0.12158460278791432,
      "grad_norm": 8.380972862243652,
      "learning_rate": 9.660230330513881e-05,
      "loss": 0.2614,
      "step": 3007
    },
    {
      "epoch": 0.12162503664318133,
      "grad_norm": 6.186496257781982,
      "learning_rate": 9.659999217123151e-05,
      "loss": 0.1883,
      "step": 3008
    },
    {
      "epoch": 0.12166547049844835,
      "grad_norm": 5.58031702041626,
      "learning_rate": 9.659768027923294e-05,
      "loss": 0.1198,
      "step": 3009
    },
    {
      "epoch": 0.12170590435371537,
      "grad_norm": 8.336541175842285,
      "learning_rate": 9.659536762918068e-05,
      "loss": 0.2246,
      "step": 3010
    },
    {
      "epoch": 0.12174633820898238,
      "grad_norm": 2.1361212730407715,
      "learning_rate": 9.65930542211124e-05,
      "loss": 0.11,
      "step": 3011
    },
    {
      "epoch": 0.1217867720642494,
      "grad_norm": 8.18678092956543,
      "learning_rate": 9.65907400550657e-05,
      "loss": 0.1897,
      "step": 3012
    },
    {
      "epoch": 0.12182720591951642,
      "grad_norm": 9.012898445129395,
      "learning_rate": 9.658842513107824e-05,
      "loss": 0.2251,
      "step": 3013
    },
    {
      "epoch": 0.12186763977478343,
      "grad_norm": 2.8977274894714355,
      "learning_rate": 9.658610944918767e-05,
      "loss": 0.1562,
      "step": 3014
    },
    {
      "epoch": 0.12190807363005045,
      "grad_norm": 4.815328121185303,
      "learning_rate": 9.658379300943166e-05,
      "loss": 0.1728,
      "step": 3015
    },
    {
      "epoch": 0.12194850748531745,
      "grad_norm": 6.522462844848633,
      "learning_rate": 9.65814758118479e-05,
      "loss": 0.2036,
      "step": 3016
    },
    {
      "epoch": 0.12198894134058447,
      "grad_norm": 6.245277404785156,
      "learning_rate": 9.65791578564741e-05,
      "loss": 0.3211,
      "step": 3017
    },
    {
      "epoch": 0.12202937519585148,
      "grad_norm": 5.822898864746094,
      "learning_rate": 9.657683914334797e-05,
      "loss": 0.206,
      "step": 3018
    },
    {
      "epoch": 0.1220698090511185,
      "grad_norm": 8.85268497467041,
      "learning_rate": 9.657451967250719e-05,
      "loss": 0.3526,
      "step": 3019
    },
    {
      "epoch": 0.12211024290638552,
      "grad_norm": 9.268869400024414,
      "learning_rate": 9.657219944398952e-05,
      "loss": 0.2979,
      "step": 3020
    },
    {
      "epoch": 0.12215067676165253,
      "grad_norm": 8.018997192382812,
      "learning_rate": 9.656987845783272e-05,
      "loss": 0.2688,
      "step": 3021
    },
    {
      "epoch": 0.12219111061691955,
      "grad_norm": 2.4185240268707275,
      "learning_rate": 9.656755671407452e-05,
      "loss": 0.1814,
      "step": 3022
    },
    {
      "epoch": 0.12223154447218657,
      "grad_norm": 5.327836513519287,
      "learning_rate": 9.656523421275272e-05,
      "loss": 0.0838,
      "step": 3023
    },
    {
      "epoch": 0.12227197832745358,
      "grad_norm": 3.221057176589966,
      "learning_rate": 9.656291095390507e-05,
      "loss": 0.1411,
      "step": 3024
    },
    {
      "epoch": 0.1223124121827206,
      "grad_norm": 3.6192314624786377,
      "learning_rate": 9.65605869375694e-05,
      "loss": 0.1651,
      "step": 3025
    },
    {
      "epoch": 0.1223528460379876,
      "grad_norm": 7.338139057159424,
      "learning_rate": 9.655826216378347e-05,
      "loss": 0.1501,
      "step": 3026
    },
    {
      "epoch": 0.12239327989325462,
      "grad_norm": 7.896450042724609,
      "learning_rate": 9.655593663258516e-05,
      "loss": 0.2228,
      "step": 3027
    },
    {
      "epoch": 0.12243371374852163,
      "grad_norm": 1.4890258312225342,
      "learning_rate": 9.655361034401225e-05,
      "loss": 0.0708,
      "step": 3028
    },
    {
      "epoch": 0.12247414760378865,
      "grad_norm": 6.506618022918701,
      "learning_rate": 9.65512832981026e-05,
      "loss": 0.1974,
      "step": 3029
    },
    {
      "epoch": 0.12251458145905567,
      "grad_norm": 4.868554592132568,
      "learning_rate": 9.654895549489407e-05,
      "loss": 0.2038,
      "step": 3030
    },
    {
      "epoch": 0.12255501531432268,
      "grad_norm": 3.964310646057129,
      "learning_rate": 9.654662693442456e-05,
      "loss": 0.1076,
      "step": 3031
    },
    {
      "epoch": 0.1225954491695897,
      "grad_norm": 3.3144307136535645,
      "learning_rate": 9.65442976167319e-05,
      "loss": 0.0975,
      "step": 3032
    },
    {
      "epoch": 0.12263588302485671,
      "grad_norm": 3.2532970905303955,
      "learning_rate": 9.654196754185398e-05,
      "loss": 0.2211,
      "step": 3033
    },
    {
      "epoch": 0.12267631688012373,
      "grad_norm": 3.789276599884033,
      "learning_rate": 9.653963670982876e-05,
      "loss": 0.1703,
      "step": 3034
    },
    {
      "epoch": 0.12271675073539075,
      "grad_norm": 6.969799518585205,
      "learning_rate": 9.653730512069412e-05,
      "loss": 0.1844,
      "step": 3035
    },
    {
      "epoch": 0.12275718459065776,
      "grad_norm": 3.704760789871216,
      "learning_rate": 9.653497277448799e-05,
      "loss": 0.1201,
      "step": 3036
    },
    {
      "epoch": 0.12279761844592477,
      "grad_norm": 8.86923885345459,
      "learning_rate": 9.653263967124835e-05,
      "loss": 0.1578,
      "step": 3037
    },
    {
      "epoch": 0.12283805230119178,
      "grad_norm": 8.414907455444336,
      "learning_rate": 9.65303058110131e-05,
      "loss": 0.3085,
      "step": 3038
    },
    {
      "epoch": 0.1228784861564588,
      "grad_norm": 4.7198872566223145,
      "learning_rate": 9.652797119382025e-05,
      "loss": 0.261,
      "step": 3039
    },
    {
      "epoch": 0.12291892001172582,
      "grad_norm": 6.661331653594971,
      "learning_rate": 9.652563581970774e-05,
      "loss": 0.1472,
      "step": 3040
    },
    {
      "epoch": 0.12295935386699283,
      "grad_norm": 3.5777363777160645,
      "learning_rate": 9.65232996887136e-05,
      "loss": 0.1988,
      "step": 3041
    },
    {
      "epoch": 0.12299978772225985,
      "grad_norm": 5.789548397064209,
      "learning_rate": 9.652096280087581e-05,
      "loss": 0.1429,
      "step": 3042
    },
    {
      "epoch": 0.12304022157752686,
      "grad_norm": 3.320195436477661,
      "learning_rate": 9.65186251562324e-05,
      "loss": 0.185,
      "step": 3043
    },
    {
      "epoch": 0.12308065543279388,
      "grad_norm": 8.352540969848633,
      "learning_rate": 9.651628675482141e-05,
      "loss": 0.2198,
      "step": 3044
    },
    {
      "epoch": 0.1231210892880609,
      "grad_norm": 6.598633289337158,
      "learning_rate": 9.651394759668084e-05,
      "loss": 0.212,
      "step": 3045
    },
    {
      "epoch": 0.12316152314332791,
      "grad_norm": 3.2555322647094727,
      "learning_rate": 9.65116076818488e-05,
      "loss": 0.1145,
      "step": 3046
    },
    {
      "epoch": 0.12320195699859493,
      "grad_norm": 4.1114091873168945,
      "learning_rate": 9.65092670103633e-05,
      "loss": 0.103,
      "step": 3047
    },
    {
      "epoch": 0.12324239085386193,
      "grad_norm": 2.6677558422088623,
      "learning_rate": 9.650692558226246e-05,
      "loss": 0.091,
      "step": 3048
    },
    {
      "epoch": 0.12328282470912895,
      "grad_norm": 7.637884140014648,
      "learning_rate": 9.650458339758435e-05,
      "loss": 0.2201,
      "step": 3049
    },
    {
      "epoch": 0.12332325856439597,
      "grad_norm": 7.138638019561768,
      "learning_rate": 9.650224045636706e-05,
      "loss": 0.1912,
      "step": 3050
    },
    {
      "epoch": 0.12336369241966298,
      "grad_norm": 4.282451152801514,
      "learning_rate": 9.649989675864874e-05,
      "loss": 0.2658,
      "step": 3051
    },
    {
      "epoch": 0.12340412627493,
      "grad_norm": 5.736961841583252,
      "learning_rate": 9.649755230446751e-05,
      "loss": 0.2306,
      "step": 3052
    },
    {
      "epoch": 0.12344456013019701,
      "grad_norm": 7.713468074798584,
      "learning_rate": 9.649520709386149e-05,
      "loss": 0.2396,
      "step": 3053
    },
    {
      "epoch": 0.12348499398546403,
      "grad_norm": 8.221893310546875,
      "learning_rate": 9.649286112686884e-05,
      "loss": 0.2756,
      "step": 3054
    },
    {
      "epoch": 0.12352542784073105,
      "grad_norm": 10.16769027709961,
      "learning_rate": 9.649051440352771e-05,
      "loss": 0.3091,
      "step": 3055
    },
    {
      "epoch": 0.12356586169599806,
      "grad_norm": 7.066163063049316,
      "learning_rate": 9.648816692387631e-05,
      "loss": 0.157,
      "step": 3056
    },
    {
      "epoch": 0.12360629555126508,
      "grad_norm": 7.838068008422852,
      "learning_rate": 9.648581868795281e-05,
      "loss": 0.2638,
      "step": 3057
    },
    {
      "epoch": 0.12364672940653208,
      "grad_norm": 2.9372270107269287,
      "learning_rate": 9.64834696957954e-05,
      "loss": 0.1646,
      "step": 3058
    },
    {
      "epoch": 0.1236871632617991,
      "grad_norm": 3.936868190765381,
      "learning_rate": 9.648111994744233e-05,
      "loss": 0.1942,
      "step": 3059
    },
    {
      "epoch": 0.12372759711706612,
      "grad_norm": 7.217771530151367,
      "learning_rate": 9.647876944293178e-05,
      "loss": 0.2212,
      "step": 3060
    },
    {
      "epoch": 0.12376803097233313,
      "grad_norm": 2.458962917327881,
      "learning_rate": 9.647641818230202e-05,
      "loss": 0.1743,
      "step": 3061
    },
    {
      "epoch": 0.12380846482760015,
      "grad_norm": 3.130805730819702,
      "learning_rate": 9.647406616559129e-05,
      "loss": 0.1443,
      "step": 3062
    },
    {
      "epoch": 0.12384889868286716,
      "grad_norm": 11.035645484924316,
      "learning_rate": 9.647171339283785e-05,
      "loss": 0.2161,
      "step": 3063
    },
    {
      "epoch": 0.12388933253813418,
      "grad_norm": 6.409191608428955,
      "learning_rate": 9.646935986407998e-05,
      "loss": 0.2962,
      "step": 3064
    },
    {
      "epoch": 0.1239297663934012,
      "grad_norm": 7.301177978515625,
      "learning_rate": 9.646700557935597e-05,
      "loss": 0.169,
      "step": 3065
    },
    {
      "epoch": 0.12397020024866821,
      "grad_norm": 8.623258590698242,
      "learning_rate": 9.64646505387041e-05,
      "loss": 0.1653,
      "step": 3066
    },
    {
      "epoch": 0.12401063410393523,
      "grad_norm": 7.678514003753662,
      "learning_rate": 9.646229474216272e-05,
      "loss": 0.2593,
      "step": 3067
    },
    {
      "epoch": 0.12405106795920225,
      "grad_norm": 7.988147735595703,
      "learning_rate": 9.645993818977011e-05,
      "loss": 0.344,
      "step": 3068
    },
    {
      "epoch": 0.12409150181446925,
      "grad_norm": 4.161752223968506,
      "learning_rate": 9.645758088156465e-05,
      "loss": 0.2419,
      "step": 3069
    },
    {
      "epoch": 0.12413193566973627,
      "grad_norm": 3.413909673690796,
      "learning_rate": 9.645522281758464e-05,
      "loss": 0.1846,
      "step": 3070
    },
    {
      "epoch": 0.12417236952500328,
      "grad_norm": 3.3409903049468994,
      "learning_rate": 9.645286399786848e-05,
      "loss": 0.2363,
      "step": 3071
    },
    {
      "epoch": 0.1242128033802703,
      "grad_norm": 7.244366645812988,
      "learning_rate": 9.645050442245453e-05,
      "loss": 0.157,
      "step": 3072
    },
    {
      "epoch": 0.12425323723553731,
      "grad_norm": 5.298910617828369,
      "learning_rate": 9.644814409138118e-05,
      "loss": 0.1283,
      "step": 3073
    },
    {
      "epoch": 0.12429367109080433,
      "grad_norm": 7.6536688804626465,
      "learning_rate": 9.644578300468683e-05,
      "loss": 0.288,
      "step": 3074
    },
    {
      "epoch": 0.12433410494607135,
      "grad_norm": 4.787538051605225,
      "learning_rate": 9.644342116240989e-05,
      "loss": 0.2343,
      "step": 3075
    },
    {
      "epoch": 0.12437453880133836,
      "grad_norm": 3.7564480304718018,
      "learning_rate": 9.644105856458876e-05,
      "loss": 0.1551,
      "step": 3076
    },
    {
      "epoch": 0.12441497265660538,
      "grad_norm": 6.956244468688965,
      "learning_rate": 9.64386952112619e-05,
      "loss": 0.2851,
      "step": 3077
    },
    {
      "epoch": 0.1244554065118724,
      "grad_norm": 8.253607749938965,
      "learning_rate": 9.643633110246775e-05,
      "loss": 0.1539,
      "step": 3078
    },
    {
      "epoch": 0.12449584036713941,
      "grad_norm": 7.334596633911133,
      "learning_rate": 9.643396623824476e-05,
      "loss": 0.233,
      "step": 3079
    },
    {
      "epoch": 0.12453627422240642,
      "grad_norm": 4.545625686645508,
      "learning_rate": 9.643160061863143e-05,
      "loss": 0.2549,
      "step": 3080
    },
    {
      "epoch": 0.12457670807767343,
      "grad_norm": 2.3526532649993896,
      "learning_rate": 9.642923424366622e-05,
      "loss": 0.1412,
      "step": 3081
    },
    {
      "epoch": 0.12461714193294045,
      "grad_norm": 7.09621000289917,
      "learning_rate": 9.642686711338762e-05,
      "loss": 0.2273,
      "step": 3082
    },
    {
      "epoch": 0.12465757578820746,
      "grad_norm": 8.304969787597656,
      "learning_rate": 9.642449922783416e-05,
      "loss": 0.2045,
      "step": 3083
    },
    {
      "epoch": 0.12469800964347448,
      "grad_norm": 4.975228786468506,
      "learning_rate": 9.642213058704434e-05,
      "loss": 0.147,
      "step": 3084
    },
    {
      "epoch": 0.1247384434987415,
      "grad_norm": 2.148052215576172,
      "learning_rate": 9.641976119105672e-05,
      "loss": 0.1881,
      "step": 3085
    },
    {
      "epoch": 0.12477887735400851,
      "grad_norm": 4.729580879211426,
      "learning_rate": 9.641739103990983e-05,
      "loss": 0.199,
      "step": 3086
    },
    {
      "epoch": 0.12481931120927553,
      "grad_norm": 2.9985358715057373,
      "learning_rate": 9.641502013364222e-05,
      "loss": 0.0836,
      "step": 3087
    },
    {
      "epoch": 0.12485974506454255,
      "grad_norm": 2.899381399154663,
      "learning_rate": 9.641264847229245e-05,
      "loss": 0.2134,
      "step": 3088
    },
    {
      "epoch": 0.12490017891980956,
      "grad_norm": 4.262881278991699,
      "learning_rate": 9.641027605589915e-05,
      "loss": 0.1309,
      "step": 3089
    },
    {
      "epoch": 0.12494061277507656,
      "grad_norm": 3.340488910675049,
      "learning_rate": 9.640790288450087e-05,
      "loss": 0.1789,
      "step": 3090
    },
    {
      "epoch": 0.12498104663034358,
      "grad_norm": 6.830298900604248,
      "learning_rate": 9.640552895813624e-05,
      "loss": 0.1735,
      "step": 3091
    },
    {
      "epoch": 0.1250214804856106,
      "grad_norm": 7.962432384490967,
      "learning_rate": 9.640315427684386e-05,
      "loss": 0.2176,
      "step": 3092
    },
    {
      "epoch": 0.12506191434087763,
      "grad_norm": 3.2954540252685547,
      "learning_rate": 9.640077884066236e-05,
      "loss": 0.0876,
      "step": 3093
    },
    {
      "epoch": 0.12510234819614463,
      "grad_norm": 4.549011707305908,
      "learning_rate": 9.639840264963042e-05,
      "loss": 0.136,
      "step": 3094
    },
    {
      "epoch": 0.12514278205141166,
      "grad_norm": 4.3669843673706055,
      "learning_rate": 9.639602570378665e-05,
      "loss": 0.1121,
      "step": 3095
    },
    {
      "epoch": 0.12518321590667866,
      "grad_norm": 5.089741230010986,
      "learning_rate": 9.639364800316976e-05,
      "loss": 0.2995,
      "step": 3096
    },
    {
      "epoch": 0.12522364976194567,
      "grad_norm": 7.412095069885254,
      "learning_rate": 9.63912695478184e-05,
      "loss": 0.2127,
      "step": 3097
    },
    {
      "epoch": 0.1252640836172127,
      "grad_norm": 5.960890769958496,
      "learning_rate": 9.638889033777128e-05,
      "loss": 0.2024,
      "step": 3098
    },
    {
      "epoch": 0.1253045174724797,
      "grad_norm": 5.155802249908447,
      "learning_rate": 9.638651037306709e-05,
      "loss": 0.2068,
      "step": 3099
    },
    {
      "epoch": 0.12534495132774673,
      "grad_norm": 5.71253776550293,
      "learning_rate": 9.638412965374455e-05,
      "loss": 0.1545,
      "step": 3100
    },
    {
      "epoch": 0.12538538518301373,
      "grad_norm": 3.6299359798431396,
      "learning_rate": 9.63817481798424e-05,
      "loss": 0.1553,
      "step": 3101
    },
    {
      "epoch": 0.12542581903828076,
      "grad_norm": 7.880157947540283,
      "learning_rate": 9.637936595139939e-05,
      "loss": 0.3067,
      "step": 3102
    },
    {
      "epoch": 0.12546625289354776,
      "grad_norm": 3.5101027488708496,
      "learning_rate": 9.637698296845426e-05,
      "loss": 0.1968,
      "step": 3103
    },
    {
      "epoch": 0.1255066867488148,
      "grad_norm": 5.385390281677246,
      "learning_rate": 9.637459923104576e-05,
      "loss": 0.2056,
      "step": 3104
    },
    {
      "epoch": 0.1255471206040818,
      "grad_norm": 4.1077680587768555,
      "learning_rate": 9.637221473921269e-05,
      "loss": 0.092,
      "step": 3105
    },
    {
      "epoch": 0.1255875544593488,
      "grad_norm": 3.250027894973755,
      "learning_rate": 9.636982949299385e-05,
      "loss": 0.1046,
      "step": 3106
    },
    {
      "epoch": 0.12562798831461583,
      "grad_norm": 4.66965913772583,
      "learning_rate": 9.636744349242804e-05,
      "loss": 0.2068,
      "step": 3107
    },
    {
      "epoch": 0.12566842216988283,
      "grad_norm": 2.465945243835449,
      "learning_rate": 9.636505673755404e-05,
      "loss": 0.1035,
      "step": 3108
    },
    {
      "epoch": 0.12570885602514986,
      "grad_norm": 4.304436683654785,
      "learning_rate": 9.636266922841072e-05,
      "loss": 0.2204,
      "step": 3109
    },
    {
      "epoch": 0.12574928988041686,
      "grad_norm": 3.3799431324005127,
      "learning_rate": 9.636028096503691e-05,
      "loss": 0.2108,
      "step": 3110
    },
    {
      "epoch": 0.1257897237356839,
      "grad_norm": 6.574578285217285,
      "learning_rate": 9.635789194747145e-05,
      "loss": 0.2165,
      "step": 3111
    },
    {
      "epoch": 0.1258301575909509,
      "grad_norm": 4.579308986663818,
      "learning_rate": 9.635550217575319e-05,
      "loss": 0.1535,
      "step": 3112
    },
    {
      "epoch": 0.12587059144621793,
      "grad_norm": 2.919924020767212,
      "learning_rate": 9.635311164992107e-05,
      "loss": 0.116,
      "step": 3113
    },
    {
      "epoch": 0.12591102530148493,
      "grad_norm": 6.491866111755371,
      "learning_rate": 9.635072037001391e-05,
      "loss": 0.1903,
      "step": 3114
    },
    {
      "epoch": 0.12595145915675196,
      "grad_norm": 6.042548179626465,
      "learning_rate": 9.634832833607066e-05,
      "loss": 0.1579,
      "step": 3115
    },
    {
      "epoch": 0.12599189301201896,
      "grad_norm": 3.5038857460021973,
      "learning_rate": 9.634593554813019e-05,
      "loss": 0.1416,
      "step": 3116
    },
    {
      "epoch": 0.12603232686728597,
      "grad_norm": 2.8658742904663086,
      "learning_rate": 9.634354200623146e-05,
      "loss": 0.0967,
      "step": 3117
    },
    {
      "epoch": 0.126072760722553,
      "grad_norm": 3.0398120880126953,
      "learning_rate": 9.63411477104134e-05,
      "loss": 0.1286,
      "step": 3118
    },
    {
      "epoch": 0.12611319457782,
      "grad_norm": 4.546473503112793,
      "learning_rate": 9.633875266071496e-05,
      "loss": 0.1374,
      "step": 3119
    },
    {
      "epoch": 0.12615362843308703,
      "grad_norm": 4.931397914886475,
      "learning_rate": 9.63363568571751e-05,
      "loss": 0.339,
      "step": 3120
    },
    {
      "epoch": 0.12619406228835403,
      "grad_norm": 6.361161708831787,
      "learning_rate": 9.633396029983279e-05,
      "loss": 0.1421,
      "step": 3121
    },
    {
      "epoch": 0.12623449614362106,
      "grad_norm": 7.341002464294434,
      "learning_rate": 9.633156298872703e-05,
      "loss": 0.2715,
      "step": 3122
    },
    {
      "epoch": 0.12627492999888806,
      "grad_norm": 6.57306432723999,
      "learning_rate": 9.632916492389681e-05,
      "loss": 0.2057,
      "step": 3123
    },
    {
      "epoch": 0.1263153638541551,
      "grad_norm": 6.175103187561035,
      "learning_rate": 9.632676610538115e-05,
      "loss": 0.1668,
      "step": 3124
    },
    {
      "epoch": 0.1263557977094221,
      "grad_norm": 6.517489910125732,
      "learning_rate": 9.632436653321905e-05,
      "loss": 0.2883,
      "step": 3125
    },
    {
      "epoch": 0.12639623156468913,
      "grad_norm": 7.410542011260986,
      "learning_rate": 9.632196620744958e-05,
      "loss": 0.1937,
      "step": 3126
    },
    {
      "epoch": 0.12643666541995613,
      "grad_norm": 5.421642303466797,
      "learning_rate": 9.631956512811177e-05,
      "loss": 0.2284,
      "step": 3127
    },
    {
      "epoch": 0.12647709927522313,
      "grad_norm": 3.102980852127075,
      "learning_rate": 9.63171632952447e-05,
      "loss": 0.1023,
      "step": 3128
    },
    {
      "epoch": 0.12651753313049016,
      "grad_norm": 4.849391460418701,
      "learning_rate": 9.63147607088874e-05,
      "loss": 0.1536,
      "step": 3129
    },
    {
      "epoch": 0.12655796698575716,
      "grad_norm": 4.618221759796143,
      "learning_rate": 9.6312357369079e-05,
      "loss": 0.1662,
      "step": 3130
    },
    {
      "epoch": 0.1265984008410242,
      "grad_norm": 2.865647315979004,
      "learning_rate": 9.630995327585858e-05,
      "loss": 0.1105,
      "step": 3131
    },
    {
      "epoch": 0.1266388346962912,
      "grad_norm": 6.706053733825684,
      "learning_rate": 9.630754842926524e-05,
      "loss": 0.2243,
      "step": 3132
    },
    {
      "epoch": 0.12667926855155823,
      "grad_norm": 1.8924041986465454,
      "learning_rate": 9.630514282933814e-05,
      "loss": 0.0911,
      "step": 3133
    },
    {
      "epoch": 0.12671970240682523,
      "grad_norm": 2.764437675476074,
      "learning_rate": 9.630273647611637e-05,
      "loss": 0.1223,
      "step": 3134
    },
    {
      "epoch": 0.12676013626209226,
      "grad_norm": 3.457174777984619,
      "learning_rate": 9.630032936963909e-05,
      "loss": 0.0992,
      "step": 3135
    },
    {
      "epoch": 0.12680057011735926,
      "grad_norm": 5.565229415893555,
      "learning_rate": 9.629792150994545e-05,
      "loss": 0.1446,
      "step": 3136
    },
    {
      "epoch": 0.1268410039726263,
      "grad_norm": 3.081270217895508,
      "learning_rate": 9.629551289707466e-05,
      "loss": 0.2114,
      "step": 3137
    },
    {
      "epoch": 0.1268814378278933,
      "grad_norm": 2.492976427078247,
      "learning_rate": 9.629310353106586e-05,
      "loss": 0.1174,
      "step": 3138
    },
    {
      "epoch": 0.1269218716831603,
      "grad_norm": 5.924192428588867,
      "learning_rate": 9.629069341195829e-05,
      "loss": 0.2741,
      "step": 3139
    },
    {
      "epoch": 0.12696230553842733,
      "grad_norm": 3.0183067321777344,
      "learning_rate": 9.628828253979111e-05,
      "loss": 0.1486,
      "step": 3140
    },
    {
      "epoch": 0.12700273939369433,
      "grad_norm": 3.814072608947754,
      "learning_rate": 9.628587091460355e-05,
      "loss": 0.1631,
      "step": 3141
    },
    {
      "epoch": 0.12704317324896136,
      "grad_norm": 5.795636177062988,
      "learning_rate": 9.628345853643488e-05,
      "loss": 0.2717,
      "step": 3142
    },
    {
      "epoch": 0.12708360710422836,
      "grad_norm": 4.301156997680664,
      "learning_rate": 9.628104540532431e-05,
      "loss": 0.3189,
      "step": 3143
    },
    {
      "epoch": 0.1271240409594954,
      "grad_norm": 3.710374116897583,
      "learning_rate": 9.62786315213111e-05,
      "loss": 0.0974,
      "step": 3144
    },
    {
      "epoch": 0.1271644748147624,
      "grad_norm": 4.541816711425781,
      "learning_rate": 9.627621688443454e-05,
      "loss": 0.2999,
      "step": 3145
    },
    {
      "epoch": 0.12720490867002943,
      "grad_norm": 2.7101540565490723,
      "learning_rate": 9.627380149473388e-05,
      "loss": 0.1467,
      "step": 3146
    },
    {
      "epoch": 0.12724534252529643,
      "grad_norm": 2.0789730548858643,
      "learning_rate": 9.627138535224843e-05,
      "loss": 0.0908,
      "step": 3147
    },
    {
      "epoch": 0.12728577638056346,
      "grad_norm": 5.319098949432373,
      "learning_rate": 9.626896845701751e-05,
      "loss": 0.1157,
      "step": 3148
    },
    {
      "epoch": 0.12732621023583046,
      "grad_norm": 3.898212194442749,
      "learning_rate": 9.62665508090804e-05,
      "loss": 0.1023,
      "step": 3149
    },
    {
      "epoch": 0.12736664409109746,
      "grad_norm": 12.242679595947266,
      "learning_rate": 9.626413240847649e-05,
      "loss": 0.3182,
      "step": 3150
    },
    {
      "epoch": 0.1274070779463645,
      "grad_norm": 6.351028919219971,
      "learning_rate": 9.626171325524506e-05,
      "loss": 0.1751,
      "step": 3151
    },
    {
      "epoch": 0.1274475118016315,
      "grad_norm": 5.071272373199463,
      "learning_rate": 9.62592933494255e-05,
      "loss": 0.1104,
      "step": 3152
    },
    {
      "epoch": 0.12748794565689853,
      "grad_norm": 6.054483413696289,
      "learning_rate": 9.625687269105716e-05,
      "loss": 0.2521,
      "step": 3153
    },
    {
      "epoch": 0.12752837951216553,
      "grad_norm": 4.155210494995117,
      "learning_rate": 9.625445128017945e-05,
      "loss": 0.1531,
      "step": 3154
    },
    {
      "epoch": 0.12756881336743256,
      "grad_norm": 2.6578447818756104,
      "learning_rate": 9.625202911683172e-05,
      "loss": 0.1153,
      "step": 3155
    },
    {
      "epoch": 0.12760924722269956,
      "grad_norm": 9.211869239807129,
      "learning_rate": 9.62496062010534e-05,
      "loss": 0.1891,
      "step": 3156
    },
    {
      "epoch": 0.1276496810779666,
      "grad_norm": 8.357471466064453,
      "learning_rate": 9.624718253288391e-05,
      "loss": 0.1749,
      "step": 3157
    },
    {
      "epoch": 0.1276901149332336,
      "grad_norm": 1.9742708206176758,
      "learning_rate": 9.624475811236266e-05,
      "loss": 0.1452,
      "step": 3158
    },
    {
      "epoch": 0.12773054878850063,
      "grad_norm": 1.7911368608474731,
      "learning_rate": 9.62423329395291e-05,
      "loss": 0.106,
      "step": 3159
    },
    {
      "epoch": 0.12777098264376763,
      "grad_norm": 3.394331693649292,
      "learning_rate": 9.623990701442269e-05,
      "loss": 0.1732,
      "step": 3160
    },
    {
      "epoch": 0.12781141649903463,
      "grad_norm": 3.867633581161499,
      "learning_rate": 9.623748033708287e-05,
      "loss": 0.1412,
      "step": 3161
    },
    {
      "epoch": 0.12785185035430166,
      "grad_norm": 9.76225471496582,
      "learning_rate": 9.623505290754915e-05,
      "loss": 0.2277,
      "step": 3162
    },
    {
      "epoch": 0.12789228420956866,
      "grad_norm": 2.505388021469116,
      "learning_rate": 9.6232624725861e-05,
      "loss": 0.103,
      "step": 3163
    },
    {
      "epoch": 0.1279327180648357,
      "grad_norm": 3.115339994430542,
      "learning_rate": 9.623019579205793e-05,
      "loss": 0.1068,
      "step": 3164
    },
    {
      "epoch": 0.1279731519201027,
      "grad_norm": 2.4636292457580566,
      "learning_rate": 9.622776610617946e-05,
      "loss": 0.1308,
      "step": 3165
    },
    {
      "epoch": 0.12801358577536973,
      "grad_norm": 9.397582054138184,
      "learning_rate": 9.622533566826508e-05,
      "loss": 0.2789,
      "step": 3166
    },
    {
      "epoch": 0.12805401963063673,
      "grad_norm": 3.361727714538574,
      "learning_rate": 9.622290447835436e-05,
      "loss": 0.2276,
      "step": 3167
    },
    {
      "epoch": 0.12809445348590376,
      "grad_norm": 3.6538307666778564,
      "learning_rate": 9.622047253648685e-05,
      "loss": 0.1503,
      "step": 3168
    },
    {
      "epoch": 0.12813488734117076,
      "grad_norm": 6.195231914520264,
      "learning_rate": 9.621803984270213e-05,
      "loss": 0.2081,
      "step": 3169
    },
    {
      "epoch": 0.12817532119643776,
      "grad_norm": 6.89579963684082,
      "learning_rate": 9.621560639703973e-05,
      "loss": 0.2191,
      "step": 3170
    },
    {
      "epoch": 0.1282157550517048,
      "grad_norm": 7.5912580490112305,
      "learning_rate": 9.621317219953926e-05,
      "loss": 0.3314,
      "step": 3171
    },
    {
      "epoch": 0.1282561889069718,
      "grad_norm": 4.156790733337402,
      "learning_rate": 9.621073725024032e-05,
      "loss": 0.099,
      "step": 3172
    },
    {
      "epoch": 0.12829662276223883,
      "grad_norm": 6.763216495513916,
      "learning_rate": 9.620830154918254e-05,
      "loss": 0.1553,
      "step": 3173
    },
    {
      "epoch": 0.12833705661750583,
      "grad_norm": 7.126609802246094,
      "learning_rate": 9.620586509640551e-05,
      "loss": 0.2449,
      "step": 3174
    },
    {
      "epoch": 0.12837749047277286,
      "grad_norm": 3.513136863708496,
      "learning_rate": 9.62034278919489e-05,
      "loss": 0.1206,
      "step": 3175
    },
    {
      "epoch": 0.12841792432803986,
      "grad_norm": 7.130175590515137,
      "learning_rate": 9.620098993585232e-05,
      "loss": 0.1746,
      "step": 3176
    },
    {
      "epoch": 0.1284583581833069,
      "grad_norm": 8.63797664642334,
      "learning_rate": 9.619855122815546e-05,
      "loss": 0.2465,
      "step": 3177
    },
    {
      "epoch": 0.1284987920385739,
      "grad_norm": 2.8862836360931396,
      "learning_rate": 9.6196111768898e-05,
      "loss": 0.1459,
      "step": 3178
    },
    {
      "epoch": 0.12853922589384092,
      "grad_norm": 6.611880779266357,
      "learning_rate": 9.61936715581196e-05,
      "loss": 0.3094,
      "step": 3179
    },
    {
      "epoch": 0.12857965974910793,
      "grad_norm": 8.889134407043457,
      "learning_rate": 9.619123059585997e-05,
      "loss": 0.2001,
      "step": 3180
    },
    {
      "epoch": 0.12862009360437493,
      "grad_norm": 8.524028778076172,
      "learning_rate": 9.618878888215881e-05,
      "loss": 0.2345,
      "step": 3181
    },
    {
      "epoch": 0.12866052745964196,
      "grad_norm": 3.461775541305542,
      "learning_rate": 9.618634641705587e-05,
      "loss": 0.0984,
      "step": 3182
    },
    {
      "epoch": 0.12870096131490896,
      "grad_norm": 4.388386249542236,
      "learning_rate": 9.618390320059084e-05,
      "loss": 0.1346,
      "step": 3183
    },
    {
      "epoch": 0.128741395170176,
      "grad_norm": 3.3514535427093506,
      "learning_rate": 9.618145923280351e-05,
      "loss": 0.0602,
      "step": 3184
    },
    {
      "epoch": 0.128781829025443,
      "grad_norm": 3.492845296859741,
      "learning_rate": 9.617901451373361e-05,
      "loss": 0.1078,
      "step": 3185
    },
    {
      "epoch": 0.12882226288071003,
      "grad_norm": 5.799691677093506,
      "learning_rate": 9.617656904342091e-05,
      "loss": 0.1075,
      "step": 3186
    },
    {
      "epoch": 0.12886269673597703,
      "grad_norm": 2.1168739795684814,
      "learning_rate": 9.617412282190523e-05,
      "loss": 0.1062,
      "step": 3187
    },
    {
      "epoch": 0.12890313059124406,
      "grad_norm": 7.542407989501953,
      "learning_rate": 9.61716758492263e-05,
      "loss": 0.1863,
      "step": 3188
    },
    {
      "epoch": 0.12894356444651106,
      "grad_norm": 4.5323805809021,
      "learning_rate": 9.6169228125424e-05,
      "loss": 0.1605,
      "step": 3189
    },
    {
      "epoch": 0.1289839983017781,
      "grad_norm": 6.319133758544922,
      "learning_rate": 9.616677965053811e-05,
      "loss": 0.1157,
      "step": 3190
    },
    {
      "epoch": 0.1290244321570451,
      "grad_norm": 8.119425773620605,
      "learning_rate": 9.616433042460845e-05,
      "loss": 0.1669,
      "step": 3191
    },
    {
      "epoch": 0.1290648660123121,
      "grad_norm": 3.649080276489258,
      "learning_rate": 9.61618804476749e-05,
      "loss": 0.2361,
      "step": 3192
    },
    {
      "epoch": 0.12910529986757913,
      "grad_norm": 2.198651075363159,
      "learning_rate": 9.61594297197773e-05,
      "loss": 0.0916,
      "step": 3193
    },
    {
      "epoch": 0.12914573372284613,
      "grad_norm": 4.540441989898682,
      "learning_rate": 9.615697824095549e-05,
      "loss": 0.1549,
      "step": 3194
    },
    {
      "epoch": 0.12918616757811316,
      "grad_norm": 4.165707111358643,
      "learning_rate": 9.61545260112494e-05,
      "loss": 0.1649,
      "step": 3195
    },
    {
      "epoch": 0.12922660143338016,
      "grad_norm": 8.447750091552734,
      "learning_rate": 9.615207303069889e-05,
      "loss": 0.1209,
      "step": 3196
    },
    {
      "epoch": 0.1292670352886472,
      "grad_norm": 9.771206855773926,
      "learning_rate": 9.614961929934388e-05,
      "loss": 0.2772,
      "step": 3197
    },
    {
      "epoch": 0.1293074691439142,
      "grad_norm": 6.438025951385498,
      "learning_rate": 9.614716481722428e-05,
      "loss": 0.1686,
      "step": 3198
    },
    {
      "epoch": 0.12934790299918122,
      "grad_norm": 7.078266143798828,
      "learning_rate": 9.614470958438002e-05,
      "loss": 0.1806,
      "step": 3199
    },
    {
      "epoch": 0.12938833685444823,
      "grad_norm": 5.145882606506348,
      "learning_rate": 9.614225360085106e-05,
      "loss": 0.1242,
      "step": 3200
    },
    {
      "epoch": 0.12942877070971526,
      "grad_norm": 3.7052180767059326,
      "learning_rate": 9.613979686667734e-05,
      "loss": 0.0842,
      "step": 3201
    },
    {
      "epoch": 0.12946920456498226,
      "grad_norm": 6.253951549530029,
      "learning_rate": 9.61373393818988e-05,
      "loss": 0.2009,
      "step": 3202
    },
    {
      "epoch": 0.12950963842024926,
      "grad_norm": 4.835132598876953,
      "learning_rate": 9.613488114655546e-05,
      "loss": 0.3234,
      "step": 3203
    },
    {
      "epoch": 0.1295500722755163,
      "grad_norm": 3.673395872116089,
      "learning_rate": 9.613242216068729e-05,
      "loss": 0.1415,
      "step": 3204
    },
    {
      "epoch": 0.1295905061307833,
      "grad_norm": 5.0938520431518555,
      "learning_rate": 9.612996242433429e-05,
      "loss": 0.1194,
      "step": 3205
    },
    {
      "epoch": 0.12963093998605033,
      "grad_norm": 5.03550910949707,
      "learning_rate": 9.612750193753651e-05,
      "loss": 0.2433,
      "step": 3206
    },
    {
      "epoch": 0.12967137384131733,
      "grad_norm": 7.57958459854126,
      "learning_rate": 9.612504070033391e-05,
      "loss": 0.1918,
      "step": 3207
    },
    {
      "epoch": 0.12971180769658436,
      "grad_norm": 7.207999229431152,
      "learning_rate": 9.61225787127666e-05,
      "loss": 0.3087,
      "step": 3208
    },
    {
      "epoch": 0.12975224155185136,
      "grad_norm": 5.037413597106934,
      "learning_rate": 9.612011597487458e-05,
      "loss": 0.2426,
      "step": 3209
    },
    {
      "epoch": 0.1297926754071184,
      "grad_norm": 3.050163745880127,
      "learning_rate": 9.611765248669795e-05,
      "loss": 0.0997,
      "step": 3210
    },
    {
      "epoch": 0.1298331092623854,
      "grad_norm": 2.765784502029419,
      "learning_rate": 9.611518824827677e-05,
      "loss": 0.0921,
      "step": 3211
    },
    {
      "epoch": 0.12987354311765242,
      "grad_norm": 8.340452194213867,
      "learning_rate": 9.611272325965111e-05,
      "loss": 0.1926,
      "step": 3212
    },
    {
      "epoch": 0.12991397697291943,
      "grad_norm": 13.78488540649414,
      "learning_rate": 9.611025752086111e-05,
      "loss": 0.2905,
      "step": 3213
    },
    {
      "epoch": 0.12995441082818643,
      "grad_norm": 4.133663654327393,
      "learning_rate": 9.610779103194685e-05,
      "loss": 0.1626,
      "step": 3214
    },
    {
      "epoch": 0.12999484468345346,
      "grad_norm": 5.196542263031006,
      "learning_rate": 9.610532379294847e-05,
      "loss": 0.1585,
      "step": 3215
    },
    {
      "epoch": 0.13003527853872046,
      "grad_norm": 6.929041862487793,
      "learning_rate": 9.610285580390612e-05,
      "loss": 0.1811,
      "step": 3216
    },
    {
      "epoch": 0.1300757123939875,
      "grad_norm": 3.4541215896606445,
      "learning_rate": 9.61003870648599e-05,
      "loss": 0.0982,
      "step": 3217
    },
    {
      "epoch": 0.1301161462492545,
      "grad_norm": 4.9665398597717285,
      "learning_rate": 9.609791757585004e-05,
      "loss": 0.1565,
      "step": 3218
    },
    {
      "epoch": 0.13015658010452152,
      "grad_norm": 4.074686527252197,
      "learning_rate": 9.609544733691666e-05,
      "loss": 0.2107,
      "step": 3219
    },
    {
      "epoch": 0.13019701395978853,
      "grad_norm": 3.0833327770233154,
      "learning_rate": 9.609297634809998e-05,
      "loss": 0.0927,
      "step": 3220
    },
    {
      "epoch": 0.13023744781505556,
      "grad_norm": 6.387527942657471,
      "learning_rate": 9.609050460944018e-05,
      "loss": 0.1051,
      "step": 3221
    },
    {
      "epoch": 0.13027788167032256,
      "grad_norm": 6.497186660766602,
      "learning_rate": 9.608803212097747e-05,
      "loss": 0.1611,
      "step": 3222
    },
    {
      "epoch": 0.1303183155255896,
      "grad_norm": 6.997556209564209,
      "learning_rate": 9.608555888275207e-05,
      "loss": 0.2122,
      "step": 3223
    },
    {
      "epoch": 0.1303587493808566,
      "grad_norm": 2.1259026527404785,
      "learning_rate": 9.608308489480425e-05,
      "loss": 0.0875,
      "step": 3224
    },
    {
      "epoch": 0.1303991832361236,
      "grad_norm": 4.309907913208008,
      "learning_rate": 9.60806101571742e-05,
      "loss": 0.1648,
      "step": 3225
    },
    {
      "epoch": 0.13043961709139062,
      "grad_norm": 8.330089569091797,
      "learning_rate": 9.607813466990221e-05,
      "loss": 0.1668,
      "step": 3226
    },
    {
      "epoch": 0.13048005094665763,
      "grad_norm": 4.964942932128906,
      "learning_rate": 9.607565843302855e-05,
      "loss": 0.1803,
      "step": 3227
    },
    {
      "epoch": 0.13052048480192466,
      "grad_norm": 4.224373817443848,
      "learning_rate": 9.607318144659351e-05,
      "loss": 0.1272,
      "step": 3228
    },
    {
      "epoch": 0.13056091865719166,
      "grad_norm": 6.426896572113037,
      "learning_rate": 9.607070371063738e-05,
      "loss": 0.1941,
      "step": 3229
    },
    {
      "epoch": 0.1306013525124587,
      "grad_norm": 4.336273670196533,
      "learning_rate": 9.606822522520044e-05,
      "loss": 0.1731,
      "step": 3230
    },
    {
      "epoch": 0.1306417863677257,
      "grad_norm": 3.774484872817993,
      "learning_rate": 9.606574599032307e-05,
      "loss": 0.1839,
      "step": 3231
    },
    {
      "epoch": 0.13068222022299272,
      "grad_norm": 6.666057109832764,
      "learning_rate": 9.606326600604556e-05,
      "loss": 0.2521,
      "step": 3232
    },
    {
      "epoch": 0.13072265407825973,
      "grad_norm": 7.5743021965026855,
      "learning_rate": 9.606078527240825e-05,
      "loss": 0.1339,
      "step": 3233
    },
    {
      "epoch": 0.13076308793352673,
      "grad_norm": 4.186329364776611,
      "learning_rate": 9.605830378945152e-05,
      "loss": 0.2303,
      "step": 3234
    },
    {
      "epoch": 0.13080352178879376,
      "grad_norm": 5.1223883628845215,
      "learning_rate": 9.605582155721573e-05,
      "loss": 0.3281,
      "step": 3235
    },
    {
      "epoch": 0.13084395564406076,
      "grad_norm": 2.8402483463287354,
      "learning_rate": 9.605333857574125e-05,
      "loss": 0.2745,
      "step": 3236
    },
    {
      "epoch": 0.1308843894993278,
      "grad_norm": 3.214343309402466,
      "learning_rate": 9.605085484506849e-05,
      "loss": 0.1087,
      "step": 3237
    },
    {
      "epoch": 0.1309248233545948,
      "grad_norm": 4.512606620788574,
      "learning_rate": 9.604837036523787e-05,
      "loss": 0.344,
      "step": 3238
    },
    {
      "epoch": 0.13096525720986182,
      "grad_norm": 16.257143020629883,
      "learning_rate": 9.604588513628975e-05,
      "loss": 0.3861,
      "step": 3239
    },
    {
      "epoch": 0.13100569106512883,
      "grad_norm": 7.765781402587891,
      "learning_rate": 9.604339915826462e-05,
      "loss": 0.2024,
      "step": 3240
    },
    {
      "epoch": 0.13104612492039586,
      "grad_norm": 5.1959547996521,
      "learning_rate": 9.604091243120288e-05,
      "loss": 0.2524,
      "step": 3241
    },
    {
      "epoch": 0.13108655877566286,
      "grad_norm": 10.115506172180176,
      "learning_rate": 9.603842495514501e-05,
      "loss": 0.2487,
      "step": 3242
    },
    {
      "epoch": 0.1311269926309299,
      "grad_norm": 3.60178542137146,
      "learning_rate": 9.603593673013148e-05,
      "loss": 0.2076,
      "step": 3243
    },
    {
      "epoch": 0.1311674264861969,
      "grad_norm": 3.5223987102508545,
      "learning_rate": 9.603344775620273e-05,
      "loss": 0.0884,
      "step": 3244
    },
    {
      "epoch": 0.1312078603414639,
      "grad_norm": 4.727479934692383,
      "learning_rate": 9.60309580333993e-05,
      "loss": 0.1834,
      "step": 3245
    },
    {
      "epoch": 0.13124829419673092,
      "grad_norm": 7.12638521194458,
      "learning_rate": 9.602846756176167e-05,
      "loss": 0.2536,
      "step": 3246
    },
    {
      "epoch": 0.13128872805199793,
      "grad_norm": 5.082055568695068,
      "learning_rate": 9.602597634133035e-05,
      "loss": 0.2168,
      "step": 3247
    },
    {
      "epoch": 0.13132916190726496,
      "grad_norm": 3.0537877082824707,
      "learning_rate": 9.602348437214588e-05,
      "loss": 0.158,
      "step": 3248
    },
    {
      "epoch": 0.13136959576253196,
      "grad_norm": 5.72343111038208,
      "learning_rate": 9.60209916542488e-05,
      "loss": 0.1344,
      "step": 3249
    },
    {
      "epoch": 0.131410029617799,
      "grad_norm": 8.558048248291016,
      "learning_rate": 9.601849818767963e-05,
      "loss": 0.1739,
      "step": 3250
    },
    {
      "epoch": 0.131450463473066,
      "grad_norm": 7.38080358505249,
      "learning_rate": 9.601600397247899e-05,
      "loss": 0.287,
      "step": 3251
    },
    {
      "epoch": 0.13149089732833302,
      "grad_norm": 4.024547576904297,
      "learning_rate": 9.60135090086874e-05,
      "loss": 0.156,
      "step": 3252
    },
    {
      "epoch": 0.13153133118360003,
      "grad_norm": 4.323020935058594,
      "learning_rate": 9.601101329634547e-05,
      "loss": 0.2007,
      "step": 3253
    },
    {
      "epoch": 0.13157176503886706,
      "grad_norm": 7.356448173522949,
      "learning_rate": 9.600851683549382e-05,
      "loss": 0.1872,
      "step": 3254
    },
    {
      "epoch": 0.13161219889413406,
      "grad_norm": 3.3663225173950195,
      "learning_rate": 9.600601962617304e-05,
      "loss": 0.0862,
      "step": 3255
    },
    {
      "epoch": 0.13165263274940106,
      "grad_norm": 3.9319443702697754,
      "learning_rate": 9.600352166842376e-05,
      "loss": 0.1828,
      "step": 3256
    },
    {
      "epoch": 0.1316930666046681,
      "grad_norm": 7.020348072052002,
      "learning_rate": 9.600102296228662e-05,
      "loss": 0.3299,
      "step": 3257
    },
    {
      "epoch": 0.1317335004599351,
      "grad_norm": 6.874411582946777,
      "learning_rate": 9.599852350780225e-05,
      "loss": 0.2006,
      "step": 3258
    },
    {
      "epoch": 0.13177393431520212,
      "grad_norm": 5.5015549659729,
      "learning_rate": 9.599602330501136e-05,
      "loss": 0.1614,
      "step": 3259
    },
    {
      "epoch": 0.13181436817046913,
      "grad_norm": 10.913169860839844,
      "learning_rate": 9.599352235395455e-05,
      "loss": 0.2445,
      "step": 3260
    },
    {
      "epoch": 0.13185480202573616,
      "grad_norm": 12.511356353759766,
      "learning_rate": 9.599102065467258e-05,
      "loss": 0.3555,
      "step": 3261
    },
    {
      "epoch": 0.13189523588100316,
      "grad_norm": 5.842842102050781,
      "learning_rate": 9.59885182072061e-05,
      "loss": 0.1681,
      "step": 3262
    },
    {
      "epoch": 0.1319356697362702,
      "grad_norm": 9.170148849487305,
      "learning_rate": 9.598601501159585e-05,
      "loss": 0.2549,
      "step": 3263
    },
    {
      "epoch": 0.1319761035915372,
      "grad_norm": 7.573867321014404,
      "learning_rate": 9.598351106788252e-05,
      "loss": 0.1446,
      "step": 3264
    },
    {
      "epoch": 0.13201653744680422,
      "grad_norm": 6.803439140319824,
      "learning_rate": 9.598100637610687e-05,
      "loss": 0.2251,
      "step": 3265
    },
    {
      "epoch": 0.13205697130207122,
      "grad_norm": 7.513017654418945,
      "learning_rate": 9.597850093630963e-05,
      "loss": 0.3054,
      "step": 3266
    },
    {
      "epoch": 0.13209740515733823,
      "grad_norm": 6.438835620880127,
      "learning_rate": 9.597599474853157e-05,
      "loss": 0.3608,
      "step": 3267
    },
    {
      "epoch": 0.13213783901260526,
      "grad_norm": 3.6551952362060547,
      "learning_rate": 9.597348781281346e-05,
      "loss": 0.1451,
      "step": 3268
    },
    {
      "epoch": 0.13217827286787226,
      "grad_norm": 2.9724977016448975,
      "learning_rate": 9.597098012919607e-05,
      "loss": 0.0954,
      "step": 3269
    },
    {
      "epoch": 0.1322187067231393,
      "grad_norm": 5.917806148529053,
      "learning_rate": 9.59684716977202e-05,
      "loss": 0.1239,
      "step": 3270
    },
    {
      "epoch": 0.1322591405784063,
      "grad_norm": 5.499895095825195,
      "learning_rate": 9.596596251842669e-05,
      "loss": 0.1266,
      "step": 3271
    },
    {
      "epoch": 0.13229957443367332,
      "grad_norm": 9.126837730407715,
      "learning_rate": 9.596345259135631e-05,
      "loss": 0.2111,
      "step": 3272
    },
    {
      "epoch": 0.13234000828894033,
      "grad_norm": 4.068196773529053,
      "learning_rate": 9.596094191654992e-05,
      "loss": 0.1439,
      "step": 3273
    },
    {
      "epoch": 0.13238044214420736,
      "grad_norm": 6.179837226867676,
      "learning_rate": 9.595843049404836e-05,
      "loss": 0.2556,
      "step": 3274
    },
    {
      "epoch": 0.13242087599947436,
      "grad_norm": 3.6682002544403076,
      "learning_rate": 9.595591832389247e-05,
      "loss": 0.1838,
      "step": 3275
    },
    {
      "epoch": 0.1324613098547414,
      "grad_norm": 3.1138885021209717,
      "learning_rate": 9.595340540612315e-05,
      "loss": 0.1396,
      "step": 3276
    },
    {
      "epoch": 0.1325017437100084,
      "grad_norm": 2.75237774848938,
      "learning_rate": 9.595089174078125e-05,
      "loss": 0.2125,
      "step": 3277
    },
    {
      "epoch": 0.1325421775652754,
      "grad_norm": 4.480936050415039,
      "learning_rate": 9.594837732790767e-05,
      "loss": 0.119,
      "step": 3278
    },
    {
      "epoch": 0.13258261142054242,
      "grad_norm": 8.116569519042969,
      "learning_rate": 9.594586216754332e-05,
      "loss": 0.2117,
      "step": 3279
    },
    {
      "epoch": 0.13262304527580943,
      "grad_norm": 3.1955950260162354,
      "learning_rate": 9.594334625972911e-05,
      "loss": 0.1611,
      "step": 3280
    },
    {
      "epoch": 0.13266347913107646,
      "grad_norm": 4.217013359069824,
      "learning_rate": 9.594082960450599e-05,
      "loss": 0.1161,
      "step": 3281
    },
    {
      "epoch": 0.13270391298634346,
      "grad_norm": 5.574992656707764,
      "learning_rate": 9.593831220191488e-05,
      "loss": 0.2795,
      "step": 3282
    },
    {
      "epoch": 0.1327443468416105,
      "grad_norm": 8.234593391418457,
      "learning_rate": 9.593579405199674e-05,
      "loss": 0.2806,
      "step": 3283
    },
    {
      "epoch": 0.1327847806968775,
      "grad_norm": 7.051079750061035,
      "learning_rate": 9.593327515479254e-05,
      "loss": 0.1671,
      "step": 3284
    },
    {
      "epoch": 0.13282521455214452,
      "grad_norm": 11.332500457763672,
      "learning_rate": 9.593075551034322e-05,
      "loss": 0.3736,
      "step": 3285
    },
    {
      "epoch": 0.13286564840741152,
      "grad_norm": 3.2538418769836426,
      "learning_rate": 9.592823511868985e-05,
      "loss": 0.1686,
      "step": 3286
    },
    {
      "epoch": 0.13290608226267855,
      "grad_norm": 2.708024024963379,
      "learning_rate": 9.592571397987335e-05,
      "loss": 0.1836,
      "step": 3287
    },
    {
      "epoch": 0.13294651611794556,
      "grad_norm": 3.5020015239715576,
      "learning_rate": 9.592319209393476e-05,
      "loss": 0.1491,
      "step": 3288
    },
    {
      "epoch": 0.13298694997321256,
      "grad_norm": 8.657442092895508,
      "learning_rate": 9.592066946091513e-05,
      "loss": 0.1895,
      "step": 3289
    },
    {
      "epoch": 0.1330273838284796,
      "grad_norm": 8.097545623779297,
      "learning_rate": 9.59181460808555e-05,
      "loss": 0.2562,
      "step": 3290
    },
    {
      "epoch": 0.1330678176837466,
      "grad_norm": 12.034658432006836,
      "learning_rate": 9.591562195379687e-05,
      "loss": 0.2701,
      "step": 3291
    },
    {
      "epoch": 0.13310825153901362,
      "grad_norm": 9.10651683807373,
      "learning_rate": 9.591309707978036e-05,
      "loss": 0.1574,
      "step": 3292
    },
    {
      "epoch": 0.13314868539428062,
      "grad_norm": 7.29874324798584,
      "learning_rate": 9.5910571458847e-05,
      "loss": 0.2007,
      "step": 3293
    },
    {
      "epoch": 0.13318911924954766,
      "grad_norm": 3.3138577938079834,
      "learning_rate": 9.59080450910379e-05,
      "loss": 0.2089,
      "step": 3294
    },
    {
      "epoch": 0.13322955310481466,
      "grad_norm": 6.86495304107666,
      "learning_rate": 9.590551797639416e-05,
      "loss": 0.161,
      "step": 3295
    },
    {
      "epoch": 0.1332699869600817,
      "grad_norm": 3.062520980834961,
      "learning_rate": 9.590299011495689e-05,
      "loss": 0.1336,
      "step": 3296
    },
    {
      "epoch": 0.1333104208153487,
      "grad_norm": 3.2224082946777344,
      "learning_rate": 9.590046150676721e-05,
      "loss": 0.1528,
      "step": 3297
    },
    {
      "epoch": 0.1333508546706157,
      "grad_norm": 6.471376895904541,
      "learning_rate": 9.589793215186624e-05,
      "loss": 0.1433,
      "step": 3298
    },
    {
      "epoch": 0.13339128852588272,
      "grad_norm": 5.662395000457764,
      "learning_rate": 9.589540205029517e-05,
      "loss": 0.1582,
      "step": 3299
    },
    {
      "epoch": 0.13343172238114973,
      "grad_norm": 3.11987042427063,
      "learning_rate": 9.589287120209511e-05,
      "loss": 0.1551,
      "step": 3300
    },
    {
      "epoch": 0.13347215623641676,
      "grad_norm": 6.835002899169922,
      "learning_rate": 9.589033960730726e-05,
      "loss": 0.2923,
      "step": 3301
    },
    {
      "epoch": 0.13351259009168376,
      "grad_norm": 3.9990367889404297,
      "learning_rate": 9.588780726597282e-05,
      "loss": 0.343,
      "step": 3302
    },
    {
      "epoch": 0.1335530239469508,
      "grad_norm": 4.245683193206787,
      "learning_rate": 9.588527417813294e-05,
      "loss": 0.204,
      "step": 3303
    },
    {
      "epoch": 0.1335934578022178,
      "grad_norm": 4.173803329467773,
      "learning_rate": 9.588274034382887e-05,
      "loss": 0.1327,
      "step": 3304
    },
    {
      "epoch": 0.13363389165748482,
      "grad_norm": 3.67171049118042,
      "learning_rate": 9.58802057631018e-05,
      "loss": 0.171,
      "step": 3305
    },
    {
      "epoch": 0.13367432551275182,
      "grad_norm": 4.190315246582031,
      "learning_rate": 9.587767043599298e-05,
      "loss": 0.1746,
      "step": 3306
    },
    {
      "epoch": 0.13371475936801885,
      "grad_norm": 4.629176139831543,
      "learning_rate": 9.587513436254367e-05,
      "loss": 0.2861,
      "step": 3307
    },
    {
      "epoch": 0.13375519322328586,
      "grad_norm": 2.6783041954040527,
      "learning_rate": 9.587259754279509e-05,
      "loss": 0.1339,
      "step": 3308
    },
    {
      "epoch": 0.13379562707855286,
      "grad_norm": 4.862961292266846,
      "learning_rate": 9.587005997678854e-05,
      "loss": 0.1301,
      "step": 3309
    },
    {
      "epoch": 0.1338360609338199,
      "grad_norm": 3.5956549644470215,
      "learning_rate": 9.586752166456529e-05,
      "loss": 0.1218,
      "step": 3310
    },
    {
      "epoch": 0.1338764947890869,
      "grad_norm": 8.196605682373047,
      "learning_rate": 9.586498260616663e-05,
      "loss": 0.3057,
      "step": 3311
    },
    {
      "epoch": 0.13391692864435392,
      "grad_norm": 3.6698269844055176,
      "learning_rate": 9.586244280163386e-05,
      "loss": 0.1537,
      "step": 3312
    },
    {
      "epoch": 0.13395736249962092,
      "grad_norm": 4.987952709197998,
      "learning_rate": 9.585990225100831e-05,
      "loss": 0.1341,
      "step": 3313
    },
    {
      "epoch": 0.13399779635488795,
      "grad_norm": 4.503644943237305,
      "learning_rate": 9.585736095433131e-05,
      "loss": 0.2348,
      "step": 3314
    },
    {
      "epoch": 0.13403823021015496,
      "grad_norm": 2.487854242324829,
      "learning_rate": 9.58548189116442e-05,
      "loss": 0.128,
      "step": 3315
    },
    {
      "epoch": 0.134078664065422,
      "grad_norm": 9.025087356567383,
      "learning_rate": 9.585227612298833e-05,
      "loss": 0.2414,
      "step": 3316
    },
    {
      "epoch": 0.134119097920689,
      "grad_norm": 6.2614521980285645,
      "learning_rate": 9.584973258840507e-05,
      "loss": 0.1984,
      "step": 3317
    },
    {
      "epoch": 0.13415953177595602,
      "grad_norm": 3.3130719661712646,
      "learning_rate": 9.584718830793579e-05,
      "loss": 0.2194,
      "step": 3318
    },
    {
      "epoch": 0.13419996563122302,
      "grad_norm": 5.85471248626709,
      "learning_rate": 9.58446432816219e-05,
      "loss": 0.1502,
      "step": 3319
    },
    {
      "epoch": 0.13424039948649003,
      "grad_norm": 7.2113237380981445,
      "learning_rate": 9.584209750950478e-05,
      "loss": 0.1533,
      "step": 3320
    },
    {
      "epoch": 0.13428083334175706,
      "grad_norm": 2.2621777057647705,
      "learning_rate": 9.583955099162586e-05,
      "loss": 0.2094,
      "step": 3321
    },
    {
      "epoch": 0.13432126719702406,
      "grad_norm": 5.248476505279541,
      "learning_rate": 9.583700372802655e-05,
      "loss": 0.1319,
      "step": 3322
    },
    {
      "epoch": 0.1343617010522911,
      "grad_norm": 10.026060104370117,
      "learning_rate": 9.583445571874832e-05,
      "loss": 0.2884,
      "step": 3323
    },
    {
      "epoch": 0.1344021349075581,
      "grad_norm": 3.7140965461730957,
      "learning_rate": 9.583190696383258e-05,
      "loss": 0.1094,
      "step": 3324
    },
    {
      "epoch": 0.13444256876282512,
      "grad_norm": 5.366124629974365,
      "learning_rate": 9.582935746332083e-05,
      "loss": 0.1405,
      "step": 3325
    },
    {
      "epoch": 0.13448300261809212,
      "grad_norm": 5.522871017456055,
      "learning_rate": 9.582680721725452e-05,
      "loss": 0.0834,
      "step": 3326
    },
    {
      "epoch": 0.13452343647335915,
      "grad_norm": 5.586090087890625,
      "learning_rate": 9.582425622567515e-05,
      "loss": 0.3464,
      "step": 3327
    },
    {
      "epoch": 0.13456387032862616,
      "grad_norm": 6.8120317459106445,
      "learning_rate": 9.582170448862423e-05,
      "loss": 0.2206,
      "step": 3328
    },
    {
      "epoch": 0.1346043041838932,
      "grad_norm": 8.491228103637695,
      "learning_rate": 9.581915200614325e-05,
      "loss": 0.3243,
      "step": 3329
    },
    {
      "epoch": 0.1346447380391602,
      "grad_norm": 6.080544471740723,
      "learning_rate": 9.581659877827374e-05,
      "loss": 0.1831,
      "step": 3330
    },
    {
      "epoch": 0.1346851718944272,
      "grad_norm": 13.643579483032227,
      "learning_rate": 9.581404480505725e-05,
      "loss": 0.2953,
      "step": 3331
    },
    {
      "epoch": 0.13472560574969422,
      "grad_norm": 6.166047096252441,
      "learning_rate": 9.581149008653531e-05,
      "loss": 0.1967,
      "step": 3332
    },
    {
      "epoch": 0.13476603960496122,
      "grad_norm": 10.076539993286133,
      "learning_rate": 9.580893462274949e-05,
      "loss": 0.2694,
      "step": 3333
    },
    {
      "epoch": 0.13480647346022825,
      "grad_norm": 5.800323009490967,
      "learning_rate": 9.580637841374137e-05,
      "loss": 0.1792,
      "step": 3334
    },
    {
      "epoch": 0.13484690731549526,
      "grad_norm": 4.676064491271973,
      "learning_rate": 9.580382145955251e-05,
      "loss": 0.1664,
      "step": 3335
    },
    {
      "epoch": 0.1348873411707623,
      "grad_norm": 5.5211029052734375,
      "learning_rate": 9.580126376022453e-05,
      "loss": 0.1253,
      "step": 3336
    },
    {
      "epoch": 0.1349277750260293,
      "grad_norm": 5.6168718338012695,
      "learning_rate": 9.579870531579902e-05,
      "loss": 0.1879,
      "step": 3337
    },
    {
      "epoch": 0.13496820888129632,
      "grad_norm": 2.7584898471832275,
      "learning_rate": 9.579614612631762e-05,
      "loss": 0.204,
      "step": 3338
    },
    {
      "epoch": 0.13500864273656332,
      "grad_norm": 6.506252765655518,
      "learning_rate": 9.579358619182195e-05,
      "loss": 0.2569,
      "step": 3339
    },
    {
      "epoch": 0.13504907659183035,
      "grad_norm": 6.172201633453369,
      "learning_rate": 9.579102551235365e-05,
      "loss": 0.2006,
      "step": 3340
    },
    {
      "epoch": 0.13508951044709736,
      "grad_norm": 8.017014503479004,
      "learning_rate": 9.578846408795441e-05,
      "loss": 0.2647,
      "step": 3341
    },
    {
      "epoch": 0.13512994430236436,
      "grad_norm": 5.504523277282715,
      "learning_rate": 9.578590191866586e-05,
      "loss": 0.1689,
      "step": 3342
    },
    {
      "epoch": 0.1351703781576314,
      "grad_norm": 7.352840900421143,
      "learning_rate": 9.57833390045297e-05,
      "loss": 0.196,
      "step": 3343
    },
    {
      "epoch": 0.1352108120128984,
      "grad_norm": 6.665403366088867,
      "learning_rate": 9.578077534558763e-05,
      "loss": 0.1448,
      "step": 3344
    },
    {
      "epoch": 0.13525124586816542,
      "grad_norm": 2.412034273147583,
      "learning_rate": 9.577821094188133e-05,
      "loss": 0.1288,
      "step": 3345
    },
    {
      "epoch": 0.13529167972343242,
      "grad_norm": 4.581907272338867,
      "learning_rate": 9.577564579345255e-05,
      "loss": 0.118,
      "step": 3346
    },
    {
      "epoch": 0.13533211357869945,
      "grad_norm": 3.5014419555664062,
      "learning_rate": 9.577307990034301e-05,
      "loss": 0.1088,
      "step": 3347
    },
    {
      "epoch": 0.13537254743396646,
      "grad_norm": 10.801563262939453,
      "learning_rate": 9.577051326259443e-05,
      "loss": 0.2627,
      "step": 3348
    },
    {
      "epoch": 0.13541298128923349,
      "grad_norm": 3.8466949462890625,
      "learning_rate": 9.576794588024859e-05,
      "loss": 0.0782,
      "step": 3349
    },
    {
      "epoch": 0.1354534151445005,
      "grad_norm": 7.30478572845459,
      "learning_rate": 9.576537775334724e-05,
      "loss": 0.2381,
      "step": 3350
    },
    {
      "epoch": 0.13549384899976752,
      "grad_norm": 10.397120475769043,
      "learning_rate": 9.576280888193217e-05,
      "loss": 0.3413,
      "step": 3351
    },
    {
      "epoch": 0.13553428285503452,
      "grad_norm": 3.953392744064331,
      "learning_rate": 9.576023926604517e-05,
      "loss": 0.2361,
      "step": 3352
    },
    {
      "epoch": 0.13557471671030152,
      "grad_norm": 3.1065471172332764,
      "learning_rate": 9.575766890572803e-05,
      "loss": 0.2197,
      "step": 3353
    },
    {
      "epoch": 0.13561515056556855,
      "grad_norm": 5.370883464813232,
      "learning_rate": 9.575509780102259e-05,
      "loss": 0.1827,
      "step": 3354
    },
    {
      "epoch": 0.13565558442083556,
      "grad_norm": 3.2279443740844727,
      "learning_rate": 9.575252595197066e-05,
      "loss": 0.1636,
      "step": 3355
    },
    {
      "epoch": 0.1356960182761026,
      "grad_norm": 2.7974250316619873,
      "learning_rate": 9.574995335861407e-05,
      "loss": 0.1853,
      "step": 3356
    },
    {
      "epoch": 0.1357364521313696,
      "grad_norm": 2.0515267848968506,
      "learning_rate": 9.574738002099468e-05,
      "loss": 0.0735,
      "step": 3357
    },
    {
      "epoch": 0.13577688598663662,
      "grad_norm": 3.0899031162261963,
      "learning_rate": 9.574480593915436e-05,
      "loss": 0.1757,
      "step": 3358
    },
    {
      "epoch": 0.13581731984190362,
      "grad_norm": 2.3329625129699707,
      "learning_rate": 9.574223111313496e-05,
      "loss": 0.1308,
      "step": 3359
    },
    {
      "epoch": 0.13585775369717065,
      "grad_norm": 2.758235454559326,
      "learning_rate": 9.573965554297841e-05,
      "loss": 0.0922,
      "step": 3360
    },
    {
      "epoch": 0.13589818755243765,
      "grad_norm": 6.687131404876709,
      "learning_rate": 9.573707922872657e-05,
      "loss": 0.3549,
      "step": 3361
    },
    {
      "epoch": 0.13593862140770466,
      "grad_norm": 7.911161422729492,
      "learning_rate": 9.573450217042138e-05,
      "loss": 0.2898,
      "step": 3362
    },
    {
      "epoch": 0.1359790552629717,
      "grad_norm": 6.317928791046143,
      "learning_rate": 9.573192436810475e-05,
      "loss": 0.2265,
      "step": 3363
    },
    {
      "epoch": 0.1360194891182387,
      "grad_norm": 9.419342041015625,
      "learning_rate": 9.572934582181861e-05,
      "loss": 0.1968,
      "step": 3364
    },
    {
      "epoch": 0.13605992297350572,
      "grad_norm": 10.527022361755371,
      "learning_rate": 9.572676653160493e-05,
      "loss": 0.3463,
      "step": 3365
    },
    {
      "epoch": 0.13610035682877272,
      "grad_norm": 2.5564329624176025,
      "learning_rate": 9.572418649750564e-05,
      "loss": 0.1612,
      "step": 3366
    },
    {
      "epoch": 0.13614079068403975,
      "grad_norm": 6.918292999267578,
      "learning_rate": 9.572160571956274e-05,
      "loss": 0.2445,
      "step": 3367
    },
    {
      "epoch": 0.13618122453930676,
      "grad_norm": 9.50683307647705,
      "learning_rate": 9.57190241978182e-05,
      "loss": 0.1904,
      "step": 3368
    },
    {
      "epoch": 0.13622165839457379,
      "grad_norm": 5.411144256591797,
      "learning_rate": 9.5716441932314e-05,
      "loss": 0.2151,
      "step": 3369
    },
    {
      "epoch": 0.1362620922498408,
      "grad_norm": 4.506725311279297,
      "learning_rate": 9.571385892309219e-05,
      "loss": 0.1949,
      "step": 3370
    },
    {
      "epoch": 0.13630252610510782,
      "grad_norm": 3.301774263381958,
      "learning_rate": 9.571127517019475e-05,
      "loss": 0.1682,
      "step": 3371
    },
    {
      "epoch": 0.13634295996037482,
      "grad_norm": 8.42991828918457,
      "learning_rate": 9.570869067366374e-05,
      "loss": 0.259,
      "step": 3372
    },
    {
      "epoch": 0.13638339381564182,
      "grad_norm": 5.217926979064941,
      "learning_rate": 9.570610543354119e-05,
      "loss": 0.1921,
      "step": 3373
    },
    {
      "epoch": 0.13642382767090885,
      "grad_norm": 5.017066478729248,
      "learning_rate": 9.570351944986917e-05,
      "loss": 0.1607,
      "step": 3374
    },
    {
      "epoch": 0.13646426152617586,
      "grad_norm": 4.415814399719238,
      "learning_rate": 9.570093272268973e-05,
      "loss": 0.1979,
      "step": 3375
    },
    {
      "epoch": 0.1365046953814429,
      "grad_norm": 2.5535149574279785,
      "learning_rate": 9.569834525204496e-05,
      "loss": 0.151,
      "step": 3376
    },
    {
      "epoch": 0.1365451292367099,
      "grad_norm": 3.367611885070801,
      "learning_rate": 9.569575703797697e-05,
      "loss": 0.2141,
      "step": 3377
    },
    {
      "epoch": 0.13658556309197692,
      "grad_norm": 1.965120792388916,
      "learning_rate": 9.569316808052783e-05,
      "loss": 0.1451,
      "step": 3378
    },
    {
      "epoch": 0.13662599694724392,
      "grad_norm": 2.9477896690368652,
      "learning_rate": 9.569057837973969e-05,
      "loss": 0.2214,
      "step": 3379
    },
    {
      "epoch": 0.13666643080251095,
      "grad_norm": 3.523401975631714,
      "learning_rate": 9.568798793565465e-05,
      "loss": 0.1932,
      "step": 3380
    },
    {
      "epoch": 0.13670686465777795,
      "grad_norm": 5.432856559753418,
      "learning_rate": 9.56853967483149e-05,
      "loss": 0.2909,
      "step": 3381
    },
    {
      "epoch": 0.13674729851304498,
      "grad_norm": 2.565180540084839,
      "learning_rate": 9.568280481776253e-05,
      "loss": 0.174,
      "step": 3382
    },
    {
      "epoch": 0.136787732368312,
      "grad_norm": 1.3865786790847778,
      "learning_rate": 9.568021214403975e-05,
      "loss": 0.0618,
      "step": 3383
    },
    {
      "epoch": 0.136828166223579,
      "grad_norm": 5.439849853515625,
      "learning_rate": 9.567761872718872e-05,
      "loss": 0.1659,
      "step": 3384
    },
    {
      "epoch": 0.13686860007884602,
      "grad_norm": 7.5103759765625,
      "learning_rate": 9.567502456725164e-05,
      "loss": 0.2557,
      "step": 3385
    },
    {
      "epoch": 0.13690903393411302,
      "grad_norm": 4.842629432678223,
      "learning_rate": 9.56724296642707e-05,
      "loss": 0.1304,
      "step": 3386
    },
    {
      "epoch": 0.13694946778938005,
      "grad_norm": 2.548671007156372,
      "learning_rate": 9.566983401828813e-05,
      "loss": 0.2184,
      "step": 3387
    },
    {
      "epoch": 0.13698990164464706,
      "grad_norm": 5.978455543518066,
      "learning_rate": 9.566723762934614e-05,
      "loss": 0.361,
      "step": 3388
    },
    {
      "epoch": 0.13703033549991409,
      "grad_norm": 6.170347690582275,
      "learning_rate": 9.566464049748698e-05,
      "loss": 0.1604,
      "step": 3389
    },
    {
      "epoch": 0.1370707693551811,
      "grad_norm": 6.19504451751709,
      "learning_rate": 9.56620426227529e-05,
      "loss": 0.3457,
      "step": 3390
    },
    {
      "epoch": 0.13711120321044812,
      "grad_norm": 3.670827627182007,
      "learning_rate": 9.565944400518613e-05,
      "loss": 0.1018,
      "step": 3391
    },
    {
      "epoch": 0.13715163706571512,
      "grad_norm": 7.104718208312988,
      "learning_rate": 9.565684464482901e-05,
      "loss": 0.4441,
      "step": 3392
    },
    {
      "epoch": 0.13719207092098215,
      "grad_norm": 2.193960428237915,
      "learning_rate": 9.565424454172376e-05,
      "loss": 0.0845,
      "step": 3393
    },
    {
      "epoch": 0.13723250477624915,
      "grad_norm": 5.507030010223389,
      "learning_rate": 9.56516436959127e-05,
      "loss": 0.3164,
      "step": 3394
    },
    {
      "epoch": 0.13727293863151616,
      "grad_norm": 5.012667179107666,
      "learning_rate": 9.564904210743817e-05,
      "loss": 0.1414,
      "step": 3395
    },
    {
      "epoch": 0.1373133724867832,
      "grad_norm": 4.671367645263672,
      "learning_rate": 9.564643977634247e-05,
      "loss": 0.1113,
      "step": 3396
    },
    {
      "epoch": 0.1373538063420502,
      "grad_norm": 3.1328396797180176,
      "learning_rate": 9.564383670266793e-05,
      "loss": 0.1376,
      "step": 3397
    },
    {
      "epoch": 0.13739424019731722,
      "grad_norm": 1.6062774658203125,
      "learning_rate": 9.56412328864569e-05,
      "loss": 0.0475,
      "step": 3398
    },
    {
      "epoch": 0.13743467405258422,
      "grad_norm": 4.274123668670654,
      "learning_rate": 9.563862832775176e-05,
      "loss": 0.1324,
      "step": 3399
    },
    {
      "epoch": 0.13747510790785125,
      "grad_norm": 4.227860450744629,
      "learning_rate": 9.563602302659485e-05,
      "loss": 0.1312,
      "step": 3400
    },
    {
      "epoch": 0.13751554176311825,
      "grad_norm": 6.541635036468506,
      "learning_rate": 9.563341698302856e-05,
      "loss": 0.2706,
      "step": 3401
    },
    {
      "epoch": 0.13755597561838528,
      "grad_norm": 4.44124174118042,
      "learning_rate": 9.563081019709529e-05,
      "loss": 0.1325,
      "step": 3402
    },
    {
      "epoch": 0.1375964094736523,
      "grad_norm": 8.909440040588379,
      "learning_rate": 9.562820266883745e-05,
      "loss": 0.2591,
      "step": 3403
    },
    {
      "epoch": 0.13763684332891932,
      "grad_norm": 3.781332492828369,
      "learning_rate": 9.562559439829746e-05,
      "loss": 0.1272,
      "step": 3404
    },
    {
      "epoch": 0.13767727718418632,
      "grad_norm": 8.2918701171875,
      "learning_rate": 9.562298538551774e-05,
      "loss": 0.2171,
      "step": 3405
    },
    {
      "epoch": 0.13771771103945332,
      "grad_norm": 6.390937805175781,
      "learning_rate": 9.562037563054077e-05,
      "loss": 0.2105,
      "step": 3406
    },
    {
      "epoch": 0.13775814489472035,
      "grad_norm": 2.494025468826294,
      "learning_rate": 9.561776513340896e-05,
      "loss": 0.0644,
      "step": 3407
    },
    {
      "epoch": 0.13779857874998735,
      "grad_norm": 6.266638278961182,
      "learning_rate": 9.56151538941648e-05,
      "loss": 0.2079,
      "step": 3408
    },
    {
      "epoch": 0.13783901260525439,
      "grad_norm": 5.68182373046875,
      "learning_rate": 9.561254191285077e-05,
      "loss": 0.1493,
      "step": 3409
    },
    {
      "epoch": 0.1378794464605214,
      "grad_norm": 3.4947195053100586,
      "learning_rate": 9.560992918950935e-05,
      "loss": 0.0957,
      "step": 3410
    },
    {
      "epoch": 0.13791988031578842,
      "grad_norm": 3.1119680404663086,
      "learning_rate": 9.560731572418304e-05,
      "loss": 0.0814,
      "step": 3411
    },
    {
      "epoch": 0.13796031417105542,
      "grad_norm": 3.2736432552337646,
      "learning_rate": 9.560470151691438e-05,
      "loss": 0.1445,
      "step": 3412
    },
    {
      "epoch": 0.13800074802632245,
      "grad_norm": 4.617390155792236,
      "learning_rate": 9.56020865677459e-05,
      "loss": 0.2313,
      "step": 3413
    },
    {
      "epoch": 0.13804118188158945,
      "grad_norm": 10.78541088104248,
      "learning_rate": 9.559947087672012e-05,
      "loss": 0.2714,
      "step": 3414
    },
    {
      "epoch": 0.13808161573685648,
      "grad_norm": 4.890484809875488,
      "learning_rate": 9.559685444387958e-05,
      "loss": 0.1479,
      "step": 3415
    },
    {
      "epoch": 0.13812204959212349,
      "grad_norm": 3.7044496536254883,
      "learning_rate": 9.559423726926689e-05,
      "loss": 0.0784,
      "step": 3416
    },
    {
      "epoch": 0.1381624834473905,
      "grad_norm": 3.7583093643188477,
      "learning_rate": 9.559161935292458e-05,
      "loss": 0.114,
      "step": 3417
    },
    {
      "epoch": 0.13820291730265752,
      "grad_norm": 6.684147834777832,
      "learning_rate": 9.558900069489524e-05,
      "loss": 0.2234,
      "step": 3418
    },
    {
      "epoch": 0.13824335115792452,
      "grad_norm": 6.423051834106445,
      "learning_rate": 9.558638129522153e-05,
      "loss": 0.258,
      "step": 3419
    },
    {
      "epoch": 0.13828378501319155,
      "grad_norm": 6.327081203460693,
      "learning_rate": 9.558376115394598e-05,
      "loss": 0.2647,
      "step": 3420
    },
    {
      "epoch": 0.13832421886845855,
      "grad_norm": 7.196712017059326,
      "learning_rate": 9.558114027111129e-05,
      "loss": 0.2364,
      "step": 3421
    },
    {
      "epoch": 0.13836465272372558,
      "grad_norm": 6.086154460906982,
      "learning_rate": 9.557851864676005e-05,
      "loss": 0.1821,
      "step": 3422
    },
    {
      "epoch": 0.1384050865789926,
      "grad_norm": 7.127618312835693,
      "learning_rate": 9.55758962809349e-05,
      "loss": 0.2735,
      "step": 3423
    },
    {
      "epoch": 0.13844552043425962,
      "grad_norm": 2.0601468086242676,
      "learning_rate": 9.557327317367854e-05,
      "loss": 0.0842,
      "step": 3424
    },
    {
      "epoch": 0.13848595428952662,
      "grad_norm": 2.8338241577148438,
      "learning_rate": 9.557064932503362e-05,
      "loss": 0.1519,
      "step": 3425
    },
    {
      "epoch": 0.13852638814479362,
      "grad_norm": 1.5631422996520996,
      "learning_rate": 9.556802473504282e-05,
      "loss": 0.0618,
      "step": 3426
    },
    {
      "epoch": 0.13856682200006065,
      "grad_norm": 5.567506790161133,
      "learning_rate": 9.556539940374887e-05,
      "loss": 0.2272,
      "step": 3427
    },
    {
      "epoch": 0.13860725585532765,
      "grad_norm": 6.285430431365967,
      "learning_rate": 9.556277333119443e-05,
      "loss": 0.1963,
      "step": 3428
    },
    {
      "epoch": 0.13864768971059468,
      "grad_norm": 4.521222114562988,
      "learning_rate": 9.556014651742227e-05,
      "loss": 0.1793,
      "step": 3429
    },
    {
      "epoch": 0.1386881235658617,
      "grad_norm": 3.21838116645813,
      "learning_rate": 9.555751896247508e-05,
      "loss": 0.0971,
      "step": 3430
    },
    {
      "epoch": 0.13872855742112872,
      "grad_norm": 6.9481072425842285,
      "learning_rate": 9.555489066639563e-05,
      "loss": 0.176,
      "step": 3431
    },
    {
      "epoch": 0.13876899127639572,
      "grad_norm": 4.603512287139893,
      "learning_rate": 9.555226162922667e-05,
      "loss": 0.1469,
      "step": 3432
    },
    {
      "epoch": 0.13880942513166275,
      "grad_norm": 4.470980644226074,
      "learning_rate": 9.554963185101098e-05,
      "loss": 0.2193,
      "step": 3433
    },
    {
      "epoch": 0.13884985898692975,
      "grad_norm": 7.165218353271484,
      "learning_rate": 9.554700133179132e-05,
      "loss": 0.2188,
      "step": 3434
    },
    {
      "epoch": 0.13889029284219678,
      "grad_norm": 3.1681671142578125,
      "learning_rate": 9.55443700716105e-05,
      "loss": 0.1615,
      "step": 3435
    },
    {
      "epoch": 0.13893072669746379,
      "grad_norm": 7.174466133117676,
      "learning_rate": 9.554173807051132e-05,
      "loss": 0.2446,
      "step": 3436
    },
    {
      "epoch": 0.1389711605527308,
      "grad_norm": 3.594766139984131,
      "learning_rate": 9.55391053285366e-05,
      "loss": 0.1922,
      "step": 3437
    },
    {
      "epoch": 0.13901159440799782,
      "grad_norm": 5.171209335327148,
      "learning_rate": 9.553647184572917e-05,
      "loss": 0.1506,
      "step": 3438
    },
    {
      "epoch": 0.13905202826326482,
      "grad_norm": 5.774380207061768,
      "learning_rate": 9.553383762213186e-05,
      "loss": 0.1538,
      "step": 3439
    },
    {
      "epoch": 0.13909246211853185,
      "grad_norm": 6.404676914215088,
      "learning_rate": 9.553120265778754e-05,
      "loss": 0.2164,
      "step": 3440
    },
    {
      "epoch": 0.13913289597379885,
      "grad_norm": 2.1848831176757812,
      "learning_rate": 9.552856695273908e-05,
      "loss": 0.0989,
      "step": 3441
    },
    {
      "epoch": 0.13917332982906588,
      "grad_norm": 6.371652126312256,
      "learning_rate": 9.552593050702934e-05,
      "loss": 0.1987,
      "step": 3442
    },
    {
      "epoch": 0.1392137636843329,
      "grad_norm": 2.929884672164917,
      "learning_rate": 9.552329332070122e-05,
      "loss": 0.1172,
      "step": 3443
    },
    {
      "epoch": 0.13925419753959992,
      "grad_norm": 6.50648832321167,
      "learning_rate": 9.552065539379762e-05,
      "loss": 0.1683,
      "step": 3444
    },
    {
      "epoch": 0.13929463139486692,
      "grad_norm": 3.370377779006958,
      "learning_rate": 9.551801672636144e-05,
      "loss": 0.0697,
      "step": 3445
    },
    {
      "epoch": 0.13933506525013395,
      "grad_norm": 1.826620101928711,
      "learning_rate": 9.551537731843563e-05,
      "loss": 0.1469,
      "step": 3446
    },
    {
      "epoch": 0.13937549910540095,
      "grad_norm": 4.42665433883667,
      "learning_rate": 9.55127371700631e-05,
      "loss": 0.1301,
      "step": 3447
    },
    {
      "epoch": 0.13941593296066795,
      "grad_norm": 4.635818004608154,
      "learning_rate": 9.551009628128684e-05,
      "loss": 0.2375,
      "step": 3448
    },
    {
      "epoch": 0.13945636681593498,
      "grad_norm": 4.2088847160339355,
      "learning_rate": 9.550745465214978e-05,
      "loss": 0.0878,
      "step": 3449
    },
    {
      "epoch": 0.139496800671202,
      "grad_norm": 3.120185375213623,
      "learning_rate": 9.55048122826949e-05,
      "loss": 0.2424,
      "step": 3450
    },
    {
      "epoch": 0.13953723452646902,
      "grad_norm": 10.87005615234375,
      "learning_rate": 9.55021691729652e-05,
      "loss": 0.2601,
      "step": 3451
    },
    {
      "epoch": 0.13957766838173602,
      "grad_norm": 6.9819817543029785,
      "learning_rate": 9.549952532300366e-05,
      "loss": 0.1622,
      "step": 3452
    },
    {
      "epoch": 0.13961810223700305,
      "grad_norm": 8.456396102905273,
      "learning_rate": 9.54968807328533e-05,
      "loss": 0.1758,
      "step": 3453
    },
    {
      "epoch": 0.13965853609227005,
      "grad_norm": 8.399441719055176,
      "learning_rate": 9.549423540255712e-05,
      "loss": 0.2194,
      "step": 3454
    },
    {
      "epoch": 0.13969896994753708,
      "grad_norm": 6.708520889282227,
      "learning_rate": 9.549158933215821e-05,
      "loss": 0.1863,
      "step": 3455
    },
    {
      "epoch": 0.13973940380280409,
      "grad_norm": 4.6737284660339355,
      "learning_rate": 9.548894252169958e-05,
      "loss": 0.2034,
      "step": 3456
    },
    {
      "epoch": 0.13977983765807112,
      "grad_norm": 4.977152347564697,
      "learning_rate": 9.548629497122427e-05,
      "loss": 0.1423,
      "step": 3457
    },
    {
      "epoch": 0.13982027151333812,
      "grad_norm": 1.0458273887634277,
      "learning_rate": 9.548364668077536e-05,
      "loss": 0.0881,
      "step": 3458
    },
    {
      "epoch": 0.13986070536860512,
      "grad_norm": 3.8723297119140625,
      "learning_rate": 9.548099765039597e-05,
      "loss": 0.0743,
      "step": 3459
    },
    {
      "epoch": 0.13990113922387215,
      "grad_norm": 4.772374629974365,
      "learning_rate": 9.547834788012915e-05,
      "loss": 0.3095,
      "step": 3460
    },
    {
      "epoch": 0.13994157307913915,
      "grad_norm": 3.9878618717193604,
      "learning_rate": 9.547569737001802e-05,
      "loss": 0.2374,
      "step": 3461
    },
    {
      "epoch": 0.13998200693440618,
      "grad_norm": 2.8656413555145264,
      "learning_rate": 9.547304612010573e-05,
      "loss": 0.2077,
      "step": 3462
    },
    {
      "epoch": 0.14002244078967319,
      "grad_norm": 5.248875617980957,
      "learning_rate": 9.547039413043535e-05,
      "loss": 0.2008,
      "step": 3463
    },
    {
      "epoch": 0.14006287464494022,
      "grad_norm": 4.484411716461182,
      "learning_rate": 9.546774140105008e-05,
      "loss": 0.2415,
      "step": 3464
    },
    {
      "epoch": 0.14010330850020722,
      "grad_norm": 2.5546905994415283,
      "learning_rate": 9.546508793199305e-05,
      "loss": 0.1531,
      "step": 3465
    },
    {
      "epoch": 0.14014374235547425,
      "grad_norm": 8.2513427734375,
      "learning_rate": 9.54624337233074e-05,
      "loss": 0.2009,
      "step": 3466
    },
    {
      "epoch": 0.14018417621074125,
      "grad_norm": 10.440117835998535,
      "learning_rate": 9.545977877503637e-05,
      "loss": 0.2599,
      "step": 3467
    },
    {
      "epoch": 0.14022461006600828,
      "grad_norm": 9.524097442626953,
      "learning_rate": 9.54571230872231e-05,
      "loss": 0.2741,
      "step": 3468
    },
    {
      "epoch": 0.14026504392127528,
      "grad_norm": 5.911411285400391,
      "learning_rate": 9.545446665991083e-05,
      "loss": 0.0908,
      "step": 3469
    },
    {
      "epoch": 0.1403054777765423,
      "grad_norm": 4.05776309967041,
      "learning_rate": 9.545180949314272e-05,
      "loss": 0.1918,
      "step": 3470
    },
    {
      "epoch": 0.14034591163180932,
      "grad_norm": 4.600869178771973,
      "learning_rate": 9.544915158696206e-05,
      "loss": 0.2609,
      "step": 3471
    },
    {
      "epoch": 0.14038634548707632,
      "grad_norm": 2.694551467895508,
      "learning_rate": 9.544649294141205e-05,
      "loss": 0.1309,
      "step": 3472
    },
    {
      "epoch": 0.14042677934234335,
      "grad_norm": 9.02026081085205,
      "learning_rate": 9.544383355653594e-05,
      "loss": 0.2373,
      "step": 3473
    },
    {
      "epoch": 0.14046721319761035,
      "grad_norm": 8.126273155212402,
      "learning_rate": 9.544117343237702e-05,
      "loss": 0.2214,
      "step": 3474
    },
    {
      "epoch": 0.14050764705287738,
      "grad_norm": 2.1096255779266357,
      "learning_rate": 9.543851256897856e-05,
      "loss": 0.0959,
      "step": 3475
    },
    {
      "epoch": 0.14054808090814438,
      "grad_norm": 8.093713760375977,
      "learning_rate": 9.543585096638381e-05,
      "loss": 0.2737,
      "step": 3476
    },
    {
      "epoch": 0.14058851476341142,
      "grad_norm": 8.19959831237793,
      "learning_rate": 9.543318862463611e-05,
      "loss": 0.1954,
      "step": 3477
    },
    {
      "epoch": 0.14062894861867842,
      "grad_norm": 3.536663770675659,
      "learning_rate": 9.543052554377876e-05,
      "loss": 0.1532,
      "step": 3478
    },
    {
      "epoch": 0.14066938247394545,
      "grad_norm": 2.866973400115967,
      "learning_rate": 9.542786172385508e-05,
      "loss": 0.2553,
      "step": 3479
    },
    {
      "epoch": 0.14070981632921245,
      "grad_norm": 8.79090690612793,
      "learning_rate": 9.542519716490841e-05,
      "loss": 0.2558,
      "step": 3480
    },
    {
      "epoch": 0.14075025018447945,
      "grad_norm": 5.096632957458496,
      "learning_rate": 9.542253186698209e-05,
      "loss": 0.1652,
      "step": 3481
    },
    {
      "epoch": 0.14079068403974648,
      "grad_norm": 5.337882041931152,
      "learning_rate": 9.541986583011948e-05,
      "loss": 0.1966,
      "step": 3482
    },
    {
      "epoch": 0.14083111789501349,
      "grad_norm": 6.145480632781982,
      "learning_rate": 9.541719905436395e-05,
      "loss": 0.2488,
      "step": 3483
    },
    {
      "epoch": 0.14087155175028052,
      "grad_norm": 7.086443901062012,
      "learning_rate": 9.541453153975888e-05,
      "loss": 0.2888,
      "step": 3484
    },
    {
      "epoch": 0.14091198560554752,
      "grad_norm": 3.4092772006988525,
      "learning_rate": 9.541186328634768e-05,
      "loss": 0.1387,
      "step": 3485
    },
    {
      "epoch": 0.14095241946081455,
      "grad_norm": 2.6334125995635986,
      "learning_rate": 9.540919429417375e-05,
      "loss": 0.14,
      "step": 3486
    },
    {
      "epoch": 0.14099285331608155,
      "grad_norm": 3.785665512084961,
      "learning_rate": 9.540652456328051e-05,
      "loss": 0.1477,
      "step": 3487
    },
    {
      "epoch": 0.14103328717134858,
      "grad_norm": 4.7195844650268555,
      "learning_rate": 9.540385409371139e-05,
      "loss": 0.2029,
      "step": 3488
    },
    {
      "epoch": 0.14107372102661558,
      "grad_norm": 6.95353364944458,
      "learning_rate": 9.540118288550983e-05,
      "loss": 0.439,
      "step": 3489
    },
    {
      "epoch": 0.14111415488188261,
      "grad_norm": 3.2179110050201416,
      "learning_rate": 9.539851093871928e-05,
      "loss": 0.1185,
      "step": 3490
    },
    {
      "epoch": 0.14115458873714962,
      "grad_norm": 2.02966570854187,
      "learning_rate": 9.539583825338323e-05,
      "loss": 0.1322,
      "step": 3491
    },
    {
      "epoch": 0.14119502259241662,
      "grad_norm": 6.967922210693359,
      "learning_rate": 9.539316482954514e-05,
      "loss": 0.1817,
      "step": 3492
    },
    {
      "epoch": 0.14123545644768365,
      "grad_norm": 4.862983703613281,
      "learning_rate": 9.53904906672485e-05,
      "loss": 0.1976,
      "step": 3493
    },
    {
      "epoch": 0.14127589030295065,
      "grad_norm": 4.221317291259766,
      "learning_rate": 9.538781576653684e-05,
      "loss": 0.201,
      "step": 3494
    },
    {
      "epoch": 0.14131632415821768,
      "grad_norm": 8.198760032653809,
      "learning_rate": 9.538514012745364e-05,
      "loss": 0.2576,
      "step": 3495
    },
    {
      "epoch": 0.14135675801348468,
      "grad_norm": 7.142951011657715,
      "learning_rate": 9.538246375004243e-05,
      "loss": 0.2291,
      "step": 3496
    },
    {
      "epoch": 0.14139719186875171,
      "grad_norm": 2.2888295650482178,
      "learning_rate": 9.537978663434678e-05,
      "loss": 0.113,
      "step": 3497
    },
    {
      "epoch": 0.14143762572401872,
      "grad_norm": 4.752518653869629,
      "learning_rate": 9.537710878041022e-05,
      "loss": 0.1898,
      "step": 3498
    },
    {
      "epoch": 0.14147805957928575,
      "grad_norm": 4.310927391052246,
      "learning_rate": 9.537443018827632e-05,
      "loss": 0.1097,
      "step": 3499
    },
    {
      "epoch": 0.14151849343455275,
      "grad_norm": 9.1107759475708,
      "learning_rate": 9.537175085798865e-05,
      "loss": 0.3414,
      "step": 3500
    },
    {
      "epoch": 0.14155892728981975,
      "grad_norm": 1.988946557044983,
      "learning_rate": 9.536907078959081e-05,
      "loss": 0.1555,
      "step": 3501
    },
    {
      "epoch": 0.14159936114508678,
      "grad_norm": 6.458177089691162,
      "learning_rate": 9.536638998312637e-05,
      "loss": 0.2348,
      "step": 3502
    },
    {
      "epoch": 0.14163979500035379,
      "grad_norm": 2.3811652660369873,
      "learning_rate": 9.536370843863897e-05,
      "loss": 0.1021,
      "step": 3503
    },
    {
      "epoch": 0.14168022885562082,
      "grad_norm": 12.195796966552734,
      "learning_rate": 9.536102615617222e-05,
      "loss": 0.246,
      "step": 3504
    },
    {
      "epoch": 0.14172066271088782,
      "grad_norm": 6.400260925292969,
      "learning_rate": 9.535834313576976e-05,
      "loss": 0.2184,
      "step": 3505
    },
    {
      "epoch": 0.14176109656615485,
      "grad_norm": 11.695612907409668,
      "learning_rate": 9.535565937747525e-05,
      "loss": 0.3749,
      "step": 3506
    },
    {
      "epoch": 0.14180153042142185,
      "grad_norm": 4.476245403289795,
      "learning_rate": 9.535297488133235e-05,
      "loss": 0.1262,
      "step": 3507
    },
    {
      "epoch": 0.14184196427668888,
      "grad_norm": 3.1585700511932373,
      "learning_rate": 9.535028964738469e-05,
      "loss": 0.1598,
      "step": 3508
    },
    {
      "epoch": 0.14188239813195588,
      "grad_norm": 3.915647029876709,
      "learning_rate": 9.534760367567598e-05,
      "loss": 0.1652,
      "step": 3509
    },
    {
      "epoch": 0.1419228319872229,
      "grad_norm": 8.169148445129395,
      "learning_rate": 9.534491696624995e-05,
      "loss": 0.1857,
      "step": 3510
    },
    {
      "epoch": 0.14196326584248992,
      "grad_norm": 2.92764949798584,
      "learning_rate": 9.534222951915025e-05,
      "loss": 0.1234,
      "step": 3511
    },
    {
      "epoch": 0.14200369969775692,
      "grad_norm": 5.074685573577881,
      "learning_rate": 9.533954133442064e-05,
      "loss": 0.226,
      "step": 3512
    },
    {
      "epoch": 0.14204413355302395,
      "grad_norm": 10.167046546936035,
      "learning_rate": 9.533685241210482e-05,
      "loss": 0.3274,
      "step": 3513
    },
    {
      "epoch": 0.14208456740829095,
      "grad_norm": 8.613748550415039,
      "learning_rate": 9.533416275224655e-05,
      "loss": 0.1769,
      "step": 3514
    },
    {
      "epoch": 0.14212500126355798,
      "grad_norm": 4.754901885986328,
      "learning_rate": 9.53314723548896e-05,
      "loss": 0.1419,
      "step": 3515
    },
    {
      "epoch": 0.14216543511882498,
      "grad_norm": 8.998748779296875,
      "learning_rate": 9.532878122007773e-05,
      "loss": 0.3114,
      "step": 3516
    },
    {
      "epoch": 0.14220586897409201,
      "grad_norm": 2.663538694381714,
      "learning_rate": 9.532608934785469e-05,
      "loss": 0.0849,
      "step": 3517
    },
    {
      "epoch": 0.14224630282935902,
      "grad_norm": 6.175215244293213,
      "learning_rate": 9.532339673826432e-05,
      "loss": 0.18,
      "step": 3518
    },
    {
      "epoch": 0.14228673668462605,
      "grad_norm": 11.401808738708496,
      "learning_rate": 9.532070339135038e-05,
      "loss": 0.2985,
      "step": 3519
    },
    {
      "epoch": 0.14232717053989305,
      "grad_norm": 3.6207220554351807,
      "learning_rate": 9.53180093071567e-05,
      "loss": 0.1343,
      "step": 3520
    },
    {
      "epoch": 0.14236760439516008,
      "grad_norm": 3.699906349182129,
      "learning_rate": 9.53153144857271e-05,
      "loss": 0.0714,
      "step": 3521
    },
    {
      "epoch": 0.14240803825042708,
      "grad_norm": 3.5816192626953125,
      "learning_rate": 9.531261892710546e-05,
      "loss": 0.1608,
      "step": 3522
    },
    {
      "epoch": 0.14244847210569409,
      "grad_norm": 4.288010120391846,
      "learning_rate": 9.530992263133558e-05,
      "loss": 0.1339,
      "step": 3523
    },
    {
      "epoch": 0.14248890596096112,
      "grad_norm": 6.6092848777771,
      "learning_rate": 9.530722559846135e-05,
      "loss": 0.3646,
      "step": 3524
    },
    {
      "epoch": 0.14252933981622812,
      "grad_norm": 6.828031539916992,
      "learning_rate": 9.530452782852666e-05,
      "loss": 0.1907,
      "step": 3525
    },
    {
      "epoch": 0.14256977367149515,
      "grad_norm": 3.3427200317382812,
      "learning_rate": 9.530182932157535e-05,
      "loss": 0.2809,
      "step": 3526
    },
    {
      "epoch": 0.14261020752676215,
      "grad_norm": 2.8049211502075195,
      "learning_rate": 9.529913007765137e-05,
      "loss": 0.1519,
      "step": 3527
    },
    {
      "epoch": 0.14265064138202918,
      "grad_norm": 3.7867610454559326,
      "learning_rate": 9.529643009679859e-05,
      "loss": 0.0819,
      "step": 3528
    },
    {
      "epoch": 0.14269107523729618,
      "grad_norm": 4.715887546539307,
      "learning_rate": 9.529372937906096e-05,
      "loss": 0.1376,
      "step": 3529
    },
    {
      "epoch": 0.1427315090925632,
      "grad_norm": 6.557296276092529,
      "learning_rate": 9.529102792448241e-05,
      "loss": 0.2091,
      "step": 3530
    },
    {
      "epoch": 0.14277194294783022,
      "grad_norm": 3.190204381942749,
      "learning_rate": 9.528832573310688e-05,
      "loss": 0.163,
      "step": 3531
    },
    {
      "epoch": 0.14281237680309725,
      "grad_norm": 7.828081130981445,
      "learning_rate": 9.528562280497835e-05,
      "loss": 0.2399,
      "step": 3532
    },
    {
      "epoch": 0.14285281065836425,
      "grad_norm": 4.016374588012695,
      "learning_rate": 9.528291914014078e-05,
      "loss": 0.2176,
      "step": 3533
    },
    {
      "epoch": 0.14289324451363125,
      "grad_norm": 4.293730735778809,
      "learning_rate": 9.528021473863814e-05,
      "loss": 0.1736,
      "step": 3534
    },
    {
      "epoch": 0.14293367836889828,
      "grad_norm": 9.165038108825684,
      "learning_rate": 9.527750960051442e-05,
      "loss": 0.225,
      "step": 3535
    },
    {
      "epoch": 0.14297411222416528,
      "grad_norm": 7.1044020652771,
      "learning_rate": 9.527480372581366e-05,
      "loss": 0.1698,
      "step": 3536
    },
    {
      "epoch": 0.14301454607943231,
      "grad_norm": 5.043811321258545,
      "learning_rate": 9.527209711457985e-05,
      "loss": 0.2385,
      "step": 3537
    },
    {
      "epoch": 0.14305497993469932,
      "grad_norm": 12.505127906799316,
      "learning_rate": 9.526938976685703e-05,
      "loss": 0.3119,
      "step": 3538
    },
    {
      "epoch": 0.14309541378996635,
      "grad_norm": 5.109428405761719,
      "learning_rate": 9.526668168268927e-05,
      "loss": 0.1464,
      "step": 3539
    },
    {
      "epoch": 0.14313584764523335,
      "grad_norm": 4.42938232421875,
      "learning_rate": 9.526397286212056e-05,
      "loss": 0.16,
      "step": 3540
    },
    {
      "epoch": 0.14317628150050038,
      "grad_norm": 6.182464599609375,
      "learning_rate": 9.526126330519505e-05,
      "loss": 0.1411,
      "step": 3541
    },
    {
      "epoch": 0.14321671535576738,
      "grad_norm": 7.565199851989746,
      "learning_rate": 9.525855301195676e-05,
      "loss": 0.3131,
      "step": 3542
    },
    {
      "epoch": 0.1432571492110344,
      "grad_norm": 2.3426923751831055,
      "learning_rate": 9.52558419824498e-05,
      "loss": 0.1222,
      "step": 3543
    },
    {
      "epoch": 0.14329758306630141,
      "grad_norm": 6.209098815917969,
      "learning_rate": 9.525313021671827e-05,
      "loss": 0.1696,
      "step": 3544
    },
    {
      "epoch": 0.14333801692156842,
      "grad_norm": 7.735261917114258,
      "learning_rate": 9.525041771480629e-05,
      "loss": 0.1602,
      "step": 3545
    },
    {
      "epoch": 0.14337845077683545,
      "grad_norm": 1.8661118745803833,
      "learning_rate": 9.5247704476758e-05,
      "loss": 0.0648,
      "step": 3546
    },
    {
      "epoch": 0.14341888463210245,
      "grad_norm": 6.103165149688721,
      "learning_rate": 9.52449905026175e-05,
      "loss": 0.2131,
      "step": 3547
    },
    {
      "epoch": 0.14345931848736948,
      "grad_norm": 4.132080554962158,
      "learning_rate": 9.524227579242897e-05,
      "loss": 0.2243,
      "step": 3548
    },
    {
      "epoch": 0.14349975234263648,
      "grad_norm": 4.822537899017334,
      "learning_rate": 9.523956034623657e-05,
      "loss": 0.2135,
      "step": 3549
    },
    {
      "epoch": 0.1435401861979035,
      "grad_norm": 7.660507678985596,
      "learning_rate": 9.523684416408448e-05,
      "loss": 0.2727,
      "step": 3550
    },
    {
      "epoch": 0.14358062005317052,
      "grad_norm": 14.923781394958496,
      "learning_rate": 9.523412724601688e-05,
      "loss": 0.3346,
      "step": 3551
    },
    {
      "epoch": 0.14362105390843755,
      "grad_norm": 10.441078186035156,
      "learning_rate": 9.523140959207797e-05,
      "loss": 0.2675,
      "step": 3552
    },
    {
      "epoch": 0.14366148776370455,
      "grad_norm": 14.549656867980957,
      "learning_rate": 9.522869120231194e-05,
      "loss": 0.3879,
      "step": 3553
    },
    {
      "epoch": 0.14370192161897158,
      "grad_norm": 13.475183486938477,
      "learning_rate": 9.522597207676305e-05,
      "loss": 0.2876,
      "step": 3554
    },
    {
      "epoch": 0.14374235547423858,
      "grad_norm": 6.732460021972656,
      "learning_rate": 9.522325221547552e-05,
      "loss": 0.2887,
      "step": 3555
    },
    {
      "epoch": 0.14378278932950558,
      "grad_norm": 7.2200188636779785,
      "learning_rate": 9.522053161849358e-05,
      "loss": 0.1818,
      "step": 3556
    },
    {
      "epoch": 0.14382322318477261,
      "grad_norm": 11.236397743225098,
      "learning_rate": 9.521781028586151e-05,
      "loss": 0.2708,
      "step": 3557
    },
    {
      "epoch": 0.14386365704003962,
      "grad_norm": 8.794376373291016,
      "learning_rate": 9.521508821762358e-05,
      "loss": 0.2037,
      "step": 3558
    },
    {
      "epoch": 0.14390409089530665,
      "grad_norm": 9.606871604919434,
      "learning_rate": 9.521236541382406e-05,
      "loss": 0.2606,
      "step": 3559
    },
    {
      "epoch": 0.14394452475057365,
      "grad_norm": 3.9226810932159424,
      "learning_rate": 9.520964187450726e-05,
      "loss": 0.1798,
      "step": 3560
    },
    {
      "epoch": 0.14398495860584068,
      "grad_norm": 6.695024013519287,
      "learning_rate": 9.520691759971746e-05,
      "loss": 0.1841,
      "step": 3561
    },
    {
      "epoch": 0.14402539246110768,
      "grad_norm": 7.2114949226379395,
      "learning_rate": 9.520419258949901e-05,
      "loss": 0.1855,
      "step": 3562
    },
    {
      "epoch": 0.1440658263163747,
      "grad_norm": 5.713003158569336,
      "learning_rate": 9.520146684389624e-05,
      "loss": 0.1536,
      "step": 3563
    },
    {
      "epoch": 0.14410626017164171,
      "grad_norm": 3.973740816116333,
      "learning_rate": 9.519874036295346e-05,
      "loss": 0.0741,
      "step": 3564
    },
    {
      "epoch": 0.14414669402690872,
      "grad_norm": 10.78139591217041,
      "learning_rate": 9.519601314671505e-05,
      "loss": 0.2662,
      "step": 3565
    },
    {
      "epoch": 0.14418712788217575,
      "grad_norm": 5.399466037750244,
      "learning_rate": 9.519328519522538e-05,
      "loss": 0.2141,
      "step": 3566
    },
    {
      "epoch": 0.14422756173744275,
      "grad_norm": 1.398573637008667,
      "learning_rate": 9.519055650852882e-05,
      "loss": 0.1206,
      "step": 3567
    },
    {
      "epoch": 0.14426799559270978,
      "grad_norm": 5.266632080078125,
      "learning_rate": 9.518782708666975e-05,
      "loss": 0.2008,
      "step": 3568
    },
    {
      "epoch": 0.14430842944797678,
      "grad_norm": 4.158613204956055,
      "learning_rate": 9.518509692969258e-05,
      "loss": 0.2065,
      "step": 3569
    },
    {
      "epoch": 0.1443488633032438,
      "grad_norm": 5.3150129318237305,
      "learning_rate": 9.518236603764172e-05,
      "loss": 0.0965,
      "step": 3570
    },
    {
      "epoch": 0.14438929715851082,
      "grad_norm": 5.025362014770508,
      "learning_rate": 9.517963441056163e-05,
      "loss": 0.1509,
      "step": 3571
    },
    {
      "epoch": 0.14442973101377785,
      "grad_norm": 6.710464954376221,
      "learning_rate": 9.51769020484967e-05,
      "loss": 0.1721,
      "step": 3572
    },
    {
      "epoch": 0.14447016486904485,
      "grad_norm": 3.5823814868927,
      "learning_rate": 9.517416895149141e-05,
      "loss": 0.147,
      "step": 3573
    },
    {
      "epoch": 0.14451059872431188,
      "grad_norm": 8.714547157287598,
      "learning_rate": 9.517143511959021e-05,
      "loss": 0.1947,
      "step": 3574
    },
    {
      "epoch": 0.14455103257957888,
      "grad_norm": 9.510795593261719,
      "learning_rate": 9.516870055283758e-05,
      "loss": 0.2448,
      "step": 3575
    },
    {
      "epoch": 0.14459146643484588,
      "grad_norm": 4.553340435028076,
      "learning_rate": 9.516596525127801e-05,
      "loss": 0.104,
      "step": 3576
    },
    {
      "epoch": 0.1446319002901129,
      "grad_norm": 3.836205005645752,
      "learning_rate": 9.516322921495599e-05,
      "loss": 0.1728,
      "step": 3577
    },
    {
      "epoch": 0.14467233414537992,
      "grad_norm": 5.005370616912842,
      "learning_rate": 9.516049244391602e-05,
      "loss": 0.1904,
      "step": 3578
    },
    {
      "epoch": 0.14471276800064695,
      "grad_norm": 5.744956970214844,
      "learning_rate": 9.515775493820265e-05,
      "loss": 0.1849,
      "step": 3579
    },
    {
      "epoch": 0.14475320185591395,
      "grad_norm": 6.338112831115723,
      "learning_rate": 9.515501669786037e-05,
      "loss": 0.2844,
      "step": 3580
    },
    {
      "epoch": 0.14479363571118098,
      "grad_norm": 3.083073377609253,
      "learning_rate": 9.515227772293377e-05,
      "loss": 0.1098,
      "step": 3581
    },
    {
      "epoch": 0.14483406956644798,
      "grad_norm": 6.373285293579102,
      "learning_rate": 9.514953801346742e-05,
      "loss": 0.2482,
      "step": 3582
    },
    {
      "epoch": 0.144874503421715,
      "grad_norm": 3.063445806503296,
      "learning_rate": 9.514679756950581e-05,
      "loss": 0.0978,
      "step": 3583
    },
    {
      "epoch": 0.14491493727698201,
      "grad_norm": 4.30936336517334,
      "learning_rate": 9.51440563910936e-05,
      "loss": 0.2081,
      "step": 3584
    },
    {
      "epoch": 0.14495537113224904,
      "grad_norm": 1.4939998388290405,
      "learning_rate": 9.514131447827537e-05,
      "loss": 0.0589,
      "step": 3585
    },
    {
      "epoch": 0.14499580498751605,
      "grad_norm": 5.381535530090332,
      "learning_rate": 9.51385718310957e-05,
      "loss": 0.27,
      "step": 3586
    },
    {
      "epoch": 0.14503623884278305,
      "grad_norm": 9.377782821655273,
      "learning_rate": 9.513582844959921e-05,
      "loss": 0.2347,
      "step": 3587
    },
    {
      "epoch": 0.14507667269805008,
      "grad_norm": 12.708470344543457,
      "learning_rate": 9.513308433383054e-05,
      "loss": 0.3596,
      "step": 3588
    },
    {
      "epoch": 0.14511710655331708,
      "grad_norm": 5.963825702667236,
      "learning_rate": 9.513033948383435e-05,
      "loss": 0.1559,
      "step": 3589
    },
    {
      "epoch": 0.1451575404085841,
      "grad_norm": 5.686419486999512,
      "learning_rate": 9.512759389965527e-05,
      "loss": 0.1641,
      "step": 3590
    },
    {
      "epoch": 0.14519797426385112,
      "grad_norm": 4.6488847732543945,
      "learning_rate": 9.512484758133795e-05,
      "loss": 0.2438,
      "step": 3591
    },
    {
      "epoch": 0.14523840811911815,
      "grad_norm": 3.6143500804901123,
      "learning_rate": 9.512210052892709e-05,
      "loss": 0.0675,
      "step": 3592
    },
    {
      "epoch": 0.14527884197438515,
      "grad_norm": 5.957464694976807,
      "learning_rate": 9.511935274246739e-05,
      "loss": 0.1846,
      "step": 3593
    },
    {
      "epoch": 0.14531927582965218,
      "grad_norm": 6.955528736114502,
      "learning_rate": 9.511660422200353e-05,
      "loss": 0.4437,
      "step": 3594
    },
    {
      "epoch": 0.14535970968491918,
      "grad_norm": 3.3821303844451904,
      "learning_rate": 9.511385496758024e-05,
      "loss": 0.0676,
      "step": 3595
    },
    {
      "epoch": 0.1454001435401862,
      "grad_norm": 8.985218048095703,
      "learning_rate": 9.511110497924221e-05,
      "loss": 0.1878,
      "step": 3596
    },
    {
      "epoch": 0.1454405773954532,
      "grad_norm": 20.3598690032959,
      "learning_rate": 9.510835425703421e-05,
      "loss": 0.2726,
      "step": 3597
    },
    {
      "epoch": 0.14548101125072022,
      "grad_norm": 11.261207580566406,
      "learning_rate": 9.510560280100099e-05,
      "loss": 0.2423,
      "step": 3598
    },
    {
      "epoch": 0.14552144510598725,
      "grad_norm": 10.633523941040039,
      "learning_rate": 9.51028506111873e-05,
      "loss": 0.2722,
      "step": 3599
    },
    {
      "epoch": 0.14556187896125425,
      "grad_norm": 4.1680006980896,
      "learning_rate": 9.510009768763791e-05,
      "loss": 0.1022,
      "step": 3600
    },
    {
      "epoch": 0.14560231281652128,
      "grad_norm": 11.2153902053833,
      "learning_rate": 9.509734403039762e-05,
      "loss": 0.3125,
      "step": 3601
    },
    {
      "epoch": 0.14564274667178828,
      "grad_norm": 11.552436828613281,
      "learning_rate": 9.509458963951122e-05,
      "loss": 0.2564,
      "step": 3602
    },
    {
      "epoch": 0.1456831805270553,
      "grad_norm": 2.2904021739959717,
      "learning_rate": 9.509183451502348e-05,
      "loss": 0.0693,
      "step": 3603
    },
    {
      "epoch": 0.14572361438232231,
      "grad_norm": 6.059868812561035,
      "learning_rate": 9.50890786569793e-05,
      "loss": 0.1747,
      "step": 3604
    },
    {
      "epoch": 0.14576404823758934,
      "grad_norm": 6.084963321685791,
      "learning_rate": 9.508632206542343e-05,
      "loss": 0.2375,
      "step": 3605
    },
    {
      "epoch": 0.14580448209285635,
      "grad_norm": 5.2960357666015625,
      "learning_rate": 9.508356474040076e-05,
      "loss": 0.2605,
      "step": 3606
    },
    {
      "epoch": 0.14584491594812338,
      "grad_norm": 1.8537983894348145,
      "learning_rate": 9.508080668195616e-05,
      "loss": 0.1482,
      "step": 3607
    },
    {
      "epoch": 0.14588534980339038,
      "grad_norm": 3.0579564571380615,
      "learning_rate": 9.507804789013446e-05,
      "loss": 0.1365,
      "step": 3608
    },
    {
      "epoch": 0.14592578365865738,
      "grad_norm": 7.764875888824463,
      "learning_rate": 9.507528836498056e-05,
      "loss": 0.1216,
      "step": 3609
    },
    {
      "epoch": 0.1459662175139244,
      "grad_norm": 3.3119819164276123,
      "learning_rate": 9.507252810653935e-05,
      "loss": 0.0799,
      "step": 3610
    },
    {
      "epoch": 0.14600665136919141,
      "grad_norm": 2.8724396228790283,
      "learning_rate": 9.506976711485572e-05,
      "loss": 0.1236,
      "step": 3611
    },
    {
      "epoch": 0.14604708522445844,
      "grad_norm": 6.645023345947266,
      "learning_rate": 9.506700538997462e-05,
      "loss": 0.1857,
      "step": 3612
    },
    {
      "epoch": 0.14608751907972545,
      "grad_norm": 5.365087032318115,
      "learning_rate": 9.506424293194094e-05,
      "loss": 0.1009,
      "step": 3613
    },
    {
      "epoch": 0.14612795293499248,
      "grad_norm": 2.345421552658081,
      "learning_rate": 9.506147974079965e-05,
      "loss": 0.0787,
      "step": 3614
    },
    {
      "epoch": 0.14616838679025948,
      "grad_norm": 4.766054153442383,
      "learning_rate": 9.505871581659566e-05,
      "loss": 0.2233,
      "step": 3615
    },
    {
      "epoch": 0.1462088206455265,
      "grad_norm": 10.421212196350098,
      "learning_rate": 9.505595115937398e-05,
      "loss": 0.3105,
      "step": 3616
    },
    {
      "epoch": 0.1462492545007935,
      "grad_norm": 2.724578857421875,
      "learning_rate": 9.505318576917958e-05,
      "loss": 0.1023,
      "step": 3617
    },
    {
      "epoch": 0.14628968835606054,
      "grad_norm": 9.253836631774902,
      "learning_rate": 9.505041964605743e-05,
      "loss": 0.2023,
      "step": 3618
    },
    {
      "epoch": 0.14633012221132755,
      "grad_norm": 2.1477210521698,
      "learning_rate": 9.504765279005252e-05,
      "loss": 0.1134,
      "step": 3619
    },
    {
      "epoch": 0.14637055606659455,
      "grad_norm": 4.7746710777282715,
      "learning_rate": 9.504488520120989e-05,
      "loss": 0.2162,
      "step": 3620
    },
    {
      "epoch": 0.14641098992186158,
      "grad_norm": 5.264647960662842,
      "learning_rate": 9.504211687957454e-05,
      "loss": 0.1453,
      "step": 3621
    },
    {
      "epoch": 0.14645142377712858,
      "grad_norm": 3.960792064666748,
      "learning_rate": 9.503934782519151e-05,
      "loss": 0.0775,
      "step": 3622
    },
    {
      "epoch": 0.1464918576323956,
      "grad_norm": 4.75243616104126,
      "learning_rate": 9.503657803810585e-05,
      "loss": 0.1374,
      "step": 3623
    },
    {
      "epoch": 0.1465322914876626,
      "grad_norm": 7.293806552886963,
      "learning_rate": 9.503380751836264e-05,
      "loss": 0.1498,
      "step": 3624
    },
    {
      "epoch": 0.14657272534292964,
      "grad_norm": 2.5165674686431885,
      "learning_rate": 9.503103626600691e-05,
      "loss": 0.1147,
      "step": 3625
    },
    {
      "epoch": 0.14661315919819665,
      "grad_norm": 4.987305164337158,
      "learning_rate": 9.502826428108378e-05,
      "loss": 0.1784,
      "step": 3626
    },
    {
      "epoch": 0.14665359305346368,
      "grad_norm": 1.9346894025802612,
      "learning_rate": 9.502549156363832e-05,
      "loss": 0.0834,
      "step": 3627
    },
    {
      "epoch": 0.14669402690873068,
      "grad_norm": 3.9735419750213623,
      "learning_rate": 9.502271811371565e-05,
      "loss": 0.1757,
      "step": 3628
    },
    {
      "epoch": 0.14673446076399768,
      "grad_norm": 2.787835121154785,
      "learning_rate": 9.501994393136088e-05,
      "loss": 0.106,
      "step": 3629
    },
    {
      "epoch": 0.1467748946192647,
      "grad_norm": 4.147335529327393,
      "learning_rate": 9.501716901661916e-05,
      "loss": 0.1001,
      "step": 3630
    },
    {
      "epoch": 0.14681532847453171,
      "grad_norm": 10.251703262329102,
      "learning_rate": 9.50143933695356e-05,
      "loss": 0.2179,
      "step": 3631
    },
    {
      "epoch": 0.14685576232979874,
      "grad_norm": 3.626887559890747,
      "learning_rate": 9.501161699015538e-05,
      "loss": 0.1358,
      "step": 3632
    },
    {
      "epoch": 0.14689619618506575,
      "grad_norm": 4.441482067108154,
      "learning_rate": 9.500883987852366e-05,
      "loss": 0.1471,
      "step": 3633
    },
    {
      "epoch": 0.14693663004033278,
      "grad_norm": 4.888454437255859,
      "learning_rate": 9.500606203468561e-05,
      "loss": 0.1835,
      "step": 3634
    },
    {
      "epoch": 0.14697706389559978,
      "grad_norm": 3.987251043319702,
      "learning_rate": 9.500328345868643e-05,
      "loss": 0.1008,
      "step": 3635
    },
    {
      "epoch": 0.1470174977508668,
      "grad_norm": 6.624938011169434,
      "learning_rate": 9.500050415057131e-05,
      "loss": 0.1675,
      "step": 3636
    },
    {
      "epoch": 0.1470579316061338,
      "grad_norm": 4.840104103088379,
      "learning_rate": 9.499772411038549e-05,
      "loss": 0.2248,
      "step": 3637
    },
    {
      "epoch": 0.14709836546140084,
      "grad_norm": 5.64577054977417,
      "learning_rate": 9.499494333817416e-05,
      "loss": 0.1265,
      "step": 3638
    },
    {
      "epoch": 0.14713879931666785,
      "grad_norm": 5.176467418670654,
      "learning_rate": 9.499216183398259e-05,
      "loss": 0.3216,
      "step": 3639
    },
    {
      "epoch": 0.14717923317193485,
      "grad_norm": 9.303807258605957,
      "learning_rate": 9.4989379597856e-05,
      "loss": 0.2406,
      "step": 3640
    },
    {
      "epoch": 0.14721966702720188,
      "grad_norm": 2.8068654537200928,
      "learning_rate": 9.498659662983969e-05,
      "loss": 0.1626,
      "step": 3641
    },
    {
      "epoch": 0.14726010088246888,
      "grad_norm": 8.067621231079102,
      "learning_rate": 9.498381292997889e-05,
      "loss": 0.1888,
      "step": 3642
    },
    {
      "epoch": 0.1473005347377359,
      "grad_norm": 8.372068405151367,
      "learning_rate": 9.498102849831892e-05,
      "loss": 0.2725,
      "step": 3643
    },
    {
      "epoch": 0.1473409685930029,
      "grad_norm": 6.026985168457031,
      "learning_rate": 9.497824333490505e-05,
      "loss": 0.2211,
      "step": 3644
    },
    {
      "epoch": 0.14738140244826994,
      "grad_norm": 5.652502059936523,
      "learning_rate": 9.497545743978262e-05,
      "loss": 0.2072,
      "step": 3645
    },
    {
      "epoch": 0.14742183630353695,
      "grad_norm": 4.133243560791016,
      "learning_rate": 9.497267081299692e-05,
      "loss": 0.1098,
      "step": 3646
    },
    {
      "epoch": 0.14746227015880398,
      "grad_norm": 6.351276874542236,
      "learning_rate": 9.49698834545933e-05,
      "loss": 0.1122,
      "step": 3647
    },
    {
      "epoch": 0.14750270401407098,
      "grad_norm": 2.116190195083618,
      "learning_rate": 9.49670953646171e-05,
      "loss": 0.0762,
      "step": 3648
    },
    {
      "epoch": 0.147543137869338,
      "grad_norm": 11.030269622802734,
      "learning_rate": 9.496430654311368e-05,
      "loss": 0.3832,
      "step": 3649
    },
    {
      "epoch": 0.147583571724605,
      "grad_norm": 6.983233451843262,
      "learning_rate": 9.496151699012842e-05,
      "loss": 0.2826,
      "step": 3650
    },
    {
      "epoch": 0.14762400557987201,
      "grad_norm": 7.577662944793701,
      "learning_rate": 9.495872670570667e-05,
      "loss": 0.2639,
      "step": 3651
    },
    {
      "epoch": 0.14766443943513904,
      "grad_norm": 8.977791786193848,
      "learning_rate": 9.495593568989385e-05,
      "loss": 0.3254,
      "step": 3652
    },
    {
      "epoch": 0.14770487329040605,
      "grad_norm": 6.280385971069336,
      "learning_rate": 9.495314394273535e-05,
      "loss": 0.1618,
      "step": 3653
    },
    {
      "epoch": 0.14774530714567308,
      "grad_norm": 3.6032893657684326,
      "learning_rate": 9.495035146427659e-05,
      "loss": 0.1329,
      "step": 3654
    },
    {
      "epoch": 0.14778574100094008,
      "grad_norm": 3.755681037902832,
      "learning_rate": 9.4947558254563e-05,
      "loss": 0.1845,
      "step": 3655
    },
    {
      "epoch": 0.1478261748562071,
      "grad_norm": 5.496229648590088,
      "learning_rate": 9.494476431364002e-05,
      "loss": 0.2273,
      "step": 3656
    },
    {
      "epoch": 0.1478666087114741,
      "grad_norm": 8.844586372375488,
      "learning_rate": 9.494196964155311e-05,
      "loss": 0.2537,
      "step": 3657
    },
    {
      "epoch": 0.14790704256674114,
      "grad_norm": 6.221546649932861,
      "learning_rate": 9.49391742383477e-05,
      "loss": 0.2545,
      "step": 3658
    },
    {
      "epoch": 0.14794747642200815,
      "grad_norm": 6.7134552001953125,
      "learning_rate": 9.493637810406932e-05,
      "loss": 0.2562,
      "step": 3659
    },
    {
      "epoch": 0.14798791027727518,
      "grad_norm": 5.941163063049316,
      "learning_rate": 9.493358123876341e-05,
      "loss": 0.417,
      "step": 3660
    },
    {
      "epoch": 0.14802834413254218,
      "grad_norm": 4.243958473205566,
      "learning_rate": 9.49307836424755e-05,
      "loss": 0.1701,
      "step": 3661
    },
    {
      "epoch": 0.14806877798780918,
      "grad_norm": 3.383915662765503,
      "learning_rate": 9.492798531525108e-05,
      "loss": 0.0662,
      "step": 3662
    },
    {
      "epoch": 0.1481092118430762,
      "grad_norm": 4.4046783447265625,
      "learning_rate": 9.49251862571357e-05,
      "loss": 0.2479,
      "step": 3663
    },
    {
      "epoch": 0.1481496456983432,
      "grad_norm": 4.992476940155029,
      "learning_rate": 9.492238646817486e-05,
      "loss": 0.1639,
      "step": 3664
    },
    {
      "epoch": 0.14819007955361024,
      "grad_norm": 4.410444736480713,
      "learning_rate": 9.491958594841413e-05,
      "loss": 0.1968,
      "step": 3665
    },
    {
      "epoch": 0.14823051340887725,
      "grad_norm": 3.337266206741333,
      "learning_rate": 9.491678469789906e-05,
      "loss": 0.1786,
      "step": 3666
    },
    {
      "epoch": 0.14827094726414428,
      "grad_norm": 5.053743362426758,
      "learning_rate": 9.491398271667522e-05,
      "loss": 0.2373,
      "step": 3667
    },
    {
      "epoch": 0.14831138111941128,
      "grad_norm": 6.058218002319336,
      "learning_rate": 9.49111800047882e-05,
      "loss": 0.1727,
      "step": 3668
    },
    {
      "epoch": 0.1483518149746783,
      "grad_norm": 4.869316101074219,
      "learning_rate": 9.490837656228361e-05,
      "loss": 0.1379,
      "step": 3669
    },
    {
      "epoch": 0.1483922488299453,
      "grad_norm": 3.3773868083953857,
      "learning_rate": 9.490557238920704e-05,
      "loss": 0.2151,
      "step": 3670
    },
    {
      "epoch": 0.14843268268521234,
      "grad_norm": 4.804349899291992,
      "learning_rate": 9.490276748560409e-05,
      "loss": 0.1482,
      "step": 3671
    },
    {
      "epoch": 0.14847311654047934,
      "grad_norm": 3.505582809448242,
      "learning_rate": 9.489996185152042e-05,
      "loss": 0.1399,
      "step": 3672
    },
    {
      "epoch": 0.14851355039574635,
      "grad_norm": 3.347930908203125,
      "learning_rate": 9.489715548700165e-05,
      "loss": 0.1907,
      "step": 3673
    },
    {
      "epoch": 0.14855398425101338,
      "grad_norm": 5.165981769561768,
      "learning_rate": 9.489434839209344e-05,
      "loss": 0.1696,
      "step": 3674
    },
    {
      "epoch": 0.14859441810628038,
      "grad_norm": 4.113088130950928,
      "learning_rate": 9.489154056684148e-05,
      "loss": 0.2119,
      "step": 3675
    },
    {
      "epoch": 0.1486348519615474,
      "grad_norm": 5.662906169891357,
      "learning_rate": 9.488873201129142e-05,
      "loss": 0.0939,
      "step": 3676
    },
    {
      "epoch": 0.1486752858168144,
      "grad_norm": 4.09470796585083,
      "learning_rate": 9.488592272548895e-05,
      "loss": 0.1829,
      "step": 3677
    },
    {
      "epoch": 0.14871571967208144,
      "grad_norm": 5.277957916259766,
      "learning_rate": 9.488311270947978e-05,
      "loss": 0.1618,
      "step": 3678
    },
    {
      "epoch": 0.14875615352734844,
      "grad_norm": 8.107226371765137,
      "learning_rate": 9.488030196330964e-05,
      "loss": 0.2113,
      "step": 3679
    },
    {
      "epoch": 0.14879658738261548,
      "grad_norm": 4.942506790161133,
      "learning_rate": 9.487749048702421e-05,
      "loss": 0.1386,
      "step": 3680
    },
    {
      "epoch": 0.14883702123788248,
      "grad_norm": 3.246331214904785,
      "learning_rate": 9.487467828066929e-05,
      "loss": 0.1087,
      "step": 3681
    },
    {
      "epoch": 0.1488774550931495,
      "grad_norm": 14.222970962524414,
      "learning_rate": 9.487186534429056e-05,
      "loss": 0.5191,
      "step": 3682
    },
    {
      "epoch": 0.1489178889484165,
      "grad_norm": 2.43156099319458,
      "learning_rate": 9.486905167793384e-05,
      "loss": 0.1218,
      "step": 3683
    },
    {
      "epoch": 0.1489583228036835,
      "grad_norm": 8.677425384521484,
      "learning_rate": 9.486623728164488e-05,
      "loss": 0.299,
      "step": 3684
    },
    {
      "epoch": 0.14899875665895054,
      "grad_norm": 3.9678187370300293,
      "learning_rate": 9.486342215546944e-05,
      "loss": 0.1864,
      "step": 3685
    },
    {
      "epoch": 0.14903919051421755,
      "grad_norm": 7.879222869873047,
      "learning_rate": 9.486060629945336e-05,
      "loss": 0.1602,
      "step": 3686
    },
    {
      "epoch": 0.14907962436948458,
      "grad_norm": 1.6493757963180542,
      "learning_rate": 9.485778971364244e-05,
      "loss": 0.1079,
      "step": 3687
    },
    {
      "epoch": 0.14912005822475158,
      "grad_norm": 4.9593071937561035,
      "learning_rate": 9.485497239808245e-05,
      "loss": 0.2614,
      "step": 3688
    },
    {
      "epoch": 0.1491604920800186,
      "grad_norm": 2.8561298847198486,
      "learning_rate": 9.485215435281928e-05,
      "loss": 0.1358,
      "step": 3689
    },
    {
      "epoch": 0.1492009259352856,
      "grad_norm": 2.9083399772644043,
      "learning_rate": 9.484933557789877e-05,
      "loss": 0.0701,
      "step": 3690
    },
    {
      "epoch": 0.14924135979055264,
      "grad_norm": 2.5807747840881348,
      "learning_rate": 9.484651607336675e-05,
      "loss": 0.0627,
      "step": 3691
    },
    {
      "epoch": 0.14928179364581964,
      "grad_norm": 5.619513034820557,
      "learning_rate": 9.484369583926908e-05,
      "loss": 0.1663,
      "step": 3692
    },
    {
      "epoch": 0.14932222750108665,
      "grad_norm": 3.7795348167419434,
      "learning_rate": 9.48408748756517e-05,
      "loss": 0.0989,
      "step": 3693
    },
    {
      "epoch": 0.14936266135635368,
      "grad_norm": 3.99269437789917,
      "learning_rate": 9.483805318256042e-05,
      "loss": 0.2,
      "step": 3694
    },
    {
      "epoch": 0.14940309521162068,
      "grad_norm": 7.8895463943481445,
      "learning_rate": 9.483523076004121e-05,
      "loss": 0.1637,
      "step": 3695
    },
    {
      "epoch": 0.1494435290668877,
      "grad_norm": 2.9790096282958984,
      "learning_rate": 9.483240760813995e-05,
      "loss": 0.155,
      "step": 3696
    },
    {
      "epoch": 0.1494839629221547,
      "grad_norm": 4.249456882476807,
      "learning_rate": 9.482958372690259e-05,
      "loss": 0.1418,
      "step": 3697
    },
    {
      "epoch": 0.14952439677742174,
      "grad_norm": 2.402366876602173,
      "learning_rate": 9.482675911637503e-05,
      "loss": 0.0906,
      "step": 3698
    },
    {
      "epoch": 0.14956483063268874,
      "grad_norm": 2.885572910308838,
      "learning_rate": 9.482393377660325e-05,
      "loss": 0.1172,
      "step": 3699
    },
    {
      "epoch": 0.14960526448795577,
      "grad_norm": 7.036840915679932,
      "learning_rate": 9.482110770763322e-05,
      "loss": 0.2843,
      "step": 3700
    },
    {
      "epoch": 0.14964569834322278,
      "grad_norm": 9.39486312866211,
      "learning_rate": 9.48182809095109e-05,
      "loss": 0.3003,
      "step": 3701
    },
    {
      "epoch": 0.1496861321984898,
      "grad_norm": 6.045462131500244,
      "learning_rate": 9.48154533822823e-05,
      "loss": 0.2122,
      "step": 3702
    },
    {
      "epoch": 0.1497265660537568,
      "grad_norm": 8.831894874572754,
      "learning_rate": 9.481262512599337e-05,
      "loss": 0.1788,
      "step": 3703
    },
    {
      "epoch": 0.1497669999090238,
      "grad_norm": 11.60100269317627,
      "learning_rate": 9.480979614069016e-05,
      "loss": 0.2568,
      "step": 3704
    },
    {
      "epoch": 0.14980743376429084,
      "grad_norm": 5.256930351257324,
      "learning_rate": 9.480696642641868e-05,
      "loss": 0.19,
      "step": 3705
    },
    {
      "epoch": 0.14984786761955785,
      "grad_norm": 6.611375331878662,
      "learning_rate": 9.480413598322497e-05,
      "loss": 0.1302,
      "step": 3706
    },
    {
      "epoch": 0.14988830147482488,
      "grad_norm": 2.5165176391601562,
      "learning_rate": 9.480130481115507e-05,
      "loss": 0.1791,
      "step": 3707
    },
    {
      "epoch": 0.14992873533009188,
      "grad_norm": 6.024064540863037,
      "learning_rate": 9.479847291025503e-05,
      "loss": 0.3289,
      "step": 3708
    },
    {
      "epoch": 0.1499691691853589,
      "grad_norm": 3.119025468826294,
      "learning_rate": 9.479564028057093e-05,
      "loss": 0.1086,
      "step": 3709
    },
    {
      "epoch": 0.1500096030406259,
      "grad_norm": 4.412282943725586,
      "learning_rate": 9.479280692214886e-05,
      "loss": 0.1793,
      "step": 3710
    },
    {
      "epoch": 0.15005003689589294,
      "grad_norm": 4.447229385375977,
      "learning_rate": 9.478997283503489e-05,
      "loss": 0.1316,
      "step": 3711
    },
    {
      "epoch": 0.15009047075115994,
      "grad_norm": 6.6623334884643555,
      "learning_rate": 9.478713801927514e-05,
      "loss": 0.1657,
      "step": 3712
    },
    {
      "epoch": 0.15013090460642697,
      "grad_norm": 3.1518280506134033,
      "learning_rate": 9.478430247491573e-05,
      "loss": 0.1799,
      "step": 3713
    },
    {
      "epoch": 0.15017133846169398,
      "grad_norm": 8.263324737548828,
      "learning_rate": 9.478146620200278e-05,
      "loss": 0.2358,
      "step": 3714
    },
    {
      "epoch": 0.15021177231696098,
      "grad_norm": 5.523159503936768,
      "learning_rate": 9.477862920058244e-05,
      "loss": 0.1831,
      "step": 3715
    },
    {
      "epoch": 0.150252206172228,
      "grad_norm": 2.409398078918457,
      "learning_rate": 9.477579147070085e-05,
      "loss": 0.1039,
      "step": 3716
    },
    {
      "epoch": 0.150292640027495,
      "grad_norm": 2.8854329586029053,
      "learning_rate": 9.477295301240419e-05,
      "loss": 0.1382,
      "step": 3717
    },
    {
      "epoch": 0.15033307388276204,
      "grad_norm": 3.339097738265991,
      "learning_rate": 9.477011382573862e-05,
      "loss": 0.1865,
      "step": 3718
    },
    {
      "epoch": 0.15037350773802904,
      "grad_norm": 4.33925199508667,
      "learning_rate": 9.476727391075033e-05,
      "loss": 0.1646,
      "step": 3719
    },
    {
      "epoch": 0.15041394159329607,
      "grad_norm": 3.271284341812134,
      "learning_rate": 9.476443326748552e-05,
      "loss": 0.2218,
      "step": 3720
    },
    {
      "epoch": 0.15045437544856308,
      "grad_norm": 5.060245990753174,
      "learning_rate": 9.476159189599041e-05,
      "loss": 0.1422,
      "step": 3721
    },
    {
      "epoch": 0.1504948093038301,
      "grad_norm": 5.723391532897949,
      "learning_rate": 9.475874979631122e-05,
      "loss": 0.1293,
      "step": 3722
    },
    {
      "epoch": 0.1505352431590971,
      "grad_norm": 5.670523643493652,
      "learning_rate": 9.475590696849421e-05,
      "loss": 0.1881,
      "step": 3723
    },
    {
      "epoch": 0.15057567701436414,
      "grad_norm": 3.682007312774658,
      "learning_rate": 9.475306341258559e-05,
      "loss": 0.1673,
      "step": 3724
    },
    {
      "epoch": 0.15061611086963114,
      "grad_norm": 3.6934540271759033,
      "learning_rate": 9.475021912863162e-05,
      "loss": 0.1988,
      "step": 3725
    },
    {
      "epoch": 0.15065654472489814,
      "grad_norm": 2.651862144470215,
      "learning_rate": 9.47473741166786e-05,
      "loss": 0.1522,
      "step": 3726
    },
    {
      "epoch": 0.15069697858016518,
      "grad_norm": 2.4241786003112793,
      "learning_rate": 9.474452837677278e-05,
      "loss": 0.1158,
      "step": 3727
    },
    {
      "epoch": 0.15073741243543218,
      "grad_norm": 3.4790494441986084,
      "learning_rate": 9.474168190896048e-05,
      "loss": 0.1669,
      "step": 3728
    },
    {
      "epoch": 0.1507778462906992,
      "grad_norm": 5.553155422210693,
      "learning_rate": 9.473883471328799e-05,
      "loss": 0.1325,
      "step": 3729
    },
    {
      "epoch": 0.1508182801459662,
      "grad_norm": 4.276768684387207,
      "learning_rate": 9.473598678980164e-05,
      "loss": 0.0784,
      "step": 3730
    },
    {
      "epoch": 0.15085871400123324,
      "grad_norm": 6.989223480224609,
      "learning_rate": 9.473313813854776e-05,
      "loss": 0.2963,
      "step": 3731
    },
    {
      "epoch": 0.15089914785650024,
      "grad_norm": 10.404269218444824,
      "learning_rate": 9.473028875957269e-05,
      "loss": 0.2557,
      "step": 3732
    },
    {
      "epoch": 0.15093958171176727,
      "grad_norm": 4.555271625518799,
      "learning_rate": 9.472743865292277e-05,
      "loss": 0.1429,
      "step": 3733
    },
    {
      "epoch": 0.15098001556703428,
      "grad_norm": 5.636605262756348,
      "learning_rate": 9.472458781864437e-05,
      "loss": 0.1176,
      "step": 3734
    },
    {
      "epoch": 0.1510204494223013,
      "grad_norm": 5.930817127227783,
      "learning_rate": 9.472173625678388e-05,
      "loss": 0.2015,
      "step": 3735
    },
    {
      "epoch": 0.1510608832775683,
      "grad_norm": 6.259466171264648,
      "learning_rate": 9.47188839673877e-05,
      "loss": 0.1456,
      "step": 3736
    },
    {
      "epoch": 0.1511013171328353,
      "grad_norm": 5.237694263458252,
      "learning_rate": 9.47160309505022e-05,
      "loss": 0.1507,
      "step": 3737
    },
    {
      "epoch": 0.15114175098810234,
      "grad_norm": 4.839032173156738,
      "learning_rate": 9.471317720617383e-05,
      "loss": 0.128,
      "step": 3738
    },
    {
      "epoch": 0.15118218484336934,
      "grad_norm": 9.266863822937012,
      "learning_rate": 9.471032273444898e-05,
      "loss": 0.3037,
      "step": 3739
    },
    {
      "epoch": 0.15122261869863637,
      "grad_norm": 8.319262504577637,
      "learning_rate": 9.470746753537409e-05,
      "loss": 0.3235,
      "step": 3740
    },
    {
      "epoch": 0.15126305255390338,
      "grad_norm": 3.619023084640503,
      "learning_rate": 9.470461160899563e-05,
      "loss": 0.163,
      "step": 3741
    },
    {
      "epoch": 0.1513034864091704,
      "grad_norm": 6.550958633422852,
      "learning_rate": 9.470175495536004e-05,
      "loss": 0.2982,
      "step": 3742
    },
    {
      "epoch": 0.1513439202644374,
      "grad_norm": 8.213431358337402,
      "learning_rate": 9.469889757451381e-05,
      "loss": 0.3599,
      "step": 3743
    },
    {
      "epoch": 0.15138435411970444,
      "grad_norm": 3.0018224716186523,
      "learning_rate": 9.46960394665034e-05,
      "loss": 0.1333,
      "step": 3744
    },
    {
      "epoch": 0.15142478797497144,
      "grad_norm": 1.6334502696990967,
      "learning_rate": 9.469318063137535e-05,
      "loss": 0.1016,
      "step": 3745
    },
    {
      "epoch": 0.15146522183023847,
      "grad_norm": 4.1590704917907715,
      "learning_rate": 9.469032106917612e-05,
      "loss": 0.3908,
      "step": 3746
    },
    {
      "epoch": 0.15150565568550547,
      "grad_norm": 5.730081081390381,
      "learning_rate": 9.468746077995225e-05,
      "loss": 0.178,
      "step": 3747
    },
    {
      "epoch": 0.15154608954077248,
      "grad_norm": 4.684865474700928,
      "learning_rate": 9.468459976375026e-05,
      "loss": 0.1481,
      "step": 3748
    },
    {
      "epoch": 0.1515865233960395,
      "grad_norm": 6.7382893562316895,
      "learning_rate": 9.46817380206167e-05,
      "loss": 0.1782,
      "step": 3749
    },
    {
      "epoch": 0.1516269572513065,
      "grad_norm": 3.130275249481201,
      "learning_rate": 9.467887555059816e-05,
      "loss": 0.1441,
      "step": 3750
    },
    {
      "epoch": 0.15166739110657354,
      "grad_norm": 10.040006637573242,
      "learning_rate": 9.467601235374114e-05,
      "loss": 0.2766,
      "step": 3751
    },
    {
      "epoch": 0.15170782496184054,
      "grad_norm": 8.853872299194336,
      "learning_rate": 9.467314843009228e-05,
      "loss": 0.1863,
      "step": 3752
    },
    {
      "epoch": 0.15174825881710757,
      "grad_norm": 7.201104640960693,
      "learning_rate": 9.467028377969813e-05,
      "loss": 0.1726,
      "step": 3753
    },
    {
      "epoch": 0.15178869267237458,
      "grad_norm": 6.390852928161621,
      "learning_rate": 9.466741840260531e-05,
      "loss": 0.2076,
      "step": 3754
    },
    {
      "epoch": 0.1518291265276416,
      "grad_norm": 7.485623836517334,
      "learning_rate": 9.466455229886041e-05,
      "loss": 0.1979,
      "step": 3755
    },
    {
      "epoch": 0.1518695603829086,
      "grad_norm": 6.928185939788818,
      "learning_rate": 9.46616854685101e-05,
      "loss": 0.1299,
      "step": 3756
    },
    {
      "epoch": 0.1519099942381756,
      "grad_norm": 7.853607654571533,
      "learning_rate": 9.465881791160098e-05,
      "loss": 0.2528,
      "step": 3757
    },
    {
      "epoch": 0.15195042809344264,
      "grad_norm": 2.5768885612487793,
      "learning_rate": 9.465594962817972e-05,
      "loss": 0.1489,
      "step": 3758
    },
    {
      "epoch": 0.15199086194870964,
      "grad_norm": 5.599687576293945,
      "learning_rate": 9.465308061829298e-05,
      "loss": 0.2543,
      "step": 3759
    },
    {
      "epoch": 0.15203129580397667,
      "grad_norm": 4.213963031768799,
      "learning_rate": 9.465021088198742e-05,
      "loss": 0.13,
      "step": 3760
    },
    {
      "epoch": 0.15207172965924368,
      "grad_norm": 6.685833930969238,
      "learning_rate": 9.464734041930974e-05,
      "loss": 0.2958,
      "step": 3761
    },
    {
      "epoch": 0.1521121635145107,
      "grad_norm": 3.4054441452026367,
      "learning_rate": 9.464446923030663e-05,
      "loss": 0.1428,
      "step": 3762
    },
    {
      "epoch": 0.1521525973697777,
      "grad_norm": 5.274366855621338,
      "learning_rate": 9.464159731502477e-05,
      "loss": 0.2438,
      "step": 3763
    },
    {
      "epoch": 0.15219303122504474,
      "grad_norm": 5.733479022979736,
      "learning_rate": 9.463872467351093e-05,
      "loss": 0.1949,
      "step": 3764
    },
    {
      "epoch": 0.15223346508031174,
      "grad_norm": 2.8698132038116455,
      "learning_rate": 9.463585130581183e-05,
      "loss": 0.1414,
      "step": 3765
    },
    {
      "epoch": 0.15227389893557877,
      "grad_norm": 6.193872928619385,
      "learning_rate": 9.463297721197421e-05,
      "loss": 0.1969,
      "step": 3766
    },
    {
      "epoch": 0.15231433279084577,
      "grad_norm": 1.670049786567688,
      "learning_rate": 9.46301023920448e-05,
      "loss": 0.1191,
      "step": 3767
    },
    {
      "epoch": 0.15235476664611278,
      "grad_norm": 4.365565776824951,
      "learning_rate": 9.46272268460704e-05,
      "loss": 0.1363,
      "step": 3768
    },
    {
      "epoch": 0.1523952005013798,
      "grad_norm": 2.783111095428467,
      "learning_rate": 9.462435057409778e-05,
      "loss": 0.1118,
      "step": 3769
    },
    {
      "epoch": 0.1524356343566468,
      "grad_norm": 3.5459089279174805,
      "learning_rate": 9.462147357617372e-05,
      "loss": 0.0974,
      "step": 3770
    },
    {
      "epoch": 0.15247606821191384,
      "grad_norm": 4.504378795623779,
      "learning_rate": 9.461859585234505e-05,
      "loss": 0.1266,
      "step": 3771
    },
    {
      "epoch": 0.15251650206718084,
      "grad_norm": 2.9683282375335693,
      "learning_rate": 9.461571740265856e-05,
      "loss": 0.1579,
      "step": 3772
    },
    {
      "epoch": 0.15255693592244787,
      "grad_norm": 3.1088032722473145,
      "learning_rate": 9.461283822716107e-05,
      "loss": 0.116,
      "step": 3773
    },
    {
      "epoch": 0.15259736977771488,
      "grad_norm": 3.8027939796447754,
      "learning_rate": 9.460995832589945e-05,
      "loss": 0.0985,
      "step": 3774
    },
    {
      "epoch": 0.1526378036329819,
      "grad_norm": 8.391703605651855,
      "learning_rate": 9.460707769892054e-05,
      "loss": 0.2382,
      "step": 3775
    },
    {
      "epoch": 0.1526782374882489,
      "grad_norm": 6.943277359008789,
      "learning_rate": 9.460419634627118e-05,
      "loss": 0.3178,
      "step": 3776
    },
    {
      "epoch": 0.15271867134351594,
      "grad_norm": 12.006828308105469,
      "learning_rate": 9.460131426799826e-05,
      "loss": 0.2545,
      "step": 3777
    },
    {
      "epoch": 0.15275910519878294,
      "grad_norm": 4.6843061447143555,
      "learning_rate": 9.459843146414867e-05,
      "loss": 0.1916,
      "step": 3778
    },
    {
      "epoch": 0.15279953905404994,
      "grad_norm": 5.476874828338623,
      "learning_rate": 9.45955479347693e-05,
      "loss": 0.1334,
      "step": 3779
    },
    {
      "epoch": 0.15283997290931697,
      "grad_norm": 4.993621349334717,
      "learning_rate": 9.459266367990708e-05,
      "loss": 0.2999,
      "step": 3780
    },
    {
      "epoch": 0.15288040676458398,
      "grad_norm": 3.7252249717712402,
      "learning_rate": 9.45897786996089e-05,
      "loss": 0.2498,
      "step": 3781
    },
    {
      "epoch": 0.152920840619851,
      "grad_norm": 6.399251937866211,
      "learning_rate": 9.45868929939217e-05,
      "loss": 0.1937,
      "step": 3782
    },
    {
      "epoch": 0.152961274475118,
      "grad_norm": 5.786250591278076,
      "learning_rate": 9.458400656289244e-05,
      "loss": 0.2588,
      "step": 3783
    },
    {
      "epoch": 0.15300170833038504,
      "grad_norm": 2.853736639022827,
      "learning_rate": 9.458111940656807e-05,
      "loss": 0.1948,
      "step": 3784
    },
    {
      "epoch": 0.15304214218565204,
      "grad_norm": 6.001939296722412,
      "learning_rate": 9.457823152499556e-05,
      "loss": 0.2975,
      "step": 3785
    },
    {
      "epoch": 0.15308257604091907,
      "grad_norm": 6.598135948181152,
      "learning_rate": 9.457534291822188e-05,
      "loss": 0.2883,
      "step": 3786
    },
    {
      "epoch": 0.15312300989618607,
      "grad_norm": 10.814576148986816,
      "learning_rate": 9.457245358629403e-05,
      "loss": 0.2054,
      "step": 3787
    },
    {
      "epoch": 0.1531634437514531,
      "grad_norm": 11.645605087280273,
      "learning_rate": 9.456956352925902e-05,
      "loss": 0.3405,
      "step": 3788
    },
    {
      "epoch": 0.1532038776067201,
      "grad_norm": 7.346028804779053,
      "learning_rate": 9.456667274716386e-05,
      "loss": 0.2927,
      "step": 3789
    },
    {
      "epoch": 0.1532443114619871,
      "grad_norm": 6.7894744873046875,
      "learning_rate": 9.456378124005557e-05,
      "loss": 0.1342,
      "step": 3790
    },
    {
      "epoch": 0.15328474531725414,
      "grad_norm": 2.9470226764678955,
      "learning_rate": 9.456088900798119e-05,
      "loss": 0.1419,
      "step": 3791
    },
    {
      "epoch": 0.15332517917252114,
      "grad_norm": 1.743079662322998,
      "learning_rate": 9.455799605098778e-05,
      "loss": 0.1084,
      "step": 3792
    },
    {
      "epoch": 0.15336561302778817,
      "grad_norm": 3.6039302349090576,
      "learning_rate": 9.45551023691224e-05,
      "loss": 0.1432,
      "step": 3793
    },
    {
      "epoch": 0.15340604688305517,
      "grad_norm": 4.440608501434326,
      "learning_rate": 9.455220796243213e-05,
      "loss": 0.2586,
      "step": 3794
    },
    {
      "epoch": 0.1534464807383222,
      "grad_norm": 9.051816940307617,
      "learning_rate": 9.454931283096405e-05,
      "loss": 0.1945,
      "step": 3795
    },
    {
      "epoch": 0.1534869145935892,
      "grad_norm": 4.880588054656982,
      "learning_rate": 9.454641697476523e-05,
      "loss": 0.1192,
      "step": 3796
    },
    {
      "epoch": 0.15352734844885624,
      "grad_norm": 5.812407970428467,
      "learning_rate": 9.454352039388283e-05,
      "loss": 0.1599,
      "step": 3797
    },
    {
      "epoch": 0.15356778230412324,
      "grad_norm": 2.2018239498138428,
      "learning_rate": 9.454062308836396e-05,
      "loss": 0.1892,
      "step": 3798
    },
    {
      "epoch": 0.15360821615939027,
      "grad_norm": 4.90041446685791,
      "learning_rate": 9.453772505825574e-05,
      "loss": 0.1861,
      "step": 3799
    },
    {
      "epoch": 0.15364865001465727,
      "grad_norm": 4.139484405517578,
      "learning_rate": 9.45348263036053e-05,
      "loss": 0.1892,
      "step": 3800
    },
    {
      "epoch": 0.15368908386992428,
      "grad_norm": 4.849770545959473,
      "learning_rate": 9.453192682445982e-05,
      "loss": 0.2252,
      "step": 3801
    },
    {
      "epoch": 0.1537295177251913,
      "grad_norm": 2.962653160095215,
      "learning_rate": 9.452902662086648e-05,
      "loss": 0.3328,
      "step": 3802
    },
    {
      "epoch": 0.1537699515804583,
      "grad_norm": 2.6649117469787598,
      "learning_rate": 9.452612569287243e-05,
      "loss": 0.0864,
      "step": 3803
    },
    {
      "epoch": 0.15381038543572534,
      "grad_norm": 4.027615070343018,
      "learning_rate": 9.452322404052489e-05,
      "loss": 0.2059,
      "step": 3804
    },
    {
      "epoch": 0.15385081929099234,
      "grad_norm": 7.38258171081543,
      "learning_rate": 9.452032166387105e-05,
      "loss": 0.2146,
      "step": 3805
    },
    {
      "epoch": 0.15389125314625937,
      "grad_norm": 5.996294975280762,
      "learning_rate": 9.451741856295813e-05,
      "loss": 0.1416,
      "step": 3806
    },
    {
      "epoch": 0.15393168700152637,
      "grad_norm": 6.179449081420898,
      "learning_rate": 9.451451473783335e-05,
      "loss": 0.1031,
      "step": 3807
    },
    {
      "epoch": 0.1539721208567934,
      "grad_norm": 3.101788282394409,
      "learning_rate": 9.451161018854396e-05,
      "loss": 0.0993,
      "step": 3808
    },
    {
      "epoch": 0.1540125547120604,
      "grad_norm": 6.623003959655762,
      "learning_rate": 9.450870491513719e-05,
      "loss": 0.2454,
      "step": 3809
    },
    {
      "epoch": 0.15405298856732744,
      "grad_norm": 6.263190746307373,
      "learning_rate": 9.450579891766034e-05,
      "loss": 0.1897,
      "step": 3810
    },
    {
      "epoch": 0.15409342242259444,
      "grad_norm": 6.16630744934082,
      "learning_rate": 9.450289219616064e-05,
      "loss": 0.1553,
      "step": 3811
    },
    {
      "epoch": 0.15413385627786144,
      "grad_norm": 3.082697868347168,
      "learning_rate": 9.449998475068542e-05,
      "loss": 0.2068,
      "step": 3812
    },
    {
      "epoch": 0.15417429013312847,
      "grad_norm": 2.7387399673461914,
      "learning_rate": 9.449707658128197e-05,
      "loss": 0.0739,
      "step": 3813
    },
    {
      "epoch": 0.15421472398839547,
      "grad_norm": 3.5621256828308105,
      "learning_rate": 9.449416768799757e-05,
      "loss": 0.2345,
      "step": 3814
    },
    {
      "epoch": 0.1542551578436625,
      "grad_norm": 5.047922134399414,
      "learning_rate": 9.449125807087957e-05,
      "loss": 0.2417,
      "step": 3815
    },
    {
      "epoch": 0.1542955916989295,
      "grad_norm": 2.8505866527557373,
      "learning_rate": 9.44883477299753e-05,
      "loss": 0.1853,
      "step": 3816
    },
    {
      "epoch": 0.15433602555419654,
      "grad_norm": 5.402769565582275,
      "learning_rate": 9.448543666533211e-05,
      "loss": 0.2206,
      "step": 3817
    },
    {
      "epoch": 0.15437645940946354,
      "grad_norm": 3.216437339782715,
      "learning_rate": 9.448252487699734e-05,
      "loss": 0.1012,
      "step": 3818
    },
    {
      "epoch": 0.15441689326473057,
      "grad_norm": 8.105317115783691,
      "learning_rate": 9.447961236501835e-05,
      "loss": 0.1624,
      "step": 3819
    },
    {
      "epoch": 0.15445732711999757,
      "grad_norm": 2.3100228309631348,
      "learning_rate": 9.447669912944257e-05,
      "loss": 0.0984,
      "step": 3820
    },
    {
      "epoch": 0.15449776097526458,
      "grad_norm": 6.778954029083252,
      "learning_rate": 9.447378517031734e-05,
      "loss": 0.1261,
      "step": 3821
    },
    {
      "epoch": 0.1545381948305316,
      "grad_norm": 5.145465850830078,
      "learning_rate": 9.44708704876901e-05,
      "loss": 0.1746,
      "step": 3822
    },
    {
      "epoch": 0.1545786286857986,
      "grad_norm": 2.688877582550049,
      "learning_rate": 9.446795508160824e-05,
      "loss": 0.1064,
      "step": 3823
    },
    {
      "epoch": 0.15461906254106564,
      "grad_norm": 3.716745615005493,
      "learning_rate": 9.44650389521192e-05,
      "loss": 0.1688,
      "step": 3824
    },
    {
      "epoch": 0.15465949639633264,
      "grad_norm": 7.248511791229248,
      "learning_rate": 9.446212209927044e-05,
      "loss": 0.1682,
      "step": 3825
    },
    {
      "epoch": 0.15469993025159967,
      "grad_norm": 4.7235307693481445,
      "learning_rate": 9.445920452310937e-05,
      "loss": 0.4129,
      "step": 3826
    },
    {
      "epoch": 0.15474036410686667,
      "grad_norm": 4.163604259490967,
      "learning_rate": 9.445628622368347e-05,
      "loss": 0.1615,
      "step": 3827
    },
    {
      "epoch": 0.1547807979621337,
      "grad_norm": 4.655007362365723,
      "learning_rate": 9.445336720104026e-05,
      "loss": 0.209,
      "step": 3828
    },
    {
      "epoch": 0.1548212318174007,
      "grad_norm": 6.235599040985107,
      "learning_rate": 9.445044745522716e-05,
      "loss": 0.2117,
      "step": 3829
    },
    {
      "epoch": 0.15486166567266774,
      "grad_norm": 5.924273490905762,
      "learning_rate": 9.44475269862917e-05,
      "loss": 0.1358,
      "step": 3830
    },
    {
      "epoch": 0.15490209952793474,
      "grad_norm": 4.347218990325928,
      "learning_rate": 9.444460579428136e-05,
      "loss": 0.1767,
      "step": 3831
    },
    {
      "epoch": 0.15494253338320174,
      "grad_norm": 1.302505373954773,
      "learning_rate": 9.444168387924372e-05,
      "loss": 0.0424,
      "step": 3832
    },
    {
      "epoch": 0.15498296723846877,
      "grad_norm": 4.043443202972412,
      "learning_rate": 9.443876124122627e-05,
      "loss": 0.1263,
      "step": 3833
    },
    {
      "epoch": 0.15502340109373577,
      "grad_norm": 7.978820323944092,
      "learning_rate": 9.443583788027657e-05,
      "loss": 0.2176,
      "step": 3834
    },
    {
      "epoch": 0.1550638349490028,
      "grad_norm": 9.002331733703613,
      "learning_rate": 9.443291379644217e-05,
      "loss": 0.1698,
      "step": 3835
    },
    {
      "epoch": 0.1551042688042698,
      "grad_norm": 1.968845248222351,
      "learning_rate": 9.442998898977064e-05,
      "loss": 0.1106,
      "step": 3836
    },
    {
      "epoch": 0.15514470265953684,
      "grad_norm": 5.474277973175049,
      "learning_rate": 9.442706346030958e-05,
      "loss": 0.1977,
      "step": 3837
    },
    {
      "epoch": 0.15518513651480384,
      "grad_norm": 6.681704998016357,
      "learning_rate": 9.442413720810657e-05,
      "loss": 0.2354,
      "step": 3838
    },
    {
      "epoch": 0.15522557037007087,
      "grad_norm": 4.654881000518799,
      "learning_rate": 9.44212102332092e-05,
      "loss": 0.2284,
      "step": 3839
    },
    {
      "epoch": 0.15526600422533787,
      "grad_norm": 5.171448230743408,
      "learning_rate": 9.441828253566509e-05,
      "loss": 0.1925,
      "step": 3840
    },
    {
      "epoch": 0.1553064380806049,
      "grad_norm": 6.151573181152344,
      "learning_rate": 9.44153541155219e-05,
      "loss": 0.2319,
      "step": 3841
    },
    {
      "epoch": 0.1553468719358719,
      "grad_norm": 4.371959209442139,
      "learning_rate": 9.441242497282722e-05,
      "loss": 0.2039,
      "step": 3842
    },
    {
      "epoch": 0.1553873057911389,
      "grad_norm": 3.06198787689209,
      "learning_rate": 9.440949510762875e-05,
      "loss": 0.2608,
      "step": 3843
    },
    {
      "epoch": 0.15542773964640594,
      "grad_norm": 1.3722504377365112,
      "learning_rate": 9.440656451997412e-05,
      "loss": 0.0817,
      "step": 3844
    },
    {
      "epoch": 0.15546817350167294,
      "grad_norm": 5.647851467132568,
      "learning_rate": 9.440363320991102e-05,
      "loss": 0.3342,
      "step": 3845
    },
    {
      "epoch": 0.15550860735693997,
      "grad_norm": 7.575093746185303,
      "learning_rate": 9.440070117748713e-05,
      "loss": 0.2847,
      "step": 3846
    },
    {
      "epoch": 0.15554904121220697,
      "grad_norm": 8.477709770202637,
      "learning_rate": 9.439776842275014e-05,
      "loss": 0.2292,
      "step": 3847
    },
    {
      "epoch": 0.155589475067474,
      "grad_norm": 16.04792594909668,
      "learning_rate": 9.439483494574779e-05,
      "loss": 0.33,
      "step": 3848
    },
    {
      "epoch": 0.155629908922741,
      "grad_norm": 2.6671972274780273,
      "learning_rate": 9.439190074652778e-05,
      "loss": 0.0694,
      "step": 3849
    },
    {
      "epoch": 0.15567034277800804,
      "grad_norm": 4.051419734954834,
      "learning_rate": 9.438896582513783e-05,
      "loss": 0.1472,
      "step": 3850
    },
    {
      "epoch": 0.15571077663327504,
      "grad_norm": 8.061614036560059,
      "learning_rate": 9.438603018162571e-05,
      "loss": 0.23,
      "step": 3851
    },
    {
      "epoch": 0.15575121048854207,
      "grad_norm": 5.55203914642334,
      "learning_rate": 9.438309381603919e-05,
      "loss": 0.2665,
      "step": 3852
    },
    {
      "epoch": 0.15579164434380907,
      "grad_norm": 9.448112487792969,
      "learning_rate": 9.4380156728426e-05,
      "loss": 0.2541,
      "step": 3853
    },
    {
      "epoch": 0.15583207819907607,
      "grad_norm": 9.801310539245605,
      "learning_rate": 9.437721891883393e-05,
      "loss": 0.2241,
      "step": 3854
    },
    {
      "epoch": 0.1558725120543431,
      "grad_norm": 9.34052562713623,
      "learning_rate": 9.43742803873108e-05,
      "loss": 0.1677,
      "step": 3855
    },
    {
      "epoch": 0.1559129459096101,
      "grad_norm": 5.697126865386963,
      "learning_rate": 9.437134113390438e-05,
      "loss": 0.2316,
      "step": 3856
    },
    {
      "epoch": 0.15595337976487714,
      "grad_norm": 4.816317558288574,
      "learning_rate": 9.43684011586625e-05,
      "loss": 0.1957,
      "step": 3857
    },
    {
      "epoch": 0.15599381362014414,
      "grad_norm": 4.081836223602295,
      "learning_rate": 9.436546046163301e-05,
      "loss": 0.123,
      "step": 3858
    },
    {
      "epoch": 0.15603424747541117,
      "grad_norm": 3.915832042694092,
      "learning_rate": 9.43625190428637e-05,
      "loss": 0.1502,
      "step": 3859
    },
    {
      "epoch": 0.15607468133067817,
      "grad_norm": 5.685534954071045,
      "learning_rate": 9.435957690240247e-05,
      "loss": 0.2383,
      "step": 3860
    },
    {
      "epoch": 0.1561151151859452,
      "grad_norm": 5.959407329559326,
      "learning_rate": 9.435663404029717e-05,
      "loss": 0.0968,
      "step": 3861
    },
    {
      "epoch": 0.1561555490412122,
      "grad_norm": 5.836497783660889,
      "learning_rate": 9.435369045659565e-05,
      "loss": 0.1274,
      "step": 3862
    },
    {
      "epoch": 0.15619598289647924,
      "grad_norm": 2.7592098712921143,
      "learning_rate": 9.435074615134583e-05,
      "loss": 0.1162,
      "step": 3863
    },
    {
      "epoch": 0.15623641675174624,
      "grad_norm": 7.13545036315918,
      "learning_rate": 9.434780112459557e-05,
      "loss": 0.2053,
      "step": 3864
    },
    {
      "epoch": 0.15627685060701324,
      "grad_norm": 5.421340465545654,
      "learning_rate": 9.434485537639282e-05,
      "loss": 0.1532,
      "step": 3865
    },
    {
      "epoch": 0.15631728446228027,
      "grad_norm": 10.374335289001465,
      "learning_rate": 9.434190890678547e-05,
      "loss": 0.461,
      "step": 3866
    },
    {
      "epoch": 0.15635771831754727,
      "grad_norm": 5.180912494659424,
      "learning_rate": 9.433896171582147e-05,
      "loss": 0.1139,
      "step": 3867
    },
    {
      "epoch": 0.1563981521728143,
      "grad_norm": 6.580761432647705,
      "learning_rate": 9.433601380354877e-05,
      "loss": 0.2805,
      "step": 3868
    },
    {
      "epoch": 0.1564385860280813,
      "grad_norm": 3.245499610900879,
      "learning_rate": 9.43330651700153e-05,
      "loss": 0.1944,
      "step": 3869
    },
    {
      "epoch": 0.15647901988334834,
      "grad_norm": 1.7182128429412842,
      "learning_rate": 9.433011581526908e-05,
      "loss": 0.0889,
      "step": 3870
    },
    {
      "epoch": 0.15651945373861534,
      "grad_norm": 4.1571879386901855,
      "learning_rate": 9.432716573935802e-05,
      "loss": 0.1873,
      "step": 3871
    },
    {
      "epoch": 0.15655988759388237,
      "grad_norm": 4.140661716461182,
      "learning_rate": 9.432421494233018e-05,
      "loss": 0.1777,
      "step": 3872
    },
    {
      "epoch": 0.15660032144914937,
      "grad_norm": 7.296037673950195,
      "learning_rate": 9.432126342423352e-05,
      "loss": 0.3098,
      "step": 3873
    },
    {
      "epoch": 0.1566407553044164,
      "grad_norm": 6.0830769538879395,
      "learning_rate": 9.431831118511607e-05,
      "loss": 0.1639,
      "step": 3874
    },
    {
      "epoch": 0.1566811891596834,
      "grad_norm": 8.831962585449219,
      "learning_rate": 9.431535822502586e-05,
      "loss": 0.1991,
      "step": 3875
    },
    {
      "epoch": 0.1567216230149504,
      "grad_norm": 7.3354597091674805,
      "learning_rate": 9.431240454401091e-05,
      "loss": 0.2752,
      "step": 3876
    },
    {
      "epoch": 0.15676205687021744,
      "grad_norm": 4.085724830627441,
      "learning_rate": 9.430945014211929e-05,
      "loss": 0.1269,
      "step": 3877
    },
    {
      "epoch": 0.15680249072548444,
      "grad_norm": 3.705946207046509,
      "learning_rate": 9.430649501939908e-05,
      "loss": 0.0859,
      "step": 3878
    },
    {
      "epoch": 0.15684292458075147,
      "grad_norm": 7.971224308013916,
      "learning_rate": 9.43035391758983e-05,
      "loss": 0.191,
      "step": 3879
    },
    {
      "epoch": 0.15688335843601847,
      "grad_norm": 2.7643980979919434,
      "learning_rate": 9.430058261166509e-05,
      "loss": 0.1398,
      "step": 3880
    },
    {
      "epoch": 0.1569237922912855,
      "grad_norm": 4.97778844833374,
      "learning_rate": 9.42976253267475e-05,
      "loss": 0.1083,
      "step": 3881
    },
    {
      "epoch": 0.1569642261465525,
      "grad_norm": 7.782564640045166,
      "learning_rate": 9.42946673211937e-05,
      "loss": 0.2331,
      "step": 3882
    },
    {
      "epoch": 0.15700466000181953,
      "grad_norm": 5.317015171051025,
      "learning_rate": 9.429170859505176e-05,
      "loss": 0.2284,
      "step": 3883
    },
    {
      "epoch": 0.15704509385708654,
      "grad_norm": 7.688607215881348,
      "learning_rate": 9.428874914836982e-05,
      "loss": 0.2238,
      "step": 3884
    },
    {
      "epoch": 0.15708552771235354,
      "grad_norm": 6.4563469886779785,
      "learning_rate": 9.428578898119605e-05,
      "loss": 0.1349,
      "step": 3885
    },
    {
      "epoch": 0.15712596156762057,
      "grad_norm": 5.776787281036377,
      "learning_rate": 9.428282809357856e-05,
      "loss": 0.154,
      "step": 3886
    },
    {
      "epoch": 0.15716639542288757,
      "grad_norm": 3.379429340362549,
      "learning_rate": 9.427986648556557e-05,
      "loss": 0.1372,
      "step": 3887
    },
    {
      "epoch": 0.1572068292781546,
      "grad_norm": 16.487733840942383,
      "learning_rate": 9.427690415720523e-05,
      "loss": 0.3227,
      "step": 3888
    },
    {
      "epoch": 0.1572472631334216,
      "grad_norm": 5.275142669677734,
      "learning_rate": 9.427394110854573e-05,
      "loss": 0.1778,
      "step": 3889
    },
    {
      "epoch": 0.15728769698868864,
      "grad_norm": 4.029419898986816,
      "learning_rate": 9.427097733963529e-05,
      "loss": 0.2868,
      "step": 3890
    },
    {
      "epoch": 0.15732813084395564,
      "grad_norm": 2.6036882400512695,
      "learning_rate": 9.426801285052212e-05,
      "loss": 0.2279,
      "step": 3891
    },
    {
      "epoch": 0.15736856469922267,
      "grad_norm": 4.313218593597412,
      "learning_rate": 9.426504764125443e-05,
      "loss": 0.1535,
      "step": 3892
    },
    {
      "epoch": 0.15740899855448967,
      "grad_norm": 6.274306774139404,
      "learning_rate": 9.426208171188048e-05,
      "loss": 0.1897,
      "step": 3893
    },
    {
      "epoch": 0.1574494324097567,
      "grad_norm": 3.7983455657958984,
      "learning_rate": 9.42591150624485e-05,
      "loss": 0.0897,
      "step": 3894
    },
    {
      "epoch": 0.1574898662650237,
      "grad_norm": 3.8698556423187256,
      "learning_rate": 9.425614769300676e-05,
      "loss": 0.1491,
      "step": 3895
    },
    {
      "epoch": 0.1575303001202907,
      "grad_norm": 2.058260679244995,
      "learning_rate": 9.425317960360356e-05,
      "loss": 0.0685,
      "step": 3896
    },
    {
      "epoch": 0.15757073397555774,
      "grad_norm": 4.933915615081787,
      "learning_rate": 9.425021079428714e-05,
      "loss": 0.1878,
      "step": 3897
    },
    {
      "epoch": 0.15761116783082474,
      "grad_norm": 4.984874725341797,
      "learning_rate": 9.424724126510582e-05,
      "loss": 0.1178,
      "step": 3898
    },
    {
      "epoch": 0.15765160168609177,
      "grad_norm": 5.095334529876709,
      "learning_rate": 9.424427101610789e-05,
      "loss": 0.2413,
      "step": 3899
    },
    {
      "epoch": 0.15769203554135877,
      "grad_norm": 7.319599628448486,
      "learning_rate": 9.42413000473417e-05,
      "loss": 0.265,
      "step": 3900
    },
    {
      "epoch": 0.1577324693966258,
      "grad_norm": 5.022380352020264,
      "learning_rate": 9.423832835885558e-05,
      "loss": 0.3016,
      "step": 3901
    },
    {
      "epoch": 0.1577729032518928,
      "grad_norm": 6.956135272979736,
      "learning_rate": 9.423535595069784e-05,
      "loss": 0.4031,
      "step": 3902
    },
    {
      "epoch": 0.15781333710715983,
      "grad_norm": 6.33903169631958,
      "learning_rate": 9.423238282291687e-05,
      "loss": 0.1973,
      "step": 3903
    },
    {
      "epoch": 0.15785377096242684,
      "grad_norm": 5.547525405883789,
      "learning_rate": 9.422940897556101e-05,
      "loss": 0.1495,
      "step": 3904
    },
    {
      "epoch": 0.15789420481769387,
      "grad_norm": 3.9827146530151367,
      "learning_rate": 9.422643440867865e-05,
      "loss": 0.0822,
      "step": 3905
    },
    {
      "epoch": 0.15793463867296087,
      "grad_norm": 7.431812286376953,
      "learning_rate": 9.422345912231819e-05,
      "loss": 0.1566,
      "step": 3906
    },
    {
      "epoch": 0.15797507252822787,
      "grad_norm": 5.663488388061523,
      "learning_rate": 9.422048311652803e-05,
      "loss": 0.1369,
      "step": 3907
    },
    {
      "epoch": 0.1580155063834949,
      "grad_norm": 3.3580334186553955,
      "learning_rate": 9.421750639135658e-05,
      "loss": 0.2481,
      "step": 3908
    },
    {
      "epoch": 0.1580559402387619,
      "grad_norm": 3.076226234436035,
      "learning_rate": 9.421452894685225e-05,
      "loss": 0.1096,
      "step": 3909
    },
    {
      "epoch": 0.15809637409402894,
      "grad_norm": 3.682767868041992,
      "learning_rate": 9.42115507830635e-05,
      "loss": 0.0888,
      "step": 3910
    },
    {
      "epoch": 0.15813680794929594,
      "grad_norm": 2.5411577224731445,
      "learning_rate": 9.420857190003876e-05,
      "loss": 0.1458,
      "step": 3911
    },
    {
      "epoch": 0.15817724180456297,
      "grad_norm": 8.812243461608887,
      "learning_rate": 9.420559229782652e-05,
      "loss": 0.2366,
      "step": 3912
    },
    {
      "epoch": 0.15821767565982997,
      "grad_norm": 6.4262566566467285,
      "learning_rate": 9.420261197647522e-05,
      "loss": 0.1949,
      "step": 3913
    },
    {
      "epoch": 0.158258109515097,
      "grad_norm": 5.141708850860596,
      "learning_rate": 9.419963093603334e-05,
      "loss": 0.3415,
      "step": 3914
    },
    {
      "epoch": 0.158298543370364,
      "grad_norm": 4.587250232696533,
      "learning_rate": 9.419664917654942e-05,
      "loss": 0.2767,
      "step": 3915
    },
    {
      "epoch": 0.15833897722563103,
      "grad_norm": 3.729949712753296,
      "learning_rate": 9.419366669807194e-05,
      "loss": 0.199,
      "step": 3916
    },
    {
      "epoch": 0.15837941108089804,
      "grad_norm": 3.074167013168335,
      "learning_rate": 9.419068350064941e-05,
      "loss": 0.201,
      "step": 3917
    },
    {
      "epoch": 0.15841984493616504,
      "grad_norm": 5.1262946128845215,
      "learning_rate": 9.418769958433036e-05,
      "loss": 0.132,
      "step": 3918
    },
    {
      "epoch": 0.15846027879143207,
      "grad_norm": 10.79518985748291,
      "learning_rate": 9.418471494916337e-05,
      "loss": 0.2845,
      "step": 3919
    },
    {
      "epoch": 0.15850071264669907,
      "grad_norm": 2.344374895095825,
      "learning_rate": 9.418172959519693e-05,
      "loss": 0.1363,
      "step": 3920
    },
    {
      "epoch": 0.1585411465019661,
      "grad_norm": 6.977919578552246,
      "learning_rate": 9.417874352247967e-05,
      "loss": 0.2963,
      "step": 3921
    },
    {
      "epoch": 0.1585815803572331,
      "grad_norm": 8.296462059020996,
      "learning_rate": 9.417575673106014e-05,
      "loss": 0.2135,
      "step": 3922
    },
    {
      "epoch": 0.15862201421250013,
      "grad_norm": 6.301678657531738,
      "learning_rate": 9.417276922098692e-05,
      "loss": 0.2034,
      "step": 3923
    },
    {
      "epoch": 0.15866244806776714,
      "grad_norm": 6.617535591125488,
      "learning_rate": 9.416978099230862e-05,
      "loss": 0.1109,
      "step": 3924
    },
    {
      "epoch": 0.15870288192303417,
      "grad_norm": 6.715869903564453,
      "learning_rate": 9.416679204507386e-05,
      "loss": 0.1647,
      "step": 3925
    },
    {
      "epoch": 0.15874331577830117,
      "grad_norm": 6.666064739227295,
      "learning_rate": 9.416380237933124e-05,
      "loss": 0.2493,
      "step": 3926
    },
    {
      "epoch": 0.1587837496335682,
      "grad_norm": 4.817368030548096,
      "learning_rate": 9.416081199512943e-05,
      "loss": 0.1737,
      "step": 3927
    },
    {
      "epoch": 0.1588241834888352,
      "grad_norm": 3.6487090587615967,
      "learning_rate": 9.415782089251707e-05,
      "loss": 0.0972,
      "step": 3928
    },
    {
      "epoch": 0.1588646173441022,
      "grad_norm": 3.3505754470825195,
      "learning_rate": 9.415482907154279e-05,
      "loss": 0.3049,
      "step": 3929
    },
    {
      "epoch": 0.15890505119936923,
      "grad_norm": 6.6698079109191895,
      "learning_rate": 9.41518365322553e-05,
      "loss": 0.1747,
      "step": 3930
    },
    {
      "epoch": 0.15894548505463624,
      "grad_norm": 3.398442506790161,
      "learning_rate": 9.414884327470326e-05,
      "loss": 0.1403,
      "step": 3931
    },
    {
      "epoch": 0.15898591890990327,
      "grad_norm": 3.3191401958465576,
      "learning_rate": 9.414584929893535e-05,
      "loss": 0.1665,
      "step": 3932
    },
    {
      "epoch": 0.15902635276517027,
      "grad_norm": 1.9655755758285522,
      "learning_rate": 9.414285460500031e-05,
      "loss": 0.1456,
      "step": 3933
    },
    {
      "epoch": 0.1590667866204373,
      "grad_norm": 2.500462532043457,
      "learning_rate": 9.413985919294685e-05,
      "loss": 0.0814,
      "step": 3934
    },
    {
      "epoch": 0.1591072204757043,
      "grad_norm": 4.359124183654785,
      "learning_rate": 9.413686306282369e-05,
      "loss": 0.2849,
      "step": 3935
    },
    {
      "epoch": 0.15914765433097133,
      "grad_norm": 3.9733524322509766,
      "learning_rate": 9.413386621467956e-05,
      "loss": 0.1116,
      "step": 3936
    },
    {
      "epoch": 0.15918808818623834,
      "grad_norm": 3.9780168533325195,
      "learning_rate": 9.413086864856324e-05,
      "loss": 0.1759,
      "step": 3937
    },
    {
      "epoch": 0.15922852204150537,
      "grad_norm": 3.003861427307129,
      "learning_rate": 9.412787036452348e-05,
      "loss": 0.2553,
      "step": 3938
    },
    {
      "epoch": 0.15926895589677237,
      "grad_norm": 5.646566390991211,
      "learning_rate": 9.412487136260906e-05,
      "loss": 0.2333,
      "step": 3939
    },
    {
      "epoch": 0.15930938975203937,
      "grad_norm": 3.712719202041626,
      "learning_rate": 9.412187164286877e-05,
      "loss": 0.1624,
      "step": 3940
    },
    {
      "epoch": 0.1593498236073064,
      "grad_norm": 4.441565036773682,
      "learning_rate": 9.411887120535139e-05,
      "loss": 0.1599,
      "step": 3941
    },
    {
      "epoch": 0.1593902574625734,
      "grad_norm": 6.754016399383545,
      "learning_rate": 9.411587005010576e-05,
      "loss": 0.1902,
      "step": 3942
    },
    {
      "epoch": 0.15943069131784043,
      "grad_norm": 3.3136258125305176,
      "learning_rate": 9.411286817718068e-05,
      "loss": 0.1462,
      "step": 3943
    },
    {
      "epoch": 0.15947112517310744,
      "grad_norm": 4.267288684844971,
      "learning_rate": 9.4109865586625e-05,
      "loss": 0.1259,
      "step": 3944
    },
    {
      "epoch": 0.15951155902837447,
      "grad_norm": 2.2339589595794678,
      "learning_rate": 9.410686227848755e-05,
      "loss": 0.1683,
      "step": 3945
    },
    {
      "epoch": 0.15955199288364147,
      "grad_norm": 4.678582191467285,
      "learning_rate": 9.410385825281721e-05,
      "loss": 0.1182,
      "step": 3946
    },
    {
      "epoch": 0.1595924267389085,
      "grad_norm": 3.4060330390930176,
      "learning_rate": 9.410085350966283e-05,
      "loss": 0.1057,
      "step": 3947
    },
    {
      "epoch": 0.1596328605941755,
      "grad_norm": 3.5238535404205322,
      "learning_rate": 9.409784804907331e-05,
      "loss": 0.1493,
      "step": 3948
    },
    {
      "epoch": 0.1596732944494425,
      "grad_norm": 2.023892879486084,
      "learning_rate": 9.409484187109752e-05,
      "loss": 0.1291,
      "step": 3949
    },
    {
      "epoch": 0.15971372830470953,
      "grad_norm": 9.805617332458496,
      "learning_rate": 9.409183497578437e-05,
      "loss": 0.2501,
      "step": 3950
    },
    {
      "epoch": 0.15975416215997654,
      "grad_norm": 9.09783935546875,
      "learning_rate": 9.40888273631828e-05,
      "loss": 0.1844,
      "step": 3951
    },
    {
      "epoch": 0.15979459601524357,
      "grad_norm": 8.58177375793457,
      "learning_rate": 9.408581903334172e-05,
      "loss": 0.2886,
      "step": 3952
    },
    {
      "epoch": 0.15983502987051057,
      "grad_norm": 7.992732524871826,
      "learning_rate": 9.408280998631008e-05,
      "loss": 0.159,
      "step": 3953
    },
    {
      "epoch": 0.1598754637257776,
      "grad_norm": 3.59867262840271,
      "learning_rate": 9.407980022213681e-05,
      "loss": 0.2331,
      "step": 3954
    },
    {
      "epoch": 0.1599158975810446,
      "grad_norm": 6.043071269989014,
      "learning_rate": 9.407678974087088e-05,
      "loss": 0.1604,
      "step": 3955
    },
    {
      "epoch": 0.15995633143631163,
      "grad_norm": 4.826127052307129,
      "learning_rate": 9.407377854256128e-05,
      "loss": 0.0678,
      "step": 3956
    },
    {
      "epoch": 0.15999676529157864,
      "grad_norm": 8.113791465759277,
      "learning_rate": 9.407076662725698e-05,
      "loss": 0.2644,
      "step": 3957
    },
    {
      "epoch": 0.16003719914684567,
      "grad_norm": 4.114380836486816,
      "learning_rate": 9.406775399500699e-05,
      "loss": 0.0987,
      "step": 3958
    },
    {
      "epoch": 0.16007763300211267,
      "grad_norm": 2.9014453887939453,
      "learning_rate": 9.40647406458603e-05,
      "loss": 0.0665,
      "step": 3959
    },
    {
      "epoch": 0.16011806685737967,
      "grad_norm": 3.109654664993286,
      "learning_rate": 9.406172657986597e-05,
      "loss": 0.1003,
      "step": 3960
    },
    {
      "epoch": 0.1601585007126467,
      "grad_norm": 4.9017767906188965,
      "learning_rate": 9.4058711797073e-05,
      "loss": 0.1737,
      "step": 3961
    },
    {
      "epoch": 0.1601989345679137,
      "grad_norm": 9.06684398651123,
      "learning_rate": 9.405569629753044e-05,
      "loss": 0.2516,
      "step": 3962
    },
    {
      "epoch": 0.16023936842318073,
      "grad_norm": 2.4437639713287354,
      "learning_rate": 9.405268008128733e-05,
      "loss": 0.1036,
      "step": 3963
    },
    {
      "epoch": 0.16027980227844774,
      "grad_norm": 2.747758388519287,
      "learning_rate": 9.404966314839279e-05,
      "loss": 0.084,
      "step": 3964
    },
    {
      "epoch": 0.16032023613371477,
      "grad_norm": 5.936177730560303,
      "learning_rate": 9.404664549889586e-05,
      "loss": 0.199,
      "step": 3965
    },
    {
      "epoch": 0.16036066998898177,
      "grad_norm": 6.738802433013916,
      "learning_rate": 9.404362713284562e-05,
      "loss": 0.2447,
      "step": 3966
    },
    {
      "epoch": 0.1604011038442488,
      "grad_norm": 4.762523174285889,
      "learning_rate": 9.40406080502912e-05,
      "loss": 0.2135,
      "step": 3967
    },
    {
      "epoch": 0.1604415376995158,
      "grad_norm": 5.981261730194092,
      "learning_rate": 9.40375882512817e-05,
      "loss": 0.1332,
      "step": 3968
    },
    {
      "epoch": 0.16048197155478283,
      "grad_norm": 2.85152268409729,
      "learning_rate": 9.403456773586627e-05,
      "loss": 0.1208,
      "step": 3969
    },
    {
      "epoch": 0.16052240541004983,
      "grad_norm": 9.853446006774902,
      "learning_rate": 9.403154650409402e-05,
      "loss": 0.1802,
      "step": 3970
    },
    {
      "epoch": 0.16056283926531684,
      "grad_norm": 7.513638973236084,
      "learning_rate": 9.402852455601412e-05,
      "loss": 0.2508,
      "step": 3971
    },
    {
      "epoch": 0.16060327312058387,
      "grad_norm": 8.227529525756836,
      "learning_rate": 9.40255018916757e-05,
      "loss": 0.2848,
      "step": 3972
    },
    {
      "epoch": 0.16064370697585087,
      "grad_norm": 5.597812175750732,
      "learning_rate": 9.402247851112798e-05,
      "loss": 0.1867,
      "step": 3973
    },
    {
      "epoch": 0.1606841408311179,
      "grad_norm": 6.644929885864258,
      "learning_rate": 9.40194544144201e-05,
      "loss": 0.1455,
      "step": 3974
    },
    {
      "epoch": 0.1607245746863849,
      "grad_norm": 2.707232713699341,
      "learning_rate": 9.401642960160126e-05,
      "loss": 0.0724,
      "step": 3975
    },
    {
      "epoch": 0.16076500854165193,
      "grad_norm": 6.885936260223389,
      "learning_rate": 9.40134040727207e-05,
      "loss": 0.213,
      "step": 3976
    },
    {
      "epoch": 0.16080544239691894,
      "grad_norm": 3.640817642211914,
      "learning_rate": 9.401037782782762e-05,
      "loss": 0.1745,
      "step": 3977
    },
    {
      "epoch": 0.16084587625218597,
      "grad_norm": 3.873887062072754,
      "learning_rate": 9.400735086697126e-05,
      "loss": 0.0423,
      "step": 3978
    },
    {
      "epoch": 0.16088631010745297,
      "grad_norm": 7.203456878662109,
      "learning_rate": 9.400432319020086e-05,
      "loss": 0.1315,
      "step": 3979
    },
    {
      "epoch": 0.16092674396272,
      "grad_norm": 5.037572383880615,
      "learning_rate": 9.400129479756565e-05,
      "loss": 0.0848,
      "step": 3980
    },
    {
      "epoch": 0.160967177817987,
      "grad_norm": 3.399183511734009,
      "learning_rate": 9.399826568911493e-05,
      "loss": 0.1483,
      "step": 3981
    },
    {
      "epoch": 0.161007611673254,
      "grad_norm": 4.448543071746826,
      "learning_rate": 9.399523586489796e-05,
      "loss": 0.1902,
      "step": 3982
    },
    {
      "epoch": 0.16104804552852103,
      "grad_norm": 6.4281487464904785,
      "learning_rate": 9.399220532496403e-05,
      "loss": 0.1032,
      "step": 3983
    },
    {
      "epoch": 0.16108847938378804,
      "grad_norm": 3.7224040031433105,
      "learning_rate": 9.398917406936245e-05,
      "loss": 0.105,
      "step": 3984
    },
    {
      "epoch": 0.16112891323905507,
      "grad_norm": 2.5578622817993164,
      "learning_rate": 9.398614209814252e-05,
      "loss": 0.0962,
      "step": 3985
    },
    {
      "epoch": 0.16116934709432207,
      "grad_norm": 6.7964019775390625,
      "learning_rate": 9.398310941135358e-05,
      "loss": 0.2603,
      "step": 3986
    },
    {
      "epoch": 0.1612097809495891,
      "grad_norm": 9.528360366821289,
      "learning_rate": 9.398007600904494e-05,
      "loss": 0.3534,
      "step": 3987
    },
    {
      "epoch": 0.1612502148048561,
      "grad_norm": 1.8921895027160645,
      "learning_rate": 9.397704189126598e-05,
      "loss": 0.0907,
      "step": 3988
    },
    {
      "epoch": 0.16129064866012313,
      "grad_norm": 9.881204605102539,
      "learning_rate": 9.397400705806604e-05,
      "loss": 0.3857,
      "step": 3989
    },
    {
      "epoch": 0.16133108251539013,
      "grad_norm": 6.119568347930908,
      "learning_rate": 9.39709715094945e-05,
      "loss": 0.271,
      "step": 3990
    },
    {
      "epoch": 0.16137151637065716,
      "grad_norm": 2.189521312713623,
      "learning_rate": 9.396793524560072e-05,
      "loss": 0.1055,
      "step": 3991
    },
    {
      "epoch": 0.16141195022592417,
      "grad_norm": 4.093836784362793,
      "learning_rate": 9.396489826643413e-05,
      "loss": 0.1046,
      "step": 3992
    },
    {
      "epoch": 0.16145238408119117,
      "grad_norm": 9.893691062927246,
      "learning_rate": 9.39618605720441e-05,
      "loss": 0.4277,
      "step": 3993
    },
    {
      "epoch": 0.1614928179364582,
      "grad_norm": 7.19428825378418,
      "learning_rate": 9.395882216248007e-05,
      "loss": 0.2858,
      "step": 3994
    },
    {
      "epoch": 0.1615332517917252,
      "grad_norm": 4.85747766494751,
      "learning_rate": 9.395578303779147e-05,
      "loss": 0.2401,
      "step": 3995
    },
    {
      "epoch": 0.16157368564699223,
      "grad_norm": 7.892520904541016,
      "learning_rate": 9.395274319802773e-05,
      "loss": 0.2245,
      "step": 3996
    },
    {
      "epoch": 0.16161411950225923,
      "grad_norm": 5.816650867462158,
      "learning_rate": 9.394970264323832e-05,
      "loss": 0.2928,
      "step": 3997
    },
    {
      "epoch": 0.16165455335752626,
      "grad_norm": 2.2702598571777344,
      "learning_rate": 9.394666137347267e-05,
      "loss": 0.1027,
      "step": 3998
    },
    {
      "epoch": 0.16169498721279327,
      "grad_norm": 5.900112628936768,
      "learning_rate": 9.394361938878027e-05,
      "loss": 0.19,
      "step": 3999
    },
    {
      "epoch": 0.1617354210680603,
      "grad_norm": 5.05006742477417,
      "learning_rate": 9.394057668921064e-05,
      "loss": 0.126,
      "step": 4000
    },
    {
      "epoch": 0.1617758549233273,
      "grad_norm": 3.264193058013916,
      "learning_rate": 9.393753327481323e-05,
      "loss": 0.1345,
      "step": 4001
    },
    {
      "epoch": 0.16181628877859433,
      "grad_norm": 2.7817890644073486,
      "learning_rate": 9.393448914563758e-05,
      "loss": 0.1762,
      "step": 4002
    },
    {
      "epoch": 0.16185672263386133,
      "grad_norm": 6.51584005355835,
      "learning_rate": 9.39314443017332e-05,
      "loss": 0.1589,
      "step": 4003
    },
    {
      "epoch": 0.16189715648912834,
      "grad_norm": 3.6729543209075928,
      "learning_rate": 9.392839874314964e-05,
      "loss": 0.1487,
      "step": 4004
    },
    {
      "epoch": 0.16193759034439537,
      "grad_norm": 11.196158409118652,
      "learning_rate": 9.392535246993642e-05,
      "loss": 0.2813,
      "step": 4005
    },
    {
      "epoch": 0.16197802419966237,
      "grad_norm": 2.1161673069000244,
      "learning_rate": 9.392230548214311e-05,
      "loss": 0.0776,
      "step": 4006
    },
    {
      "epoch": 0.1620184580549294,
      "grad_norm": 6.121769428253174,
      "learning_rate": 9.39192577798193e-05,
      "loss": 0.1524,
      "step": 4007
    },
    {
      "epoch": 0.1620588919101964,
      "grad_norm": 4.022584438323975,
      "learning_rate": 9.391620936301451e-05,
      "loss": 0.0862,
      "step": 4008
    },
    {
      "epoch": 0.16209932576546343,
      "grad_norm": 7.750242710113525,
      "learning_rate": 9.391316023177841e-05,
      "loss": 0.3122,
      "step": 4009
    },
    {
      "epoch": 0.16213975962073043,
      "grad_norm": 6.883797645568848,
      "learning_rate": 9.391011038616053e-05,
      "loss": 0.2554,
      "step": 4010
    },
    {
      "epoch": 0.16218019347599746,
      "grad_norm": 1.9359537363052368,
      "learning_rate": 9.390705982621055e-05,
      "loss": 0.0921,
      "step": 4011
    },
    {
      "epoch": 0.16222062733126447,
      "grad_norm": 6.423151969909668,
      "learning_rate": 9.390400855197806e-05,
      "loss": 0.2218,
      "step": 4012
    },
    {
      "epoch": 0.1622610611865315,
      "grad_norm": 2.8673722743988037,
      "learning_rate": 9.390095656351268e-05,
      "loss": 0.1214,
      "step": 4013
    },
    {
      "epoch": 0.1623014950417985,
      "grad_norm": 6.8109660148620605,
      "learning_rate": 9.38979038608641e-05,
      "loss": 0.3266,
      "step": 4014
    },
    {
      "epoch": 0.1623419288970655,
      "grad_norm": 4.855719089508057,
      "learning_rate": 9.389485044408197e-05,
      "loss": 0.1654,
      "step": 4015
    },
    {
      "epoch": 0.16238236275233253,
      "grad_norm": 9.902756690979004,
      "learning_rate": 9.389179631321596e-05,
      "loss": 0.2305,
      "step": 4016
    },
    {
      "epoch": 0.16242279660759953,
      "grad_norm": 1.3652812242507935,
      "learning_rate": 9.388874146831575e-05,
      "loss": 0.0408,
      "step": 4017
    },
    {
      "epoch": 0.16246323046286656,
      "grad_norm": 9.730671882629395,
      "learning_rate": 9.388568590943103e-05,
      "loss": 0.2755,
      "step": 4018
    },
    {
      "epoch": 0.16250366431813357,
      "grad_norm": 6.552483558654785,
      "learning_rate": 9.388262963661152e-05,
      "loss": 0.1897,
      "step": 4019
    },
    {
      "epoch": 0.1625440981734006,
      "grad_norm": 2.495316505432129,
      "learning_rate": 9.387957264990695e-05,
      "loss": 0.091,
      "step": 4020
    },
    {
      "epoch": 0.1625845320286676,
      "grad_norm": 11.022533416748047,
      "learning_rate": 9.387651494936702e-05,
      "loss": 0.2907,
      "step": 4021
    },
    {
      "epoch": 0.16262496588393463,
      "grad_norm": 3.0573341846466064,
      "learning_rate": 9.387345653504152e-05,
      "loss": 0.1683,
      "step": 4022
    },
    {
      "epoch": 0.16266539973920163,
      "grad_norm": 6.264918804168701,
      "learning_rate": 9.387039740698015e-05,
      "loss": 0.209,
      "step": 4023
    },
    {
      "epoch": 0.16270583359446864,
      "grad_norm": 8.521798133850098,
      "learning_rate": 9.386733756523271e-05,
      "loss": 0.2492,
      "step": 4024
    },
    {
      "epoch": 0.16274626744973567,
      "grad_norm": 3.8069207668304443,
      "learning_rate": 9.386427700984897e-05,
      "loss": 0.1227,
      "step": 4025
    },
    {
      "epoch": 0.16278670130500267,
      "grad_norm": 8.90477466583252,
      "learning_rate": 9.386121574087872e-05,
      "loss": 0.2574,
      "step": 4026
    },
    {
      "epoch": 0.1628271351602697,
      "grad_norm": 5.040698528289795,
      "learning_rate": 9.385815375837177e-05,
      "loss": 0.2057,
      "step": 4027
    },
    {
      "epoch": 0.1628675690155367,
      "grad_norm": 3.8400304317474365,
      "learning_rate": 9.385509106237791e-05,
      "loss": 0.2229,
      "step": 4028
    },
    {
      "epoch": 0.16290800287080373,
      "grad_norm": 3.3539371490478516,
      "learning_rate": 9.385202765294698e-05,
      "loss": 0.1971,
      "step": 4029
    },
    {
      "epoch": 0.16294843672607073,
      "grad_norm": 8.180174827575684,
      "learning_rate": 9.384896353012881e-05,
      "loss": 0.1903,
      "step": 4030
    },
    {
      "epoch": 0.16298887058133776,
      "grad_norm": 3.6797358989715576,
      "learning_rate": 9.384589869397325e-05,
      "loss": 0.18,
      "step": 4031
    },
    {
      "epoch": 0.16302930443660477,
      "grad_norm": 5.707553386688232,
      "learning_rate": 9.384283314453015e-05,
      "loss": 0.2076,
      "step": 4032
    },
    {
      "epoch": 0.1630697382918718,
      "grad_norm": 4.814053058624268,
      "learning_rate": 9.383976688184941e-05,
      "loss": 0.167,
      "step": 4033
    },
    {
      "epoch": 0.1631101721471388,
      "grad_norm": 6.259121894836426,
      "learning_rate": 9.383669990598087e-05,
      "loss": 0.165,
      "step": 4034
    },
    {
      "epoch": 0.1631506060024058,
      "grad_norm": 3.8790555000305176,
      "learning_rate": 9.383363221697446e-05,
      "loss": 0.1692,
      "step": 4035
    },
    {
      "epoch": 0.16319103985767283,
      "grad_norm": 2.8552632331848145,
      "learning_rate": 9.383056381488007e-05,
      "loss": 0.1542,
      "step": 4036
    },
    {
      "epoch": 0.16323147371293983,
      "grad_norm": 2.435856580734253,
      "learning_rate": 9.38274946997476e-05,
      "loss": 0.1439,
      "step": 4037
    },
    {
      "epoch": 0.16327190756820686,
      "grad_norm": 2.4025697708129883,
      "learning_rate": 9.382442487162702e-05,
      "loss": 0.1699,
      "step": 4038
    },
    {
      "epoch": 0.16331234142347387,
      "grad_norm": 4.372707366943359,
      "learning_rate": 9.382135433056825e-05,
      "loss": 0.1468,
      "step": 4039
    },
    {
      "epoch": 0.1633527752787409,
      "grad_norm": 7.44069766998291,
      "learning_rate": 9.381828307662121e-05,
      "loss": 0.1705,
      "step": 4040
    },
    {
      "epoch": 0.1633932091340079,
      "grad_norm": 3.7876334190368652,
      "learning_rate": 9.381521110983591e-05,
      "loss": 0.2368,
      "step": 4041
    },
    {
      "epoch": 0.16343364298927493,
      "grad_norm": 3.353705644607544,
      "learning_rate": 9.381213843026229e-05,
      "loss": 0.1495,
      "step": 4042
    },
    {
      "epoch": 0.16347407684454193,
      "grad_norm": 3.619739294052124,
      "learning_rate": 9.380906503795038e-05,
      "loss": 0.2158,
      "step": 4043
    },
    {
      "epoch": 0.16351451069980896,
      "grad_norm": 4.689696311950684,
      "learning_rate": 9.380599093295012e-05,
      "loss": 0.1546,
      "step": 4044
    },
    {
      "epoch": 0.16355494455507597,
      "grad_norm": 3.950038194656372,
      "learning_rate": 9.380291611531156e-05,
      "loss": 0.2187,
      "step": 4045
    },
    {
      "epoch": 0.16359537841034297,
      "grad_norm": 5.583806037902832,
      "learning_rate": 9.379984058508471e-05,
      "loss": 0.1911,
      "step": 4046
    },
    {
      "epoch": 0.16363581226561,
      "grad_norm": 2.362182140350342,
      "learning_rate": 9.379676434231962e-05,
      "loss": 0.0546,
      "step": 4047
    },
    {
      "epoch": 0.163676246120877,
      "grad_norm": 4.095893859863281,
      "learning_rate": 9.37936873870663e-05,
      "loss": 0.2995,
      "step": 4048
    },
    {
      "epoch": 0.16371667997614403,
      "grad_norm": 3.7327582836151123,
      "learning_rate": 9.379060971937484e-05,
      "loss": 0.1622,
      "step": 4049
    },
    {
      "epoch": 0.16375711383141103,
      "grad_norm": 5.634705543518066,
      "learning_rate": 9.378753133929528e-05,
      "loss": 0.1089,
      "step": 4050
    },
    {
      "epoch": 0.16379754768667806,
      "grad_norm": 4.705047130584717,
      "learning_rate": 9.37844522468777e-05,
      "loss": 0.2469,
      "step": 4051
    },
    {
      "epoch": 0.16383798154194507,
      "grad_norm": 10.688558578491211,
      "learning_rate": 9.378137244217224e-05,
      "loss": 0.2271,
      "step": 4052
    },
    {
      "epoch": 0.1638784153972121,
      "grad_norm": 3.337273359298706,
      "learning_rate": 9.377829192522894e-05,
      "loss": 0.0931,
      "step": 4053
    },
    {
      "epoch": 0.1639188492524791,
      "grad_norm": 5.6125874519348145,
      "learning_rate": 9.377521069609794e-05,
      "loss": 0.1082,
      "step": 4054
    },
    {
      "epoch": 0.16395928310774613,
      "grad_norm": 7.784252643585205,
      "learning_rate": 9.377212875482937e-05,
      "loss": 0.1467,
      "step": 4055
    },
    {
      "epoch": 0.16399971696301313,
      "grad_norm": 7.106649875640869,
      "learning_rate": 9.376904610147336e-05,
      "loss": 0.2012,
      "step": 4056
    },
    {
      "epoch": 0.16404015081828013,
      "grad_norm": 7.788704872131348,
      "learning_rate": 9.376596273608007e-05,
      "loss": 0.2871,
      "step": 4057
    },
    {
      "epoch": 0.16408058467354716,
      "grad_norm": 5.916694641113281,
      "learning_rate": 9.376287865869963e-05,
      "loss": 0.2346,
      "step": 4058
    },
    {
      "epoch": 0.16412101852881417,
      "grad_norm": 8.366349220275879,
      "learning_rate": 9.375979386938225e-05,
      "loss": 0.1654,
      "step": 4059
    },
    {
      "epoch": 0.1641614523840812,
      "grad_norm": 10.842005729675293,
      "learning_rate": 9.37567083681781e-05,
      "loss": 0.3457,
      "step": 4060
    },
    {
      "epoch": 0.1642018862393482,
      "grad_norm": 7.831226825714111,
      "learning_rate": 9.375362215513737e-05,
      "loss": 0.1092,
      "step": 4061
    },
    {
      "epoch": 0.16424232009461523,
      "grad_norm": 6.945502758026123,
      "learning_rate": 9.375053523031026e-05,
      "loss": 0.2162,
      "step": 4062
    },
    {
      "epoch": 0.16428275394988223,
      "grad_norm": 6.939932346343994,
      "learning_rate": 9.3747447593747e-05,
      "loss": 0.1898,
      "step": 4063
    },
    {
      "epoch": 0.16432318780514926,
      "grad_norm": 2.9538354873657227,
      "learning_rate": 9.374435924549781e-05,
      "loss": 0.1944,
      "step": 4064
    },
    {
      "epoch": 0.16436362166041626,
      "grad_norm": 4.205642223358154,
      "learning_rate": 9.374127018561295e-05,
      "loss": 0.0533,
      "step": 4065
    },
    {
      "epoch": 0.1644040555156833,
      "grad_norm": 12.222182273864746,
      "learning_rate": 9.373818041414267e-05,
      "loss": 0.3408,
      "step": 4066
    },
    {
      "epoch": 0.1644444893709503,
      "grad_norm": 4.1083526611328125,
      "learning_rate": 9.37350899311372e-05,
      "loss": 0.0683,
      "step": 4067
    },
    {
      "epoch": 0.1644849232262173,
      "grad_norm": 7.7268805503845215,
      "learning_rate": 9.373199873664684e-05,
      "loss": 0.418,
      "step": 4068
    },
    {
      "epoch": 0.16452535708148433,
      "grad_norm": 10.906867027282715,
      "learning_rate": 9.37289068307219e-05,
      "loss": 0.3153,
      "step": 4069
    },
    {
      "epoch": 0.16456579093675133,
      "grad_norm": 5.773379802703857,
      "learning_rate": 9.372581421341265e-05,
      "loss": 0.2242,
      "step": 4070
    },
    {
      "epoch": 0.16460622479201836,
      "grad_norm": 2.4142744541168213,
      "learning_rate": 9.372272088476941e-05,
      "loss": 0.1456,
      "step": 4071
    },
    {
      "epoch": 0.16464665864728537,
      "grad_norm": 4.618976593017578,
      "learning_rate": 9.37196268448425e-05,
      "loss": 0.0625,
      "step": 4072
    },
    {
      "epoch": 0.1646870925025524,
      "grad_norm": 9.237908363342285,
      "learning_rate": 9.371653209368226e-05,
      "loss": 0.3241,
      "step": 4073
    },
    {
      "epoch": 0.1647275263578194,
      "grad_norm": 3.1524558067321777,
      "learning_rate": 9.371343663133904e-05,
      "loss": 0.1845,
      "step": 4074
    },
    {
      "epoch": 0.16476796021308643,
      "grad_norm": 4.888955116271973,
      "learning_rate": 9.371034045786318e-05,
      "loss": 0.1595,
      "step": 4075
    },
    {
      "epoch": 0.16480839406835343,
      "grad_norm": 3.4094254970550537,
      "learning_rate": 9.370724357330504e-05,
      "loss": 0.1827,
      "step": 4076
    },
    {
      "epoch": 0.16484882792362046,
      "grad_norm": 5.033443450927734,
      "learning_rate": 9.370414597771505e-05,
      "loss": 0.1823,
      "step": 4077
    },
    {
      "epoch": 0.16488926177888746,
      "grad_norm": 3.9670279026031494,
      "learning_rate": 9.370104767114357e-05,
      "loss": 0.1097,
      "step": 4078
    },
    {
      "epoch": 0.16492969563415447,
      "grad_norm": 3.107948064804077,
      "learning_rate": 9.369794865364096e-05,
      "loss": 0.1002,
      "step": 4079
    },
    {
      "epoch": 0.1649701294894215,
      "grad_norm": 7.0311408042907715,
      "learning_rate": 9.369484892525771e-05,
      "loss": 0.1588,
      "step": 4080
    },
    {
      "epoch": 0.1650105633446885,
      "grad_norm": 5.641593933105469,
      "learning_rate": 9.369174848604422e-05,
      "loss": 0.2112,
      "step": 4081
    },
    {
      "epoch": 0.16505099719995553,
      "grad_norm": 6.195979118347168,
      "learning_rate": 9.368864733605092e-05,
      "loss": 0.3547,
      "step": 4082
    },
    {
      "epoch": 0.16509143105522253,
      "grad_norm": 5.260648250579834,
      "learning_rate": 9.368554547532824e-05,
      "loss": 0.1921,
      "step": 4083
    },
    {
      "epoch": 0.16513186491048956,
      "grad_norm": 2.259248733520508,
      "learning_rate": 9.368244290392668e-05,
      "loss": 0.1279,
      "step": 4084
    },
    {
      "epoch": 0.16517229876575656,
      "grad_norm": 9.39963150024414,
      "learning_rate": 9.36793396218967e-05,
      "loss": 0.2336,
      "step": 4085
    },
    {
      "epoch": 0.1652127326210236,
      "grad_norm": 6.7742486000061035,
      "learning_rate": 9.367623562928877e-05,
      "loss": 0.1521,
      "step": 4086
    },
    {
      "epoch": 0.1652531664762906,
      "grad_norm": 3.194586753845215,
      "learning_rate": 9.36731309261534e-05,
      "loss": 0.1732,
      "step": 4087
    },
    {
      "epoch": 0.1652936003315576,
      "grad_norm": 1.893242597579956,
      "learning_rate": 9.367002551254111e-05,
      "loss": 0.0686,
      "step": 4088
    },
    {
      "epoch": 0.16533403418682463,
      "grad_norm": 6.9081950187683105,
      "learning_rate": 9.366691938850239e-05,
      "loss": 0.156,
      "step": 4089
    },
    {
      "epoch": 0.16537446804209163,
      "grad_norm": 7.37139368057251,
      "learning_rate": 9.366381255408778e-05,
      "loss": 0.2242,
      "step": 4090
    },
    {
      "epoch": 0.16541490189735866,
      "grad_norm": 3.9485387802124023,
      "learning_rate": 9.366070500934782e-05,
      "loss": 0.2549,
      "step": 4091
    },
    {
      "epoch": 0.16545533575262567,
      "grad_norm": 4.5765204429626465,
      "learning_rate": 9.36575967543331e-05,
      "loss": 0.1136,
      "step": 4092
    },
    {
      "epoch": 0.1654957696078927,
      "grad_norm": 3.419405221939087,
      "learning_rate": 9.365448778909413e-05,
      "loss": 0.0928,
      "step": 4093
    },
    {
      "epoch": 0.1655362034631597,
      "grad_norm": 7.081779956817627,
      "learning_rate": 9.365137811368152e-05,
      "loss": 0.1806,
      "step": 4094
    },
    {
      "epoch": 0.16557663731842673,
      "grad_norm": 3.6009933948516846,
      "learning_rate": 9.364826772814584e-05,
      "loss": 0.123,
      "step": 4095
    },
    {
      "epoch": 0.16561707117369373,
      "grad_norm": 3.3587167263031006,
      "learning_rate": 9.364515663253773e-05,
      "loss": 0.06,
      "step": 4096
    },
    {
      "epoch": 0.16565750502896076,
      "grad_norm": 5.469005107879639,
      "learning_rate": 9.364204482690774e-05,
      "loss": 0.1189,
      "step": 4097
    },
    {
      "epoch": 0.16569793888422776,
      "grad_norm": 4.061158657073975,
      "learning_rate": 9.363893231130653e-05,
      "loss": 0.0808,
      "step": 4098
    },
    {
      "epoch": 0.16573837273949477,
      "grad_norm": 2.441441774368286,
      "learning_rate": 9.363581908578473e-05,
      "loss": 0.118,
      "step": 4099
    },
    {
      "epoch": 0.1657788065947618,
      "grad_norm": 12.77608871459961,
      "learning_rate": 9.363270515039299e-05,
      "loss": 0.3718,
      "step": 4100
    },
    {
      "epoch": 0.1658192404500288,
      "grad_norm": 5.2358078956604,
      "learning_rate": 9.362959050518196e-05,
      "loss": 0.2221,
      "step": 4101
    },
    {
      "epoch": 0.16585967430529583,
      "grad_norm": 3.956847667694092,
      "learning_rate": 9.362647515020232e-05,
      "loss": 0.2585,
      "step": 4102
    },
    {
      "epoch": 0.16590010816056283,
      "grad_norm": 5.41862678527832,
      "learning_rate": 9.362335908550473e-05,
      "loss": 0.1707,
      "step": 4103
    },
    {
      "epoch": 0.16594054201582986,
      "grad_norm": 4.865588665008545,
      "learning_rate": 9.36202423111399e-05,
      "loss": 0.2932,
      "step": 4104
    },
    {
      "epoch": 0.16598097587109686,
      "grad_norm": 4.630239963531494,
      "learning_rate": 9.361712482715855e-05,
      "loss": 0.1314,
      "step": 4105
    },
    {
      "epoch": 0.1660214097263639,
      "grad_norm": 7.275437355041504,
      "learning_rate": 9.361400663361135e-05,
      "loss": 0.2438,
      "step": 4106
    },
    {
      "epoch": 0.1660618435816309,
      "grad_norm": 7.846410751342773,
      "learning_rate": 9.361088773054906e-05,
      "loss": 0.216,
      "step": 4107
    },
    {
      "epoch": 0.16610227743689793,
      "grad_norm": 4.453481674194336,
      "learning_rate": 9.360776811802241e-05,
      "loss": 0.1881,
      "step": 4108
    },
    {
      "epoch": 0.16614271129216493,
      "grad_norm": 3.978501081466675,
      "learning_rate": 9.360464779608214e-05,
      "loss": 0.1509,
      "step": 4109
    },
    {
      "epoch": 0.16618314514743193,
      "grad_norm": 4.650883197784424,
      "learning_rate": 9.360152676477904e-05,
      "loss": 0.1739,
      "step": 4110
    },
    {
      "epoch": 0.16622357900269896,
      "grad_norm": 10.382174491882324,
      "learning_rate": 9.359840502416385e-05,
      "loss": 0.2377,
      "step": 4111
    },
    {
      "epoch": 0.16626401285796596,
      "grad_norm": 9.206741333007812,
      "learning_rate": 9.359528257428737e-05,
      "loss": 0.3585,
      "step": 4112
    },
    {
      "epoch": 0.166304446713233,
      "grad_norm": 4.562549591064453,
      "learning_rate": 9.35921594152004e-05,
      "loss": 0.2039,
      "step": 4113
    },
    {
      "epoch": 0.1663448805685,
      "grad_norm": 3.503042221069336,
      "learning_rate": 9.358903554695376e-05,
      "loss": 0.1018,
      "step": 4114
    },
    {
      "epoch": 0.16638531442376703,
      "grad_norm": 5.606146812438965,
      "learning_rate": 9.358591096959822e-05,
      "loss": 0.1983,
      "step": 4115
    },
    {
      "epoch": 0.16642574827903403,
      "grad_norm": 2.617079496383667,
      "learning_rate": 9.358278568318466e-05,
      "loss": 0.0947,
      "step": 4116
    },
    {
      "epoch": 0.16646618213430106,
      "grad_norm": 9.674681663513184,
      "learning_rate": 9.35796596877639e-05,
      "loss": 0.3131,
      "step": 4117
    },
    {
      "epoch": 0.16650661598956806,
      "grad_norm": 4.74222469329834,
      "learning_rate": 9.357653298338681e-05,
      "loss": 0.2355,
      "step": 4118
    },
    {
      "epoch": 0.1665470498448351,
      "grad_norm": 16.32895851135254,
      "learning_rate": 9.357340557010423e-05,
      "loss": 0.376,
      "step": 4119
    },
    {
      "epoch": 0.1665874837001021,
      "grad_norm": 20.4910831451416,
      "learning_rate": 9.357027744796707e-05,
      "loss": 0.5617,
      "step": 4120
    },
    {
      "epoch": 0.1666279175553691,
      "grad_norm": 7.677221775054932,
      "learning_rate": 9.35671486170262e-05,
      "loss": 0.1763,
      "step": 4121
    },
    {
      "epoch": 0.16666835141063613,
      "grad_norm": 3.6269924640655518,
      "learning_rate": 9.35640190773325e-05,
      "loss": 0.2018,
      "step": 4122
    },
    {
      "epoch": 0.16670878526590313,
      "grad_norm": 3.6009154319763184,
      "learning_rate": 9.356088882893692e-05,
      "loss": 0.117,
      "step": 4123
    },
    {
      "epoch": 0.16674921912117016,
      "grad_norm": 20.679492950439453,
      "learning_rate": 9.355775787189037e-05,
      "loss": 0.4606,
      "step": 4124
    },
    {
      "epoch": 0.16678965297643716,
      "grad_norm": 1.9282439947128296,
      "learning_rate": 9.355462620624375e-05,
      "loss": 0.1441,
      "step": 4125
    },
    {
      "epoch": 0.1668300868317042,
      "grad_norm": 6.748626708984375,
      "learning_rate": 9.355149383204806e-05,
      "loss": 0.2288,
      "step": 4126
    },
    {
      "epoch": 0.1668705206869712,
      "grad_norm": 4.915131092071533,
      "learning_rate": 9.354836074935422e-05,
      "loss": 0.2762,
      "step": 4127
    },
    {
      "epoch": 0.16691095454223823,
      "grad_norm": 3.272979497909546,
      "learning_rate": 9.354522695821322e-05,
      "loss": 0.1707,
      "step": 4128
    },
    {
      "epoch": 0.16695138839750523,
      "grad_norm": 7.285824298858643,
      "learning_rate": 9.354209245867604e-05,
      "loss": 0.1899,
      "step": 4129
    },
    {
      "epoch": 0.16699182225277226,
      "grad_norm": 3.948040008544922,
      "learning_rate": 9.353895725079367e-05,
      "loss": 0.1259,
      "step": 4130
    },
    {
      "epoch": 0.16703225610803926,
      "grad_norm": 6.543684482574463,
      "learning_rate": 9.353582133461709e-05,
      "loss": 0.1377,
      "step": 4131
    },
    {
      "epoch": 0.16707268996330626,
      "grad_norm": 3.8433380126953125,
      "learning_rate": 9.353268471019734e-05,
      "loss": 0.1697,
      "step": 4132
    },
    {
      "epoch": 0.1671131238185733,
      "grad_norm": 7.369483470916748,
      "learning_rate": 9.352954737758544e-05,
      "loss": 0.2108,
      "step": 4133
    },
    {
      "epoch": 0.1671535576738403,
      "grad_norm": 3.9952173233032227,
      "learning_rate": 9.352640933683243e-05,
      "loss": 0.2414,
      "step": 4134
    },
    {
      "epoch": 0.16719399152910733,
      "grad_norm": 3.5957727432250977,
      "learning_rate": 9.352327058798937e-05,
      "loss": 0.1907,
      "step": 4135
    },
    {
      "epoch": 0.16723442538437433,
      "grad_norm": 4.5001325607299805,
      "learning_rate": 9.35201311311073e-05,
      "loss": 0.2342,
      "step": 4136
    },
    {
      "epoch": 0.16727485923964136,
      "grad_norm": 3.152419328689575,
      "learning_rate": 9.351699096623732e-05,
      "loss": 0.1889,
      "step": 4137
    },
    {
      "epoch": 0.16731529309490836,
      "grad_norm": 5.859038352966309,
      "learning_rate": 9.351385009343047e-05,
      "loss": 0.2065,
      "step": 4138
    },
    {
      "epoch": 0.1673557269501754,
      "grad_norm": 3.3203799724578857,
      "learning_rate": 9.351070851273789e-05,
      "loss": 0.1082,
      "step": 4139
    },
    {
      "epoch": 0.1673961608054424,
      "grad_norm": 7.509016990661621,
      "learning_rate": 9.350756622421068e-05,
      "loss": 0.2672,
      "step": 4140
    },
    {
      "epoch": 0.16743659466070943,
      "grad_norm": 5.910201549530029,
      "learning_rate": 9.350442322789993e-05,
      "loss": 0.212,
      "step": 4141
    },
    {
      "epoch": 0.16747702851597643,
      "grad_norm": 7.004674911499023,
      "learning_rate": 9.35012795238568e-05,
      "loss": 0.1727,
      "step": 4142
    },
    {
      "epoch": 0.16751746237124343,
      "grad_norm": 3.541886568069458,
      "learning_rate": 9.349813511213242e-05,
      "loss": 0.1606,
      "step": 4143
    },
    {
      "epoch": 0.16755789622651046,
      "grad_norm": 2.2568612098693848,
      "learning_rate": 9.349498999277795e-05,
      "loss": 0.1309,
      "step": 4144
    },
    {
      "epoch": 0.16759833008177746,
      "grad_norm": 3.2968485355377197,
      "learning_rate": 9.349184416584456e-05,
      "loss": 0.146,
      "step": 4145
    },
    {
      "epoch": 0.1676387639370445,
      "grad_norm": 5.28663969039917,
      "learning_rate": 9.348869763138339e-05,
      "loss": 0.1516,
      "step": 4146
    },
    {
      "epoch": 0.1676791977923115,
      "grad_norm": 5.238011837005615,
      "learning_rate": 9.348555038944569e-05,
      "loss": 0.1859,
      "step": 4147
    },
    {
      "epoch": 0.16771963164757853,
      "grad_norm": 4.443617820739746,
      "learning_rate": 9.34824024400826e-05,
      "loss": 0.3229,
      "step": 4148
    },
    {
      "epoch": 0.16776006550284553,
      "grad_norm": 3.4060475826263428,
      "learning_rate": 9.347925378334536e-05,
      "loss": 0.1185,
      "step": 4149
    },
    {
      "epoch": 0.16780049935811256,
      "grad_norm": 3.25897216796875,
      "learning_rate": 9.347610441928519e-05,
      "loss": 0.1061,
      "step": 4150
    },
    {
      "epoch": 0.16784093321337956,
      "grad_norm": 8.36097526550293,
      "learning_rate": 9.347295434795334e-05,
      "loss": 0.2167,
      "step": 4151
    },
    {
      "epoch": 0.16788136706864656,
      "grad_norm": 7.413125514984131,
      "learning_rate": 9.3469803569401e-05,
      "loss": 0.28,
      "step": 4152
    },
    {
      "epoch": 0.1679218009239136,
      "grad_norm": 6.540614604949951,
      "learning_rate": 9.34666520836795e-05,
      "loss": 0.2012,
      "step": 4153
    },
    {
      "epoch": 0.1679622347791806,
      "grad_norm": 8.881750106811523,
      "learning_rate": 9.346349989084006e-05,
      "loss": 0.2121,
      "step": 4154
    },
    {
      "epoch": 0.16800266863444763,
      "grad_norm": 7.008378028869629,
      "learning_rate": 9.346034699093398e-05,
      "loss": 0.152,
      "step": 4155
    },
    {
      "epoch": 0.16804310248971463,
      "grad_norm": 3.496488571166992,
      "learning_rate": 9.345719338401254e-05,
      "loss": 0.1894,
      "step": 4156
    },
    {
      "epoch": 0.16808353634498166,
      "grad_norm": 4.120840549468994,
      "learning_rate": 9.345403907012705e-05,
      "loss": 0.1599,
      "step": 4157
    },
    {
      "epoch": 0.16812397020024866,
      "grad_norm": 5.307690143585205,
      "learning_rate": 9.345088404932881e-05,
      "loss": 0.3111,
      "step": 4158
    },
    {
      "epoch": 0.1681644040555157,
      "grad_norm": 7.748332977294922,
      "learning_rate": 9.344772832166917e-05,
      "loss": 0.2654,
      "step": 4159
    },
    {
      "epoch": 0.1682048379107827,
      "grad_norm": 5.4857048988342285,
      "learning_rate": 9.344457188719946e-05,
      "loss": 0.251,
      "step": 4160
    },
    {
      "epoch": 0.16824527176604973,
      "grad_norm": 7.963584899902344,
      "learning_rate": 9.344141474597101e-05,
      "loss": 0.2709,
      "step": 4161
    },
    {
      "epoch": 0.16828570562131673,
      "grad_norm": 3.4513471126556396,
      "learning_rate": 9.343825689803522e-05,
      "loss": 0.1515,
      "step": 4162
    },
    {
      "epoch": 0.16832613947658373,
      "grad_norm": 3.2975668907165527,
      "learning_rate": 9.343509834344342e-05,
      "loss": 0.1552,
      "step": 4163
    },
    {
      "epoch": 0.16836657333185076,
      "grad_norm": 6.655523777008057,
      "learning_rate": 9.343193908224702e-05,
      "loss": 0.1847,
      "step": 4164
    },
    {
      "epoch": 0.16840700718711776,
      "grad_norm": 6.628211975097656,
      "learning_rate": 9.34287791144974e-05,
      "loss": 0.2379,
      "step": 4165
    },
    {
      "epoch": 0.1684474410423848,
      "grad_norm": 3.2387912273406982,
      "learning_rate": 9.342561844024598e-05,
      "loss": 0.21,
      "step": 4166
    },
    {
      "epoch": 0.1684878748976518,
      "grad_norm": 7.962008476257324,
      "learning_rate": 9.342245705954417e-05,
      "loss": 0.1796,
      "step": 4167
    },
    {
      "epoch": 0.16852830875291883,
      "grad_norm": 5.182526111602783,
      "learning_rate": 9.34192949724434e-05,
      "loss": 0.3087,
      "step": 4168
    },
    {
      "epoch": 0.16856874260818583,
      "grad_norm": 3.2711098194122314,
      "learning_rate": 9.341613217899511e-05,
      "loss": 0.0998,
      "step": 4169
    },
    {
      "epoch": 0.16860917646345286,
      "grad_norm": 4.778079509735107,
      "learning_rate": 9.341296867925075e-05,
      "loss": 0.2159,
      "step": 4170
    },
    {
      "epoch": 0.16864961031871986,
      "grad_norm": 5.385419845581055,
      "learning_rate": 9.34098044732618e-05,
      "loss": 0.1545,
      "step": 4171
    },
    {
      "epoch": 0.1686900441739869,
      "grad_norm": 8.724830627441406,
      "learning_rate": 9.340663956107971e-05,
      "loss": 0.2058,
      "step": 4172
    },
    {
      "epoch": 0.1687304780292539,
      "grad_norm": 6.293845176696777,
      "learning_rate": 9.3403473942756e-05,
      "loss": 0.2159,
      "step": 4173
    },
    {
      "epoch": 0.1687709118845209,
      "grad_norm": 3.1135973930358887,
      "learning_rate": 9.340030761834213e-05,
      "loss": 0.1587,
      "step": 4174
    },
    {
      "epoch": 0.16881134573978793,
      "grad_norm": 4.1553425788879395,
      "learning_rate": 9.339714058788964e-05,
      "loss": 0.2022,
      "step": 4175
    },
    {
      "epoch": 0.16885177959505493,
      "grad_norm": 4.323387622833252,
      "learning_rate": 9.339397285145004e-05,
      "loss": 0.1282,
      "step": 4176
    },
    {
      "epoch": 0.16889221345032196,
      "grad_norm": 12.129281997680664,
      "learning_rate": 9.339080440907486e-05,
      "loss": 0.2172,
      "step": 4177
    },
    {
      "epoch": 0.16893264730558896,
      "grad_norm": 2.337456464767456,
      "learning_rate": 9.338763526081564e-05,
      "loss": 0.1137,
      "step": 4178
    },
    {
      "epoch": 0.168973081160856,
      "grad_norm": 1.895100474357605,
      "learning_rate": 9.338446540672396e-05,
      "loss": 0.1304,
      "step": 4179
    },
    {
      "epoch": 0.169013515016123,
      "grad_norm": 5.937620639801025,
      "learning_rate": 9.338129484685137e-05,
      "loss": 0.2237,
      "step": 4180
    },
    {
      "epoch": 0.16905394887139003,
      "grad_norm": 4.117339134216309,
      "learning_rate": 9.337812358124943e-05,
      "loss": 0.1411,
      "step": 4181
    },
    {
      "epoch": 0.16909438272665703,
      "grad_norm": 1.7244703769683838,
      "learning_rate": 9.337495160996977e-05,
      "loss": 0.1228,
      "step": 4182
    },
    {
      "epoch": 0.16913481658192406,
      "grad_norm": 4.076307773590088,
      "learning_rate": 9.337177893306396e-05,
      "loss": 0.1351,
      "step": 4183
    },
    {
      "epoch": 0.16917525043719106,
      "grad_norm": 3.2556886672973633,
      "learning_rate": 9.336860555058363e-05,
      "loss": 0.0941,
      "step": 4184
    },
    {
      "epoch": 0.16921568429245806,
      "grad_norm": 4.185187339782715,
      "learning_rate": 9.33654314625804e-05,
      "loss": 0.1073,
      "step": 4185
    },
    {
      "epoch": 0.1692561181477251,
      "grad_norm": 4.883532524108887,
      "learning_rate": 9.336225666910591e-05,
      "loss": 0.212,
      "step": 4186
    },
    {
      "epoch": 0.1692965520029921,
      "grad_norm": 4.827666759490967,
      "learning_rate": 9.33590811702118e-05,
      "loss": 0.1582,
      "step": 4187
    },
    {
      "epoch": 0.16933698585825913,
      "grad_norm": 2.8138620853424072,
      "learning_rate": 9.335590496594974e-05,
      "loss": 0.0705,
      "step": 4188
    },
    {
      "epoch": 0.16937741971352613,
      "grad_norm": 7.179746627807617,
      "learning_rate": 9.335272805637139e-05,
      "loss": 0.1218,
      "step": 4189
    },
    {
      "epoch": 0.16941785356879316,
      "grad_norm": 5.452759265899658,
      "learning_rate": 9.334955044152842e-05,
      "loss": 0.238,
      "step": 4190
    },
    {
      "epoch": 0.16945828742406016,
      "grad_norm": 5.021225929260254,
      "learning_rate": 9.334637212147256e-05,
      "loss": 0.1926,
      "step": 4191
    },
    {
      "epoch": 0.1694987212793272,
      "grad_norm": 4.731101036071777,
      "learning_rate": 9.334319309625549e-05,
      "loss": 0.212,
      "step": 4192
    },
    {
      "epoch": 0.1695391551345942,
      "grad_norm": 3.5529215335845947,
      "learning_rate": 9.334001336592893e-05,
      "loss": 0.1338,
      "step": 4193
    },
    {
      "epoch": 0.16957958898986122,
      "grad_norm": 3.3683011531829834,
      "learning_rate": 9.333683293054461e-05,
      "loss": 0.0857,
      "step": 4194
    },
    {
      "epoch": 0.16962002284512823,
      "grad_norm": 4.287429332733154,
      "learning_rate": 9.333365179015426e-05,
      "loss": 0.2365,
      "step": 4195
    },
    {
      "epoch": 0.16966045670039523,
      "grad_norm": 5.061662197113037,
      "learning_rate": 9.333046994480965e-05,
      "loss": 0.1215,
      "step": 4196
    },
    {
      "epoch": 0.16970089055566226,
      "grad_norm": 4.878950119018555,
      "learning_rate": 9.332728739456253e-05,
      "loss": 0.13,
      "step": 4197
    },
    {
      "epoch": 0.16974132441092926,
      "grad_norm": 6.464277744293213,
      "learning_rate": 9.332410413946469e-05,
      "loss": 0.1894,
      "step": 4198
    },
    {
      "epoch": 0.1697817582661963,
      "grad_norm": 4.508922576904297,
      "learning_rate": 9.332092017956789e-05,
      "loss": 0.2023,
      "step": 4199
    },
    {
      "epoch": 0.1698221921214633,
      "grad_norm": 3.710608720779419,
      "learning_rate": 9.331773551492394e-05,
      "loss": 0.115,
      "step": 4200
    },
    {
      "epoch": 0.16986262597673032,
      "grad_norm": 5.339834690093994,
      "learning_rate": 9.331455014558463e-05,
      "loss": 0.2373,
      "step": 4201
    },
    {
      "epoch": 0.16990305983199733,
      "grad_norm": 5.5679707527160645,
      "learning_rate": 9.331136407160182e-05,
      "loss": 0.1397,
      "step": 4202
    },
    {
      "epoch": 0.16994349368726436,
      "grad_norm": 4.896673679351807,
      "learning_rate": 9.33081772930273e-05,
      "loss": 0.1418,
      "step": 4203
    },
    {
      "epoch": 0.16998392754253136,
      "grad_norm": 3.931295871734619,
      "learning_rate": 9.330498980991293e-05,
      "loss": 0.1508,
      "step": 4204
    },
    {
      "epoch": 0.1700243613977984,
      "grad_norm": 4.503824710845947,
      "learning_rate": 9.330180162231058e-05,
      "loss": 0.0873,
      "step": 4205
    },
    {
      "epoch": 0.1700647952530654,
      "grad_norm": 7.870980262756348,
      "learning_rate": 9.329861273027209e-05,
      "loss": 0.2273,
      "step": 4206
    },
    {
      "epoch": 0.1701052291083324,
      "grad_norm": 4.269221782684326,
      "learning_rate": 9.329542313384936e-05,
      "loss": 0.1471,
      "step": 4207
    },
    {
      "epoch": 0.17014566296359943,
      "grad_norm": 3.2440121173858643,
      "learning_rate": 9.329223283309425e-05,
      "loss": 0.1383,
      "step": 4208
    },
    {
      "epoch": 0.17018609681886643,
      "grad_norm": 5.257524013519287,
      "learning_rate": 9.328904182805867e-05,
      "loss": 0.2536,
      "step": 4209
    },
    {
      "epoch": 0.17022653067413346,
      "grad_norm": 9.347061157226562,
      "learning_rate": 9.328585011879456e-05,
      "loss": 0.2429,
      "step": 4210
    },
    {
      "epoch": 0.17026696452940046,
      "grad_norm": 3.4242444038391113,
      "learning_rate": 9.328265770535381e-05,
      "loss": 0.1515,
      "step": 4211
    },
    {
      "epoch": 0.1703073983846675,
      "grad_norm": 3.9475457668304443,
      "learning_rate": 9.327946458778835e-05,
      "loss": 0.1206,
      "step": 4212
    },
    {
      "epoch": 0.1703478322399345,
      "grad_norm": 6.114081382751465,
      "learning_rate": 9.327627076615015e-05,
      "loss": 0.3847,
      "step": 4213
    },
    {
      "epoch": 0.17038826609520152,
      "grad_norm": 5.788756847381592,
      "learning_rate": 9.327307624049117e-05,
      "loss": 0.2526,
      "step": 4214
    },
    {
      "epoch": 0.17042869995046853,
      "grad_norm": 3.1934118270874023,
      "learning_rate": 9.326988101086336e-05,
      "loss": 0.0798,
      "step": 4215
    },
    {
      "epoch": 0.17046913380573553,
      "grad_norm": 3.8001770973205566,
      "learning_rate": 9.326668507731871e-05,
      "loss": 0.1322,
      "step": 4216
    },
    {
      "epoch": 0.17050956766100256,
      "grad_norm": 2.9874961376190186,
      "learning_rate": 9.32634884399092e-05,
      "loss": 0.1643,
      "step": 4217
    },
    {
      "epoch": 0.17055000151626956,
      "grad_norm": 3.6159911155700684,
      "learning_rate": 9.326029109868683e-05,
      "loss": 0.1844,
      "step": 4218
    },
    {
      "epoch": 0.1705904353715366,
      "grad_norm": 3.3814284801483154,
      "learning_rate": 9.325709305370365e-05,
      "loss": 0.1822,
      "step": 4219
    },
    {
      "epoch": 0.1706308692268036,
      "grad_norm": 3.7024478912353516,
      "learning_rate": 9.325389430501163e-05,
      "loss": 0.161,
      "step": 4220
    },
    {
      "epoch": 0.17067130308207062,
      "grad_norm": 2.9878463745117188,
      "learning_rate": 9.325069485266287e-05,
      "loss": 0.1425,
      "step": 4221
    },
    {
      "epoch": 0.17071173693733763,
      "grad_norm": 6.2307047843933105,
      "learning_rate": 9.324749469670937e-05,
      "loss": 0.2819,
      "step": 4222
    },
    {
      "epoch": 0.17075217079260466,
      "grad_norm": 11.113974571228027,
      "learning_rate": 9.324429383720323e-05,
      "loss": 0.2752,
      "step": 4223
    },
    {
      "epoch": 0.17079260464787166,
      "grad_norm": 3.757018566131592,
      "learning_rate": 9.324109227419649e-05,
      "loss": 0.1194,
      "step": 4224
    },
    {
      "epoch": 0.1708330385031387,
      "grad_norm": 2.943429946899414,
      "learning_rate": 9.323789000774124e-05,
      "loss": 0.1595,
      "step": 4225
    },
    {
      "epoch": 0.1708734723584057,
      "grad_norm": 4.2585129737854,
      "learning_rate": 9.323468703788957e-05,
      "loss": 0.2781,
      "step": 4226
    },
    {
      "epoch": 0.1709139062136727,
      "grad_norm": 3.240901470184326,
      "learning_rate": 9.323148336469361e-05,
      "loss": 0.3243,
      "step": 4227
    },
    {
      "epoch": 0.17095434006893973,
      "grad_norm": 4.582461833953857,
      "learning_rate": 9.322827898820543e-05,
      "loss": 0.0835,
      "step": 4228
    },
    {
      "epoch": 0.17099477392420673,
      "grad_norm": 3.7030351161956787,
      "learning_rate": 9.322507390847723e-05,
      "loss": 0.2157,
      "step": 4229
    },
    {
      "epoch": 0.17103520777947376,
      "grad_norm": 8.30810832977295,
      "learning_rate": 9.322186812556108e-05,
      "loss": 0.1986,
      "step": 4230
    },
    {
      "epoch": 0.17107564163474076,
      "grad_norm": 8.439467430114746,
      "learning_rate": 9.321866163950919e-05,
      "loss": 0.1519,
      "step": 4231
    },
    {
      "epoch": 0.1711160754900078,
      "grad_norm": 5.007877826690674,
      "learning_rate": 9.321545445037368e-05,
      "loss": 0.2482,
      "step": 4232
    },
    {
      "epoch": 0.1711565093452748,
      "grad_norm": 5.475041389465332,
      "learning_rate": 9.321224655820674e-05,
      "loss": 0.1145,
      "step": 4233
    },
    {
      "epoch": 0.17119694320054182,
      "grad_norm": 3.5374202728271484,
      "learning_rate": 9.320903796306055e-05,
      "loss": 0.1081,
      "step": 4234
    },
    {
      "epoch": 0.17123737705580883,
      "grad_norm": 1.7798510789871216,
      "learning_rate": 9.320582866498732e-05,
      "loss": 0.1077,
      "step": 4235
    },
    {
      "epoch": 0.17127781091107586,
      "grad_norm": 3.0172362327575684,
      "learning_rate": 9.320261866403928e-05,
      "loss": 0.0996,
      "step": 4236
    },
    {
      "epoch": 0.17131824476634286,
      "grad_norm": 4.201366424560547,
      "learning_rate": 9.31994079602686e-05,
      "loss": 0.2365,
      "step": 4237
    },
    {
      "epoch": 0.17135867862160986,
      "grad_norm": 5.801873207092285,
      "learning_rate": 9.319619655372754e-05,
      "loss": 0.1639,
      "step": 4238
    },
    {
      "epoch": 0.1713991124768769,
      "grad_norm": 4.757242202758789,
      "learning_rate": 9.319298444446834e-05,
      "loss": 0.1625,
      "step": 4239
    },
    {
      "epoch": 0.1714395463321439,
      "grad_norm": 11.004742622375488,
      "learning_rate": 9.318977163254323e-05,
      "loss": 0.1795,
      "step": 4240
    },
    {
      "epoch": 0.17147998018741092,
      "grad_norm": 2.2677500247955322,
      "learning_rate": 9.318655811800453e-05,
      "loss": 0.2209,
      "step": 4241
    },
    {
      "epoch": 0.17152041404267793,
      "grad_norm": 3.6767468452453613,
      "learning_rate": 9.318334390090449e-05,
      "loss": 0.1168,
      "step": 4242
    },
    {
      "epoch": 0.17156084789794496,
      "grad_norm": 4.721237659454346,
      "learning_rate": 9.318012898129538e-05,
      "loss": 0.1894,
      "step": 4243
    },
    {
      "epoch": 0.17160128175321196,
      "grad_norm": 2.5642311573028564,
      "learning_rate": 9.31769133592295e-05,
      "loss": 0.1004,
      "step": 4244
    },
    {
      "epoch": 0.171641715608479,
      "grad_norm": 4.532460689544678,
      "learning_rate": 9.317369703475923e-05,
      "loss": 0.1744,
      "step": 4245
    },
    {
      "epoch": 0.171682149463746,
      "grad_norm": 4.493870258331299,
      "learning_rate": 9.31704800079368e-05,
      "loss": 0.1253,
      "step": 4246
    },
    {
      "epoch": 0.17172258331901302,
      "grad_norm": 5.104554653167725,
      "learning_rate": 9.316726227881462e-05,
      "loss": 0.2191,
      "step": 4247
    },
    {
      "epoch": 0.17176301717428002,
      "grad_norm": 5.458381652832031,
      "learning_rate": 9.316404384744498e-05,
      "loss": 0.1712,
      "step": 4248
    },
    {
      "epoch": 0.17180345102954703,
      "grad_norm": 10.221200942993164,
      "learning_rate": 9.316082471388027e-05,
      "loss": 0.2988,
      "step": 4249
    },
    {
      "epoch": 0.17184388488481406,
      "grad_norm": 2.4633235931396484,
      "learning_rate": 9.315760487817285e-05,
      "loss": 0.0628,
      "step": 4250
    },
    {
      "epoch": 0.17188431874008106,
      "grad_norm": 4.862361431121826,
      "learning_rate": 9.31543843403751e-05,
      "loss": 0.2233,
      "step": 4251
    },
    {
      "epoch": 0.1719247525953481,
      "grad_norm": 5.330404281616211,
      "learning_rate": 9.315116310053941e-05,
      "loss": 0.1691,
      "step": 4252
    },
    {
      "epoch": 0.1719651864506151,
      "grad_norm": 6.101861953735352,
      "learning_rate": 9.314794115871819e-05,
      "loss": 0.2096,
      "step": 4253
    },
    {
      "epoch": 0.17200562030588212,
      "grad_norm": 4.500319957733154,
      "learning_rate": 9.314471851496385e-05,
      "loss": 0.1606,
      "step": 4254
    },
    {
      "epoch": 0.17204605416114913,
      "grad_norm": 3.420912981033325,
      "learning_rate": 9.314149516932882e-05,
      "loss": 0.1162,
      "step": 4255
    },
    {
      "epoch": 0.17208648801641616,
      "grad_norm": 4.41970682144165,
      "learning_rate": 9.313827112186553e-05,
      "loss": 0.1242,
      "step": 4256
    },
    {
      "epoch": 0.17212692187168316,
      "grad_norm": 4.2780890464782715,
      "learning_rate": 9.313504637262642e-05,
      "loss": 0.0911,
      "step": 4257
    },
    {
      "epoch": 0.1721673557269502,
      "grad_norm": 3.653949499130249,
      "learning_rate": 9.3131820921664e-05,
      "loss": 0.1225,
      "step": 4258
    },
    {
      "epoch": 0.1722077895822172,
      "grad_norm": 4.752174377441406,
      "learning_rate": 9.312859476903068e-05,
      "loss": 0.2319,
      "step": 4259
    },
    {
      "epoch": 0.1722482234374842,
      "grad_norm": 7.364134311676025,
      "learning_rate": 9.312536791477897e-05,
      "loss": 0.1698,
      "step": 4260
    },
    {
      "epoch": 0.17228865729275122,
      "grad_norm": 5.593738079071045,
      "learning_rate": 9.312214035896136e-05,
      "loss": 0.2178,
      "step": 4261
    },
    {
      "epoch": 0.17232909114801823,
      "grad_norm": 3.5313076972961426,
      "learning_rate": 9.311891210163037e-05,
      "loss": 0.1339,
      "step": 4262
    },
    {
      "epoch": 0.17236952500328526,
      "grad_norm": 7.705352783203125,
      "learning_rate": 9.31156831428385e-05,
      "loss": 0.1667,
      "step": 4263
    },
    {
      "epoch": 0.17240995885855226,
      "grad_norm": 8.788098335266113,
      "learning_rate": 9.311245348263828e-05,
      "loss": 0.2478,
      "step": 4264
    },
    {
      "epoch": 0.1724503927138193,
      "grad_norm": 4.861062526702881,
      "learning_rate": 9.310922312108228e-05,
      "loss": 0.2739,
      "step": 4265
    },
    {
      "epoch": 0.1724908265690863,
      "grad_norm": 4.322573661804199,
      "learning_rate": 9.310599205822302e-05,
      "loss": 0.2411,
      "step": 4266
    },
    {
      "epoch": 0.17253126042435332,
      "grad_norm": 4.000153064727783,
      "learning_rate": 9.310276029411306e-05,
      "loss": 0.1398,
      "step": 4267
    },
    {
      "epoch": 0.17257169427962032,
      "grad_norm": 3.8298544883728027,
      "learning_rate": 9.3099527828805e-05,
      "loss": 0.068,
      "step": 4268
    },
    {
      "epoch": 0.17261212813488735,
      "grad_norm": 2.2562005519866943,
      "learning_rate": 9.30962946623514e-05,
      "loss": 0.1012,
      "step": 4269
    },
    {
      "epoch": 0.17265256199015436,
      "grad_norm": 2.1671154499053955,
      "learning_rate": 9.309306079480487e-05,
      "loss": 0.0911,
      "step": 4270
    },
    {
      "epoch": 0.17269299584542136,
      "grad_norm": 3.7108774185180664,
      "learning_rate": 9.3089826226218e-05,
      "loss": 0.2418,
      "step": 4271
    },
    {
      "epoch": 0.1727334297006884,
      "grad_norm": 9.546771049499512,
      "learning_rate": 9.308659095664345e-05,
      "loss": 0.3097,
      "step": 4272
    },
    {
      "epoch": 0.1727738635559554,
      "grad_norm": 4.715212345123291,
      "learning_rate": 9.308335498613383e-05,
      "loss": 0.324,
      "step": 4273
    },
    {
      "epoch": 0.17281429741122242,
      "grad_norm": 6.354020595550537,
      "learning_rate": 9.308011831474178e-05,
      "loss": 0.2029,
      "step": 4274
    },
    {
      "epoch": 0.17285473126648943,
      "grad_norm": 4.600224018096924,
      "learning_rate": 9.307688094251996e-05,
      "loss": 0.1051,
      "step": 4275
    },
    {
      "epoch": 0.17289516512175646,
      "grad_norm": 4.116301536560059,
      "learning_rate": 9.307364286952101e-05,
      "loss": 0.2044,
      "step": 4276
    },
    {
      "epoch": 0.17293559897702346,
      "grad_norm": 6.4686174392700195,
      "learning_rate": 9.307040409579766e-05,
      "loss": 0.1718,
      "step": 4277
    },
    {
      "epoch": 0.1729760328322905,
      "grad_norm": 9.388900756835938,
      "learning_rate": 9.306716462140255e-05,
      "loss": 0.2191,
      "step": 4278
    },
    {
      "epoch": 0.1730164666875575,
      "grad_norm": 3.040276288986206,
      "learning_rate": 9.306392444638839e-05,
      "loss": 0.1539,
      "step": 4279
    },
    {
      "epoch": 0.1730569005428245,
      "grad_norm": 3.685434103012085,
      "learning_rate": 9.306068357080793e-05,
      "loss": 0.1228,
      "step": 4280
    },
    {
      "epoch": 0.17309733439809152,
      "grad_norm": 4.386507034301758,
      "learning_rate": 9.305744199471382e-05,
      "loss": 0.1599,
      "step": 4281
    },
    {
      "epoch": 0.17313776825335853,
      "grad_norm": 6.306042194366455,
      "learning_rate": 9.305419971815887e-05,
      "loss": 0.2045,
      "step": 4282
    },
    {
      "epoch": 0.17317820210862556,
      "grad_norm": 5.766143321990967,
      "learning_rate": 9.305095674119579e-05,
      "loss": 0.1601,
      "step": 4283
    },
    {
      "epoch": 0.17321863596389256,
      "grad_norm": 5.761847496032715,
      "learning_rate": 9.304771306387732e-05,
      "loss": 0.1425,
      "step": 4284
    },
    {
      "epoch": 0.1732590698191596,
      "grad_norm": 3.766319990158081,
      "learning_rate": 9.304446868625625e-05,
      "loss": 0.1502,
      "step": 4285
    },
    {
      "epoch": 0.1732995036744266,
      "grad_norm": 3.3468494415283203,
      "learning_rate": 9.304122360838538e-05,
      "loss": 0.1817,
      "step": 4286
    },
    {
      "epoch": 0.17333993752969362,
      "grad_norm": 3.1448676586151123,
      "learning_rate": 9.303797783031746e-05,
      "loss": 0.1022,
      "step": 4287
    },
    {
      "epoch": 0.17338037138496062,
      "grad_norm": 2.456040143966675,
      "learning_rate": 9.30347313521053e-05,
      "loss": 0.086,
      "step": 4288
    },
    {
      "epoch": 0.17342080524022765,
      "grad_norm": 7.523097038269043,
      "learning_rate": 9.303148417380174e-05,
      "loss": 0.2406,
      "step": 4289
    },
    {
      "epoch": 0.17346123909549466,
      "grad_norm": 7.747953414916992,
      "learning_rate": 9.302823629545959e-05,
      "loss": 0.1952,
      "step": 4290
    },
    {
      "epoch": 0.17350167295076166,
      "grad_norm": 2.9230031967163086,
      "learning_rate": 9.302498771713169e-05,
      "loss": 0.1285,
      "step": 4291
    },
    {
      "epoch": 0.1735421068060287,
      "grad_norm": 3.815782308578491,
      "learning_rate": 9.302173843887087e-05,
      "loss": 0.2163,
      "step": 4292
    },
    {
      "epoch": 0.1735825406612957,
      "grad_norm": 5.342432022094727,
      "learning_rate": 9.301848846073e-05,
      "loss": 0.1013,
      "step": 4293
    },
    {
      "epoch": 0.17362297451656272,
      "grad_norm": 3.381138324737549,
      "learning_rate": 9.301523778276198e-05,
      "loss": 0.146,
      "step": 4294
    },
    {
      "epoch": 0.17366340837182973,
      "grad_norm": 6.02354621887207,
      "learning_rate": 9.301198640501964e-05,
      "loss": 0.1839,
      "step": 4295
    },
    {
      "epoch": 0.17370384222709676,
      "grad_norm": 12.485997200012207,
      "learning_rate": 9.300873432755592e-05,
      "loss": 0.3377,
      "step": 4296
    },
    {
      "epoch": 0.17374427608236376,
      "grad_norm": 5.250466823577881,
      "learning_rate": 9.30054815504237e-05,
      "loss": 0.2394,
      "step": 4297
    },
    {
      "epoch": 0.1737847099376308,
      "grad_norm": 4.898703098297119,
      "learning_rate": 9.300222807367589e-05,
      "loss": 0.1139,
      "step": 4298
    },
    {
      "epoch": 0.1738251437928978,
      "grad_norm": 3.082012891769409,
      "learning_rate": 9.299897389736544e-05,
      "loss": 0.0696,
      "step": 4299
    },
    {
      "epoch": 0.17386557764816482,
      "grad_norm": 5.07089900970459,
      "learning_rate": 9.299571902154527e-05,
      "loss": 0.1884,
      "step": 4300
    },
    {
      "epoch": 0.17390601150343182,
      "grad_norm": 5.757972240447998,
      "learning_rate": 9.299246344626835e-05,
      "loss": 0.1633,
      "step": 4301
    },
    {
      "epoch": 0.17394644535869883,
      "grad_norm": 7.7953667640686035,
      "learning_rate": 9.298920717158764e-05,
      "loss": 0.2473,
      "step": 4302
    },
    {
      "epoch": 0.17398687921396586,
      "grad_norm": 4.733716011047363,
      "learning_rate": 9.298595019755607e-05,
      "loss": 0.1661,
      "step": 4303
    },
    {
      "epoch": 0.17402731306923286,
      "grad_norm": 1.346541404724121,
      "learning_rate": 9.298269252422668e-05,
      "loss": 0.0655,
      "step": 4304
    },
    {
      "epoch": 0.1740677469244999,
      "grad_norm": 3.9009954929351807,
      "learning_rate": 9.297943415165245e-05,
      "loss": 0.0977,
      "step": 4305
    },
    {
      "epoch": 0.1741081807797669,
      "grad_norm": 5.143433570861816,
      "learning_rate": 9.297617507988638e-05,
      "loss": 0.1915,
      "step": 4306
    },
    {
      "epoch": 0.17414861463503392,
      "grad_norm": 6.312656879425049,
      "learning_rate": 9.297291530898148e-05,
      "loss": 0.2442,
      "step": 4307
    },
    {
      "epoch": 0.17418904849030092,
      "grad_norm": 6.862684726715088,
      "learning_rate": 9.296965483899081e-05,
      "loss": 0.3307,
      "step": 4308
    },
    {
      "epoch": 0.17422948234556795,
      "grad_norm": 2.7620761394500732,
      "learning_rate": 9.296639366996738e-05,
      "loss": 0.1469,
      "step": 4309
    },
    {
      "epoch": 0.17426991620083496,
      "grad_norm": 3.3018126487731934,
      "learning_rate": 9.296313180196425e-05,
      "loss": 0.1643,
      "step": 4310
    },
    {
      "epoch": 0.174310350056102,
      "grad_norm": 5.8738179206848145,
      "learning_rate": 9.29598692350345e-05,
      "loss": 0.176,
      "step": 4311
    },
    {
      "epoch": 0.174350783911369,
      "grad_norm": 11.20573902130127,
      "learning_rate": 9.29566059692312e-05,
      "loss": 0.2988,
      "step": 4312
    },
    {
      "epoch": 0.174391217766636,
      "grad_norm": 6.585662364959717,
      "learning_rate": 9.295334200460742e-05,
      "loss": 0.1605,
      "step": 4313
    },
    {
      "epoch": 0.17443165162190302,
      "grad_norm": 3.2608625888824463,
      "learning_rate": 9.295007734121626e-05,
      "loss": 0.2258,
      "step": 4314
    },
    {
      "epoch": 0.17447208547717002,
      "grad_norm": 3.5847327709198,
      "learning_rate": 9.294681197911086e-05,
      "loss": 0.1582,
      "step": 4315
    },
    {
      "epoch": 0.17451251933243705,
      "grad_norm": 3.3019821643829346,
      "learning_rate": 9.294354591834431e-05,
      "loss": 0.2087,
      "step": 4316
    },
    {
      "epoch": 0.17455295318770406,
      "grad_norm": 10.59052562713623,
      "learning_rate": 9.294027915896977e-05,
      "loss": 0.3477,
      "step": 4317
    },
    {
      "epoch": 0.1745933870429711,
      "grad_norm": 5.488692760467529,
      "learning_rate": 9.293701170104036e-05,
      "loss": 0.1556,
      "step": 4318
    },
    {
      "epoch": 0.1746338208982381,
      "grad_norm": 6.677713394165039,
      "learning_rate": 9.293374354460924e-05,
      "loss": 0.2642,
      "step": 4319
    },
    {
      "epoch": 0.17467425475350512,
      "grad_norm": 4.488563060760498,
      "learning_rate": 9.293047468972957e-05,
      "loss": 0.173,
      "step": 4320
    },
    {
      "epoch": 0.17471468860877212,
      "grad_norm": 4.447488784790039,
      "learning_rate": 9.292720513645455e-05,
      "loss": 0.2035,
      "step": 4321
    },
    {
      "epoch": 0.17475512246403915,
      "grad_norm": 6.935637950897217,
      "learning_rate": 9.292393488483736e-05,
      "loss": 0.2279,
      "step": 4322
    },
    {
      "epoch": 0.17479555631930616,
      "grad_norm": 3.453447103500366,
      "learning_rate": 9.29206639349312e-05,
      "loss": 0.1422,
      "step": 4323
    },
    {
      "epoch": 0.17483599017457316,
      "grad_norm": 5.5558061599731445,
      "learning_rate": 9.291739228678927e-05,
      "loss": 0.1514,
      "step": 4324
    },
    {
      "epoch": 0.1748764240298402,
      "grad_norm": 5.012943744659424,
      "learning_rate": 9.29141199404648e-05,
      "loss": 0.1571,
      "step": 4325
    },
    {
      "epoch": 0.1749168578851072,
      "grad_norm": 5.739306449890137,
      "learning_rate": 9.291084689601103e-05,
      "loss": 0.1895,
      "step": 4326
    },
    {
      "epoch": 0.17495729174037422,
      "grad_norm": 1.8362292051315308,
      "learning_rate": 9.29075731534812e-05,
      "loss": 0.0653,
      "step": 4327
    },
    {
      "epoch": 0.17499772559564122,
      "grad_norm": 4.9207444190979,
      "learning_rate": 9.290429871292858e-05,
      "loss": 0.1646,
      "step": 4328
    },
    {
      "epoch": 0.17503815945090825,
      "grad_norm": 7.967036724090576,
      "learning_rate": 9.290102357440642e-05,
      "loss": 0.2599,
      "step": 4329
    },
    {
      "epoch": 0.17507859330617526,
      "grad_norm": 1.8148075342178345,
      "learning_rate": 9.289774773796802e-05,
      "loss": 0.1071,
      "step": 4330
    },
    {
      "epoch": 0.1751190271614423,
      "grad_norm": 3.339397668838501,
      "learning_rate": 9.289447120366666e-05,
      "loss": 0.0786,
      "step": 4331
    },
    {
      "epoch": 0.1751594610167093,
      "grad_norm": 5.249443054199219,
      "learning_rate": 9.289119397155563e-05,
      "loss": 0.2243,
      "step": 4332
    },
    {
      "epoch": 0.17519989487197632,
      "grad_norm": 4.90446662902832,
      "learning_rate": 9.288791604168827e-05,
      "loss": 0.1098,
      "step": 4333
    },
    {
      "epoch": 0.17524032872724332,
      "grad_norm": 5.0833563804626465,
      "learning_rate": 9.288463741411791e-05,
      "loss": 0.2269,
      "step": 4334
    },
    {
      "epoch": 0.17528076258251032,
      "grad_norm": 6.35407018661499,
      "learning_rate": 9.288135808889784e-05,
      "loss": 0.3076,
      "step": 4335
    },
    {
      "epoch": 0.17532119643777735,
      "grad_norm": 3.652513265609741,
      "learning_rate": 9.287807806608146e-05,
      "loss": 0.1591,
      "step": 4336
    },
    {
      "epoch": 0.17536163029304436,
      "grad_norm": 5.488552093505859,
      "learning_rate": 9.287479734572211e-05,
      "loss": 0.1762,
      "step": 4337
    },
    {
      "epoch": 0.1754020641483114,
      "grad_norm": 4.081103324890137,
      "learning_rate": 9.287151592787315e-05,
      "loss": 0.0843,
      "step": 4338
    },
    {
      "epoch": 0.1754424980035784,
      "grad_norm": 5.227813720703125,
      "learning_rate": 9.286823381258796e-05,
      "loss": 0.1702,
      "step": 4339
    },
    {
      "epoch": 0.17548293185884542,
      "grad_norm": 2.7027454376220703,
      "learning_rate": 9.286495099991996e-05,
      "loss": 0.1314,
      "step": 4340
    },
    {
      "epoch": 0.17552336571411242,
      "grad_norm": 2.521045207977295,
      "learning_rate": 9.286166748992254e-05,
      "loss": 0.1176,
      "step": 4341
    },
    {
      "epoch": 0.17556379956937945,
      "grad_norm": 1.9059206247329712,
      "learning_rate": 9.285838328264912e-05,
      "loss": 0.0873,
      "step": 4342
    },
    {
      "epoch": 0.17560423342464646,
      "grad_norm": 3.914314031600952,
      "learning_rate": 9.285509837815312e-05,
      "loss": 0.112,
      "step": 4343
    },
    {
      "epoch": 0.17564466727991346,
      "grad_norm": 4.497342109680176,
      "learning_rate": 9.285181277648798e-05,
      "loss": 0.1379,
      "step": 4344
    },
    {
      "epoch": 0.1756851011351805,
      "grad_norm": 10.084634780883789,
      "learning_rate": 9.284852647770716e-05,
      "loss": 0.2378,
      "step": 4345
    },
    {
      "epoch": 0.1757255349904475,
      "grad_norm": 7.86751651763916,
      "learning_rate": 9.284523948186412e-05,
      "loss": 0.1788,
      "step": 4346
    },
    {
      "epoch": 0.17576596884571452,
      "grad_norm": 5.939727306365967,
      "learning_rate": 9.284195178901231e-05,
      "loss": 0.2232,
      "step": 4347
    },
    {
      "epoch": 0.17580640270098152,
      "grad_norm": 3.316871404647827,
      "learning_rate": 9.283866339920526e-05,
      "loss": 0.1253,
      "step": 4348
    },
    {
      "epoch": 0.17584683655624855,
      "grad_norm": 6.530004978179932,
      "learning_rate": 9.283537431249643e-05,
      "loss": 0.3258,
      "step": 4349
    },
    {
      "epoch": 0.17588727041151556,
      "grad_norm": 3.963029146194458,
      "learning_rate": 9.283208452893932e-05,
      "loss": 0.1965,
      "step": 4350
    },
    {
      "epoch": 0.1759277042667826,
      "grad_norm": 5.384074687957764,
      "learning_rate": 9.282879404858748e-05,
      "loss": 0.3195,
      "step": 4351
    },
    {
      "epoch": 0.1759681381220496,
      "grad_norm": 3.381303548812866,
      "learning_rate": 9.282550287149441e-05,
      "loss": 0.156,
      "step": 4352
    },
    {
      "epoch": 0.17600857197731662,
      "grad_norm": 5.072704792022705,
      "learning_rate": 9.282221099771367e-05,
      "loss": 0.1426,
      "step": 4353
    },
    {
      "epoch": 0.17604900583258362,
      "grad_norm": 6.263906955718994,
      "learning_rate": 9.281891842729882e-05,
      "loss": 0.138,
      "step": 4354
    },
    {
      "epoch": 0.17608943968785062,
      "grad_norm": 11.107349395751953,
      "learning_rate": 9.281562516030339e-05,
      "loss": 0.3396,
      "step": 4355
    },
    {
      "epoch": 0.17612987354311765,
      "grad_norm": 10.235193252563477,
      "learning_rate": 9.281233119678098e-05,
      "loss": 0.1808,
      "step": 4356
    },
    {
      "epoch": 0.17617030739838466,
      "grad_norm": 6.195062637329102,
      "learning_rate": 9.280903653678519e-05,
      "loss": 0.1293,
      "step": 4357
    },
    {
      "epoch": 0.1762107412536517,
      "grad_norm": 10.960775375366211,
      "learning_rate": 9.280574118036959e-05,
      "loss": 0.1875,
      "step": 4358
    },
    {
      "epoch": 0.1762511751089187,
      "grad_norm": 7.035867214202881,
      "learning_rate": 9.28024451275878e-05,
      "loss": 0.1976,
      "step": 4359
    },
    {
      "epoch": 0.17629160896418572,
      "grad_norm": 11.931193351745605,
      "learning_rate": 9.279914837849344e-05,
      "loss": 0.3056,
      "step": 4360
    },
    {
      "epoch": 0.17633204281945272,
      "grad_norm": 5.746275424957275,
      "learning_rate": 9.279585093314013e-05,
      "loss": 0.2065,
      "step": 4361
    },
    {
      "epoch": 0.17637247667471975,
      "grad_norm": 4.864289283752441,
      "learning_rate": 9.279255279158153e-05,
      "loss": 0.116,
      "step": 4362
    },
    {
      "epoch": 0.17641291052998676,
      "grad_norm": 3.739030122756958,
      "learning_rate": 9.27892539538713e-05,
      "loss": 0.2686,
      "step": 4363
    },
    {
      "epoch": 0.17645334438525379,
      "grad_norm": 10.259282112121582,
      "learning_rate": 9.278595442006308e-05,
      "loss": 0.2726,
      "step": 4364
    },
    {
      "epoch": 0.1764937782405208,
      "grad_norm": 5.805829048156738,
      "learning_rate": 9.278265419021057e-05,
      "loss": 0.1371,
      "step": 4365
    },
    {
      "epoch": 0.1765342120957878,
      "grad_norm": 5.157618522644043,
      "learning_rate": 9.277935326436744e-05,
      "loss": 0.188,
      "step": 4366
    },
    {
      "epoch": 0.17657464595105482,
      "grad_norm": 6.679461479187012,
      "learning_rate": 9.277605164258742e-05,
      "loss": 0.3253,
      "step": 4367
    },
    {
      "epoch": 0.17661507980632182,
      "grad_norm": 3.4621670246124268,
      "learning_rate": 9.277274932492418e-05,
      "loss": 0.1451,
      "step": 4368
    },
    {
      "epoch": 0.17665551366158885,
      "grad_norm": 4.758169651031494,
      "learning_rate": 9.276944631143148e-05,
      "loss": 0.1778,
      "step": 4369
    },
    {
      "epoch": 0.17669594751685586,
      "grad_norm": 6.1408233642578125,
      "learning_rate": 9.276614260216301e-05,
      "loss": 0.1583,
      "step": 4370
    },
    {
      "epoch": 0.17673638137212289,
      "grad_norm": 2.638777017593384,
      "learning_rate": 9.276283819717256e-05,
      "loss": 0.1908,
      "step": 4371
    },
    {
      "epoch": 0.1767768152273899,
      "grad_norm": 4.531090259552002,
      "learning_rate": 9.275953309651386e-05,
      "loss": 0.2333,
      "step": 4372
    },
    {
      "epoch": 0.17681724908265692,
      "grad_norm": 3.307896614074707,
      "learning_rate": 9.27562273002407e-05,
      "loss": 0.111,
      "step": 4373
    },
    {
      "epoch": 0.17685768293792392,
      "grad_norm": 4.843299388885498,
      "learning_rate": 9.275292080840683e-05,
      "loss": 0.1052,
      "step": 4374
    },
    {
      "epoch": 0.17689811679319095,
      "grad_norm": 1.3967288732528687,
      "learning_rate": 9.274961362106604e-05,
      "loss": 0.0572,
      "step": 4375
    },
    {
      "epoch": 0.17693855064845795,
      "grad_norm": 3.7211451530456543,
      "learning_rate": 9.274630573827217e-05,
      "loss": 0.1336,
      "step": 4376
    },
    {
      "epoch": 0.17697898450372496,
      "grad_norm": 6.906421661376953,
      "learning_rate": 9.274299716007901e-05,
      "loss": 0.1822,
      "step": 4377
    },
    {
      "epoch": 0.177019418358992,
      "grad_norm": 7.113742351531982,
      "learning_rate": 9.273968788654037e-05,
      "loss": 0.2257,
      "step": 4378
    },
    {
      "epoch": 0.177059852214259,
      "grad_norm": 5.520051002502441,
      "learning_rate": 9.27363779177101e-05,
      "loss": 0.1393,
      "step": 4379
    },
    {
      "epoch": 0.17710028606952602,
      "grad_norm": 7.218448638916016,
      "learning_rate": 9.273306725364204e-05,
      "loss": 0.1495,
      "step": 4380
    },
    {
      "epoch": 0.17714071992479302,
      "grad_norm": 5.937231540679932,
      "learning_rate": 9.272975589439006e-05,
      "loss": 0.2133,
      "step": 4381
    },
    {
      "epoch": 0.17718115378006005,
      "grad_norm": 3.1106293201446533,
      "learning_rate": 9.272644384000802e-05,
      "loss": 0.1431,
      "step": 4382
    },
    {
      "epoch": 0.17722158763532705,
      "grad_norm": 5.594428539276123,
      "learning_rate": 9.272313109054979e-05,
      "loss": 0.1308,
      "step": 4383
    },
    {
      "epoch": 0.17726202149059408,
      "grad_norm": 7.027848720550537,
      "learning_rate": 9.271981764606929e-05,
      "loss": 0.1626,
      "step": 4384
    },
    {
      "epoch": 0.1773024553458611,
      "grad_norm": 3.3546087741851807,
      "learning_rate": 9.27165035066204e-05,
      "loss": 0.091,
      "step": 4385
    },
    {
      "epoch": 0.17734288920112812,
      "grad_norm": 6.9290547370910645,
      "learning_rate": 9.271318867225706e-05,
      "loss": 0.2042,
      "step": 4386
    },
    {
      "epoch": 0.17738332305639512,
      "grad_norm": 8.209240913391113,
      "learning_rate": 9.270987314303316e-05,
      "loss": 0.2337,
      "step": 4387
    },
    {
      "epoch": 0.17742375691166212,
      "grad_norm": 9.686502456665039,
      "learning_rate": 9.270655691900266e-05,
      "loss": 0.261,
      "step": 4388
    },
    {
      "epoch": 0.17746419076692915,
      "grad_norm": 6.6962714195251465,
      "learning_rate": 9.270324000021951e-05,
      "loss": 0.2486,
      "step": 4389
    },
    {
      "epoch": 0.17750462462219616,
      "grad_norm": 3.453240156173706,
      "learning_rate": 9.269992238673766e-05,
      "loss": 0.1245,
      "step": 4390
    },
    {
      "epoch": 0.17754505847746319,
      "grad_norm": 4.729225158691406,
      "learning_rate": 9.269660407861109e-05,
      "loss": 0.0953,
      "step": 4391
    },
    {
      "epoch": 0.1775854923327302,
      "grad_norm": 2.739119052886963,
      "learning_rate": 9.269328507589376e-05,
      "loss": 0.0654,
      "step": 4392
    },
    {
      "epoch": 0.17762592618799722,
      "grad_norm": 3.271611452102661,
      "learning_rate": 9.268996537863969e-05,
      "loss": 0.1844,
      "step": 4393
    },
    {
      "epoch": 0.17766636004326422,
      "grad_norm": 5.702142715454102,
      "learning_rate": 9.268664498690289e-05,
      "loss": 0.138,
      "step": 4394
    },
    {
      "epoch": 0.17770679389853125,
      "grad_norm": 3.716280937194824,
      "learning_rate": 9.268332390073735e-05,
      "loss": 0.2911,
      "step": 4395
    },
    {
      "epoch": 0.17774722775379825,
      "grad_norm": 4.484737873077393,
      "learning_rate": 9.268000212019711e-05,
      "loss": 0.1723,
      "step": 4396
    },
    {
      "epoch": 0.17778766160906528,
      "grad_norm": 5.22304105758667,
      "learning_rate": 9.267667964533622e-05,
      "loss": 0.2263,
      "step": 4397
    },
    {
      "epoch": 0.1778280954643323,
      "grad_norm": 3.2524282932281494,
      "learning_rate": 9.267335647620871e-05,
      "loss": 0.1606,
      "step": 4398
    },
    {
      "epoch": 0.1778685293195993,
      "grad_norm": 10.37735366821289,
      "learning_rate": 9.267003261286866e-05,
      "loss": 0.277,
      "step": 4399
    },
    {
      "epoch": 0.17790896317486632,
      "grad_norm": 3.5255298614501953,
      "learning_rate": 9.266670805537012e-05,
      "loss": 0.1328,
      "step": 4400
    },
    {
      "epoch": 0.17794939703013332,
      "grad_norm": 3.905160665512085,
      "learning_rate": 9.266338280376719e-05,
      "loss": 0.1164,
      "step": 4401
    },
    {
      "epoch": 0.17798983088540035,
      "grad_norm": 2.3716328144073486,
      "learning_rate": 9.266005685811397e-05,
      "loss": 0.1094,
      "step": 4402
    },
    {
      "epoch": 0.17803026474066735,
      "grad_norm": 7.7211079597473145,
      "learning_rate": 9.265673021846455e-05,
      "loss": 0.2613,
      "step": 4403
    },
    {
      "epoch": 0.17807069859593438,
      "grad_norm": 5.067431449890137,
      "learning_rate": 9.265340288487307e-05,
      "loss": 0.1447,
      "step": 4404
    },
    {
      "epoch": 0.1781111324512014,
      "grad_norm": 3.62111759185791,
      "learning_rate": 9.265007485739366e-05,
      "loss": 0.1294,
      "step": 4405
    },
    {
      "epoch": 0.17815156630646842,
      "grad_norm": 3.8930881023406982,
      "learning_rate": 9.264674613608042e-05,
      "loss": 0.111,
      "step": 4406
    },
    {
      "epoch": 0.17819200016173542,
      "grad_norm": 6.474362373352051,
      "learning_rate": 9.264341672098754e-05,
      "loss": 0.161,
      "step": 4407
    },
    {
      "epoch": 0.17823243401700242,
      "grad_norm": 7.12742805480957,
      "learning_rate": 9.264008661216916e-05,
      "loss": 0.1894,
      "step": 4408
    },
    {
      "epoch": 0.17827286787226945,
      "grad_norm": 7.134616851806641,
      "learning_rate": 9.263675580967949e-05,
      "loss": 0.162,
      "step": 4409
    },
    {
      "epoch": 0.17831330172753646,
      "grad_norm": 8.037281036376953,
      "learning_rate": 9.263342431357269e-05,
      "loss": 0.379,
      "step": 4410
    },
    {
      "epoch": 0.17835373558280349,
      "grad_norm": 4.308222770690918,
      "learning_rate": 9.263009212390295e-05,
      "loss": 0.2025,
      "step": 4411
    },
    {
      "epoch": 0.1783941694380705,
      "grad_norm": 4.719995021820068,
      "learning_rate": 9.262675924072448e-05,
      "loss": 0.1054,
      "step": 4412
    },
    {
      "epoch": 0.17843460329333752,
      "grad_norm": 5.862801551818848,
      "learning_rate": 9.262342566409152e-05,
      "loss": 0.1162,
      "step": 4413
    },
    {
      "epoch": 0.17847503714860452,
      "grad_norm": 3.7556369304656982,
      "learning_rate": 9.262009139405827e-05,
      "loss": 0.2357,
      "step": 4414
    },
    {
      "epoch": 0.17851547100387155,
      "grad_norm": 6.281651496887207,
      "learning_rate": 9.261675643067902e-05,
      "loss": 0.1462,
      "step": 4415
    },
    {
      "epoch": 0.17855590485913855,
      "grad_norm": 5.726341247558594,
      "learning_rate": 9.261342077400798e-05,
      "loss": 0.1127,
      "step": 4416
    },
    {
      "epoch": 0.17859633871440558,
      "grad_norm": 5.502821922302246,
      "learning_rate": 9.261008442409942e-05,
      "loss": 0.1506,
      "step": 4417
    },
    {
      "epoch": 0.17863677256967259,
      "grad_norm": 7.505772590637207,
      "learning_rate": 9.260674738100763e-05,
      "loss": 0.4401,
      "step": 4418
    },
    {
      "epoch": 0.1786772064249396,
      "grad_norm": 6.100683212280273,
      "learning_rate": 9.260340964478689e-05,
      "loss": 0.1149,
      "step": 4419
    },
    {
      "epoch": 0.17871764028020662,
      "grad_norm": 4.358375549316406,
      "learning_rate": 9.26000712154915e-05,
      "loss": 0.1445,
      "step": 4420
    },
    {
      "epoch": 0.17875807413547362,
      "grad_norm": 4.7953643798828125,
      "learning_rate": 9.259673209317579e-05,
      "loss": 0.2549,
      "step": 4421
    },
    {
      "epoch": 0.17879850799074065,
      "grad_norm": 4.245721817016602,
      "learning_rate": 9.259339227789403e-05,
      "loss": 0.1406,
      "step": 4422
    },
    {
      "epoch": 0.17883894184600765,
      "grad_norm": 2.4155139923095703,
      "learning_rate": 9.25900517697006e-05,
      "loss": 0.0597,
      "step": 4423
    },
    {
      "epoch": 0.17887937570127468,
      "grad_norm": 6.459728240966797,
      "learning_rate": 9.258671056864983e-05,
      "loss": 0.2287,
      "step": 4424
    },
    {
      "epoch": 0.1789198095565417,
      "grad_norm": 6.364502429962158,
      "learning_rate": 9.258336867479605e-05,
      "loss": 0.2586,
      "step": 4425
    },
    {
      "epoch": 0.17896024341180872,
      "grad_norm": 3.3493449687957764,
      "learning_rate": 9.258002608819366e-05,
      "loss": 0.1415,
      "step": 4426
    },
    {
      "epoch": 0.17900067726707572,
      "grad_norm": 8.526144981384277,
      "learning_rate": 9.257668280889703e-05,
      "loss": 0.226,
      "step": 4427
    },
    {
      "epoch": 0.17904111112234275,
      "grad_norm": 7.9027628898620605,
      "learning_rate": 9.257333883696053e-05,
      "loss": 0.2901,
      "step": 4428
    },
    {
      "epoch": 0.17908154497760975,
      "grad_norm": 5.973458290100098,
      "learning_rate": 9.256999417243859e-05,
      "loss": 0.1696,
      "step": 4429
    },
    {
      "epoch": 0.17912197883287675,
      "grad_norm": 3.1591739654541016,
      "learning_rate": 9.25666488153856e-05,
      "loss": 0.1322,
      "step": 4430
    },
    {
      "epoch": 0.17916241268814379,
      "grad_norm": 11.17094898223877,
      "learning_rate": 9.256330276585598e-05,
      "loss": 0.3371,
      "step": 4431
    },
    {
      "epoch": 0.1792028465434108,
      "grad_norm": 2.272008180618286,
      "learning_rate": 9.255995602390417e-05,
      "loss": 0.1484,
      "step": 4432
    },
    {
      "epoch": 0.17924328039867782,
      "grad_norm": 4.741401672363281,
      "learning_rate": 9.255660858958462e-05,
      "loss": 0.1562,
      "step": 4433
    },
    {
      "epoch": 0.17928371425394482,
      "grad_norm": 4.808089733123779,
      "learning_rate": 9.255326046295179e-05,
      "loss": 0.1012,
      "step": 4434
    },
    {
      "epoch": 0.17932414810921185,
      "grad_norm": 5.381745338439941,
      "learning_rate": 9.254991164406013e-05,
      "loss": 0.1983,
      "step": 4435
    },
    {
      "epoch": 0.17936458196447885,
      "grad_norm": 5.0504631996154785,
      "learning_rate": 9.254656213296414e-05,
      "loss": 0.1874,
      "step": 4436
    },
    {
      "epoch": 0.17940501581974588,
      "grad_norm": 5.829972267150879,
      "learning_rate": 9.254321192971829e-05,
      "loss": 0.1657,
      "step": 4437
    },
    {
      "epoch": 0.17944544967501289,
      "grad_norm": 3.2080633640289307,
      "learning_rate": 9.253986103437709e-05,
      "loss": 0.1905,
      "step": 4438
    },
    {
      "epoch": 0.17948588353027992,
      "grad_norm": 9.317557334899902,
      "learning_rate": 9.253650944699504e-05,
      "loss": 0.3111,
      "step": 4439
    },
    {
      "epoch": 0.17952631738554692,
      "grad_norm": 6.989106178283691,
      "learning_rate": 9.25331571676267e-05,
      "loss": 0.3218,
      "step": 4440
    },
    {
      "epoch": 0.17956675124081392,
      "grad_norm": 3.025265693664551,
      "learning_rate": 9.252980419632656e-05,
      "loss": 0.1196,
      "step": 4441
    },
    {
      "epoch": 0.17960718509608095,
      "grad_norm": 2.5479373931884766,
      "learning_rate": 9.25264505331492e-05,
      "loss": 0.0889,
      "step": 4442
    },
    {
      "epoch": 0.17964761895134795,
      "grad_norm": 4.70698881149292,
      "learning_rate": 9.252309617814917e-05,
      "loss": 0.2799,
      "step": 4443
    },
    {
      "epoch": 0.17968805280661498,
      "grad_norm": 2.40370774269104,
      "learning_rate": 9.251974113138102e-05,
      "loss": 0.0548,
      "step": 4444
    },
    {
      "epoch": 0.179728486661882,
      "grad_norm": 3.7621653079986572,
      "learning_rate": 9.251638539289935e-05,
      "loss": 0.1263,
      "step": 4445
    },
    {
      "epoch": 0.17976892051714902,
      "grad_norm": 6.082239627838135,
      "learning_rate": 9.251302896275875e-05,
      "loss": 0.2187,
      "step": 4446
    },
    {
      "epoch": 0.17980935437241602,
      "grad_norm": 6.723329067230225,
      "learning_rate": 9.25096718410138e-05,
      "loss": 0.2074,
      "step": 4447
    },
    {
      "epoch": 0.17984978822768305,
      "grad_norm": 13.329764366149902,
      "learning_rate": 9.250631402771914e-05,
      "loss": 0.3045,
      "step": 4448
    },
    {
      "epoch": 0.17989022208295005,
      "grad_norm": 8.896671295166016,
      "learning_rate": 9.25029555229294e-05,
      "loss": 0.2814,
      "step": 4449
    },
    {
      "epoch": 0.17993065593821708,
      "grad_norm": 10.127045631408691,
      "learning_rate": 9.249959632669918e-05,
      "loss": 0.4082,
      "step": 4450
    },
    {
      "epoch": 0.17997108979348408,
      "grad_norm": 3.0729527473449707,
      "learning_rate": 9.249623643908317e-05,
      "loss": 0.1267,
      "step": 4451
    },
    {
      "epoch": 0.1800115236487511,
      "grad_norm": 2.694528341293335,
      "learning_rate": 9.249287586013601e-05,
      "loss": 0.1344,
      "step": 4452
    },
    {
      "epoch": 0.18005195750401812,
      "grad_norm": 3.431840658187866,
      "learning_rate": 9.248951458991236e-05,
      "loss": 0.1565,
      "step": 4453
    },
    {
      "epoch": 0.18009239135928512,
      "grad_norm": 11.072198867797852,
      "learning_rate": 9.248615262846692e-05,
      "loss": 0.3531,
      "step": 4454
    },
    {
      "epoch": 0.18013282521455215,
      "grad_norm": 3.617475986480713,
      "learning_rate": 9.248278997585437e-05,
      "loss": 0.1678,
      "step": 4455
    },
    {
      "epoch": 0.18017325906981915,
      "grad_norm": 7.446293830871582,
      "learning_rate": 9.247942663212941e-05,
      "loss": 0.176,
      "step": 4456
    },
    {
      "epoch": 0.18021369292508618,
      "grad_norm": 4.605300426483154,
      "learning_rate": 9.247606259734677e-05,
      "loss": 0.1385,
      "step": 4457
    },
    {
      "epoch": 0.18025412678035319,
      "grad_norm": 3.6539905071258545,
      "learning_rate": 9.247269787156116e-05,
      "loss": 0.1357,
      "step": 4458
    },
    {
      "epoch": 0.18029456063562022,
      "grad_norm": 4.3361287117004395,
      "learning_rate": 9.246933245482734e-05,
      "loss": 0.2021,
      "step": 4459
    },
    {
      "epoch": 0.18033499449088722,
      "grad_norm": 4.364213466644287,
      "learning_rate": 9.246596634720004e-05,
      "loss": 0.2826,
      "step": 4460
    },
    {
      "epoch": 0.18037542834615425,
      "grad_norm": 7.95551061630249,
      "learning_rate": 9.246259954873402e-05,
      "loss": 0.1849,
      "step": 4461
    },
    {
      "epoch": 0.18041586220142125,
      "grad_norm": 5.591546058654785,
      "learning_rate": 9.245923205948406e-05,
      "loss": 0.2574,
      "step": 4462
    },
    {
      "epoch": 0.18045629605668825,
      "grad_norm": 7.810454845428467,
      "learning_rate": 9.245586387950494e-05,
      "loss": 0.2132,
      "step": 4463
    },
    {
      "epoch": 0.18049672991195528,
      "grad_norm": 5.358616828918457,
      "learning_rate": 9.245249500885144e-05,
      "loss": 0.2102,
      "step": 4464
    },
    {
      "epoch": 0.1805371637672223,
      "grad_norm": 13.032264709472656,
      "learning_rate": 9.24491254475784e-05,
      "loss": 0.453,
      "step": 4465
    },
    {
      "epoch": 0.18057759762248932,
      "grad_norm": 10.73732852935791,
      "learning_rate": 9.244575519574062e-05,
      "loss": 0.263,
      "step": 4466
    },
    {
      "epoch": 0.18061803147775632,
      "grad_norm": 5.4014129638671875,
      "learning_rate": 9.244238425339289e-05,
      "loss": 0.1952,
      "step": 4467
    },
    {
      "epoch": 0.18065846533302335,
      "grad_norm": 6.680059909820557,
      "learning_rate": 9.243901262059009e-05,
      "loss": 0.1445,
      "step": 4468
    },
    {
      "epoch": 0.18069889918829035,
      "grad_norm": 3.8557798862457275,
      "learning_rate": 9.243564029738707e-05,
      "loss": 0.1257,
      "step": 4469
    },
    {
      "epoch": 0.18073933304355738,
      "grad_norm": 2.3113248348236084,
      "learning_rate": 9.243226728383867e-05,
      "loss": 0.1598,
      "step": 4470
    },
    {
      "epoch": 0.18077976689882438,
      "grad_norm": 8.643477439880371,
      "learning_rate": 9.242889357999978e-05,
      "loss": 0.2859,
      "step": 4471
    },
    {
      "epoch": 0.1808202007540914,
      "grad_norm": 9.467827796936035,
      "learning_rate": 9.242551918592527e-05,
      "loss": 0.2103,
      "step": 4472
    },
    {
      "epoch": 0.18086063460935842,
      "grad_norm": 10.297444343566895,
      "learning_rate": 9.242214410167007e-05,
      "loss": 0.2412,
      "step": 4473
    },
    {
      "epoch": 0.18090106846462542,
      "grad_norm": 12.236175537109375,
      "learning_rate": 9.241876832728902e-05,
      "loss": 0.3237,
      "step": 4474
    },
    {
      "epoch": 0.18094150231989245,
      "grad_norm": 3.610776901245117,
      "learning_rate": 9.24153918628371e-05,
      "loss": 0.1116,
      "step": 4475
    },
    {
      "epoch": 0.18098193617515945,
      "grad_norm": 14.347509384155273,
      "learning_rate": 9.24120147083692e-05,
      "loss": 0.4172,
      "step": 4476
    },
    {
      "epoch": 0.18102237003042648,
      "grad_norm": 7.501463413238525,
      "learning_rate": 9.240863686394027e-05,
      "loss": 0.2119,
      "step": 4477
    },
    {
      "epoch": 0.18106280388569349,
      "grad_norm": 6.820019721984863,
      "learning_rate": 9.240525832960527e-05,
      "loss": 0.197,
      "step": 4478
    },
    {
      "epoch": 0.18110323774096052,
      "grad_norm": 8.644341468811035,
      "learning_rate": 9.240187910541917e-05,
      "loss": 0.3481,
      "step": 4479
    },
    {
      "epoch": 0.18114367159622752,
      "grad_norm": 2.9839518070220947,
      "learning_rate": 9.239849919143692e-05,
      "loss": 0.2375,
      "step": 4480
    },
    {
      "epoch": 0.18118410545149455,
      "grad_norm": 4.960442543029785,
      "learning_rate": 9.23951185877135e-05,
      "loss": 0.0926,
      "step": 4481
    },
    {
      "epoch": 0.18122453930676155,
      "grad_norm": 5.43528413772583,
      "learning_rate": 9.239173729430394e-05,
      "loss": 0.1214,
      "step": 4482
    },
    {
      "epoch": 0.18126497316202855,
      "grad_norm": 8.581709861755371,
      "learning_rate": 9.238835531126322e-05,
      "loss": 0.2002,
      "step": 4483
    },
    {
      "epoch": 0.18130540701729558,
      "grad_norm": 5.703802108764648,
      "learning_rate": 9.238497263864637e-05,
      "loss": 0.1848,
      "step": 4484
    },
    {
      "epoch": 0.18134584087256259,
      "grad_norm": 7.704061508178711,
      "learning_rate": 9.238158927650842e-05,
      "loss": 0.2698,
      "step": 4485
    },
    {
      "epoch": 0.18138627472782962,
      "grad_norm": 5.25504207611084,
      "learning_rate": 9.237820522490439e-05,
      "loss": 0.0843,
      "step": 4486
    },
    {
      "epoch": 0.18142670858309662,
      "grad_norm": 3.9673051834106445,
      "learning_rate": 9.237482048388936e-05,
      "loss": 0.1664,
      "step": 4487
    },
    {
      "epoch": 0.18146714243836365,
      "grad_norm": 6.050958156585693,
      "learning_rate": 9.237143505351837e-05,
      "loss": 0.2559,
      "step": 4488
    },
    {
      "epoch": 0.18150757629363065,
      "grad_norm": 4.628909587860107,
      "learning_rate": 9.236804893384652e-05,
      "loss": 0.0971,
      "step": 4489
    },
    {
      "epoch": 0.18154801014889768,
      "grad_norm": 4.418152809143066,
      "learning_rate": 9.236466212492887e-05,
      "loss": 0.1366,
      "step": 4490
    },
    {
      "epoch": 0.18158844400416468,
      "grad_norm": 5.126473903656006,
      "learning_rate": 9.236127462682054e-05,
      "loss": 0.2007,
      "step": 4491
    },
    {
      "epoch": 0.18162887785943171,
      "grad_norm": 3.009795665740967,
      "learning_rate": 9.235788643957662e-05,
      "loss": 0.1051,
      "step": 4492
    },
    {
      "epoch": 0.18166931171469872,
      "grad_norm": 8.781230926513672,
      "learning_rate": 9.235449756325223e-05,
      "loss": 0.2537,
      "step": 4493
    },
    {
      "epoch": 0.18170974556996572,
      "grad_norm": 4.4113359451293945,
      "learning_rate": 9.23511079979025e-05,
      "loss": 0.2343,
      "step": 4494
    },
    {
      "epoch": 0.18175017942523275,
      "grad_norm": 1.6731832027435303,
      "learning_rate": 9.234771774358259e-05,
      "loss": 0.0452,
      "step": 4495
    },
    {
      "epoch": 0.18179061328049975,
      "grad_norm": 5.662269115447998,
      "learning_rate": 9.234432680034763e-05,
      "loss": 0.1223,
      "step": 4496
    },
    {
      "epoch": 0.18183104713576678,
      "grad_norm": 8.05603313446045,
      "learning_rate": 9.234093516825281e-05,
      "loss": 0.2555,
      "step": 4497
    },
    {
      "epoch": 0.18187148099103378,
      "grad_norm": 3.5211825370788574,
      "learning_rate": 9.233754284735327e-05,
      "loss": 0.1478,
      "step": 4498
    },
    {
      "epoch": 0.18191191484630082,
      "grad_norm": 6.514138221740723,
      "learning_rate": 9.233414983770423e-05,
      "loss": 0.2746,
      "step": 4499
    },
    {
      "epoch": 0.18195234870156782,
      "grad_norm": 7.850934028625488,
      "learning_rate": 9.233075613936086e-05,
      "loss": 0.2453,
      "step": 4500
    },
    {
      "epoch": 0.18199278255683485,
      "grad_norm": 4.616213321685791,
      "learning_rate": 9.232736175237839e-05,
      "loss": 0.0995,
      "step": 4501
    },
    {
      "epoch": 0.18203321641210185,
      "grad_norm": 5.7273077964782715,
      "learning_rate": 9.232396667681205e-05,
      "loss": 0.2667,
      "step": 4502
    },
    {
      "epoch": 0.18207365026736888,
      "grad_norm": 3.7441630363464355,
      "learning_rate": 9.232057091271701e-05,
      "loss": 0.2396,
      "step": 4503
    },
    {
      "epoch": 0.18211408412263588,
      "grad_norm": 8.691762924194336,
      "learning_rate": 9.23171744601486e-05,
      "loss": 0.4902,
      "step": 4504
    },
    {
      "epoch": 0.18215451797790289,
      "grad_norm": 3.4395346641540527,
      "learning_rate": 9.231377731916202e-05,
      "loss": 0.1511,
      "step": 4505
    },
    {
      "epoch": 0.18219495183316992,
      "grad_norm": 2.6527535915374756,
      "learning_rate": 9.231037948981254e-05,
      "loss": 0.1285,
      "step": 4506
    },
    {
      "epoch": 0.18223538568843692,
      "grad_norm": 4.4770283699035645,
      "learning_rate": 9.230698097215543e-05,
      "loss": 0.2403,
      "step": 4507
    },
    {
      "epoch": 0.18227581954370395,
      "grad_norm": 4.379354000091553,
      "learning_rate": 9.2303581766246e-05,
      "loss": 0.131,
      "step": 4508
    },
    {
      "epoch": 0.18231625339897095,
      "grad_norm": 10.44453239440918,
      "learning_rate": 9.230018187213953e-05,
      "loss": 0.405,
      "step": 4509
    },
    {
      "epoch": 0.18235668725423798,
      "grad_norm": 6.7079877853393555,
      "learning_rate": 9.229678128989134e-05,
      "loss": 0.1139,
      "step": 4510
    },
    {
      "epoch": 0.18239712110950498,
      "grad_norm": 14.319596290588379,
      "learning_rate": 9.229338001955674e-05,
      "loss": 0.3394,
      "step": 4511
    },
    {
      "epoch": 0.18243755496477201,
      "grad_norm": 8.255341529846191,
      "learning_rate": 9.228997806119109e-05,
      "loss": 0.2688,
      "step": 4512
    },
    {
      "epoch": 0.18247798882003902,
      "grad_norm": 7.323184013366699,
      "learning_rate": 9.228657541484968e-05,
      "loss": 0.1781,
      "step": 4513
    },
    {
      "epoch": 0.18251842267530605,
      "grad_norm": 9.519299507141113,
      "learning_rate": 9.228317208058792e-05,
      "loss": 0.2542,
      "step": 4514
    },
    {
      "epoch": 0.18255885653057305,
      "grad_norm": 9.309796333312988,
      "learning_rate": 9.227976805846115e-05,
      "loss": 0.2529,
      "step": 4515
    },
    {
      "epoch": 0.18259929038584005,
      "grad_norm": 4.288893699645996,
      "learning_rate": 9.227636334852475e-05,
      "loss": 0.2094,
      "step": 4516
    },
    {
      "epoch": 0.18263972424110708,
      "grad_norm": 4.25059700012207,
      "learning_rate": 9.227295795083411e-05,
      "loss": 0.2017,
      "step": 4517
    },
    {
      "epoch": 0.18268015809637408,
      "grad_norm": 6.7985734939575195,
      "learning_rate": 9.226955186544461e-05,
      "loss": 0.2589,
      "step": 4518
    },
    {
      "epoch": 0.18272059195164111,
      "grad_norm": 3.894500970840454,
      "learning_rate": 9.226614509241169e-05,
      "loss": 0.2097,
      "step": 4519
    },
    {
      "epoch": 0.18276102580690812,
      "grad_norm": 4.494843006134033,
      "learning_rate": 9.226273763179076e-05,
      "loss": 0.1389,
      "step": 4520
    },
    {
      "epoch": 0.18280145966217515,
      "grad_norm": 9.281052589416504,
      "learning_rate": 9.225932948363724e-05,
      "loss": 0.3235,
      "step": 4521
    },
    {
      "epoch": 0.18284189351744215,
      "grad_norm": 3.2094805240631104,
      "learning_rate": 9.22559206480066e-05,
      "loss": 0.1135,
      "step": 4522
    },
    {
      "epoch": 0.18288232737270918,
      "grad_norm": 5.420346736907959,
      "learning_rate": 9.225251112495427e-05,
      "loss": 0.1109,
      "step": 4523
    },
    {
      "epoch": 0.18292276122797618,
      "grad_norm": 2.5231359004974365,
      "learning_rate": 9.224910091453574e-05,
      "loss": 0.1436,
      "step": 4524
    },
    {
      "epoch": 0.1829631950832432,
      "grad_norm": 2.869990825653076,
      "learning_rate": 9.224569001680645e-05,
      "loss": 0.0677,
      "step": 4525
    },
    {
      "epoch": 0.18300362893851022,
      "grad_norm": 4.309022426605225,
      "learning_rate": 9.224227843182194e-05,
      "loss": 0.215,
      "step": 4526
    },
    {
      "epoch": 0.18304406279377722,
      "grad_norm": 3.0023324489593506,
      "learning_rate": 9.223886615963766e-05,
      "loss": 0.1733,
      "step": 4527
    },
    {
      "epoch": 0.18308449664904425,
      "grad_norm": 9.868016242980957,
      "learning_rate": 9.223545320030915e-05,
      "loss": 0.2129,
      "step": 4528
    },
    {
      "epoch": 0.18312493050431125,
      "grad_norm": 3.653777837753296,
      "learning_rate": 9.223203955389193e-05,
      "loss": 0.1554,
      "step": 4529
    },
    {
      "epoch": 0.18316536435957828,
      "grad_norm": 2.9675493240356445,
      "learning_rate": 9.222862522044154e-05,
      "loss": 0.1451,
      "step": 4530
    },
    {
      "epoch": 0.18320579821484528,
      "grad_norm": 6.437636375427246,
      "learning_rate": 9.222521020001349e-05,
      "loss": 0.1638,
      "step": 4531
    },
    {
      "epoch": 0.1832462320701123,
      "grad_norm": 7.919242858886719,
      "learning_rate": 9.222179449266337e-05,
      "loss": 0.3082,
      "step": 4532
    },
    {
      "epoch": 0.18328666592537932,
      "grad_norm": 5.665343284606934,
      "learning_rate": 9.221837809844673e-05,
      "loss": 0.2367,
      "step": 4533
    },
    {
      "epoch": 0.18332709978064635,
      "grad_norm": 8.419405937194824,
      "learning_rate": 9.221496101741918e-05,
      "loss": 0.2273,
      "step": 4534
    },
    {
      "epoch": 0.18336753363591335,
      "grad_norm": 2.2848074436187744,
      "learning_rate": 9.221154324963625e-05,
      "loss": 0.0978,
      "step": 4535
    },
    {
      "epoch": 0.18340796749118035,
      "grad_norm": 3.359976053237915,
      "learning_rate": 9.220812479515359e-05,
      "loss": 0.1048,
      "step": 4536
    },
    {
      "epoch": 0.18344840134644738,
      "grad_norm": 3.1032097339630127,
      "learning_rate": 9.220470565402682e-05,
      "loss": 0.227,
      "step": 4537
    },
    {
      "epoch": 0.18348883520171438,
      "grad_norm": 6.265812397003174,
      "learning_rate": 9.22012858263115e-05,
      "loss": 0.1931,
      "step": 4538
    },
    {
      "epoch": 0.18352926905698141,
      "grad_norm": 8.832064628601074,
      "learning_rate": 9.219786531206332e-05,
      "loss": 0.2083,
      "step": 4539
    },
    {
      "epoch": 0.18356970291224842,
      "grad_norm": 4.0390472412109375,
      "learning_rate": 9.21944441113379e-05,
      "loss": 0.091,
      "step": 4540
    },
    {
      "epoch": 0.18361013676751545,
      "grad_norm": 3.15034818649292,
      "learning_rate": 9.219102222419091e-05,
      "loss": 0.1896,
      "step": 4541
    },
    {
      "epoch": 0.18365057062278245,
      "grad_norm": 5.715325832366943,
      "learning_rate": 9.218759965067802e-05,
      "loss": 0.1443,
      "step": 4542
    },
    {
      "epoch": 0.18369100447804948,
      "grad_norm": 3.6684372425079346,
      "learning_rate": 9.21841763908549e-05,
      "loss": 0.2793,
      "step": 4543
    },
    {
      "epoch": 0.18373143833331648,
      "grad_norm": 4.460073471069336,
      "learning_rate": 9.218075244477722e-05,
      "loss": 0.1121,
      "step": 4544
    },
    {
      "epoch": 0.1837718721885835,
      "grad_norm": 4.779770374298096,
      "learning_rate": 9.217732781250072e-05,
      "loss": 0.2153,
      "step": 4545
    },
    {
      "epoch": 0.18381230604385052,
      "grad_norm": 8.943330764770508,
      "learning_rate": 9.217390249408108e-05,
      "loss": 0.1687,
      "step": 4546
    },
    {
      "epoch": 0.18385273989911752,
      "grad_norm": 3.1600182056427,
      "learning_rate": 9.217047648957404e-05,
      "loss": 0.0906,
      "step": 4547
    },
    {
      "epoch": 0.18389317375438455,
      "grad_norm": 4.624349594116211,
      "learning_rate": 9.216704979903534e-05,
      "loss": 0.1303,
      "step": 4548
    },
    {
      "epoch": 0.18393360760965155,
      "grad_norm": 5.40071964263916,
      "learning_rate": 9.21636224225207e-05,
      "loss": 0.1463,
      "step": 4549
    },
    {
      "epoch": 0.18397404146491858,
      "grad_norm": 3.986436128616333,
      "learning_rate": 9.21601943600859e-05,
      "loss": 0.1377,
      "step": 4550
    },
    {
      "epoch": 0.18401447532018558,
      "grad_norm": 8.610089302062988,
      "learning_rate": 9.21567656117867e-05,
      "loss": 0.3714,
      "step": 4551
    },
    {
      "epoch": 0.1840549091754526,
      "grad_norm": 2.8590712547302246,
      "learning_rate": 9.21533361776789e-05,
      "loss": 0.1522,
      "step": 4552
    },
    {
      "epoch": 0.18409534303071962,
      "grad_norm": 7.371978282928467,
      "learning_rate": 9.214990605781824e-05,
      "loss": 0.1517,
      "step": 4553
    },
    {
      "epoch": 0.18413577688598665,
      "grad_norm": 6.478354454040527,
      "learning_rate": 9.214647525226057e-05,
      "loss": 0.2346,
      "step": 4554
    },
    {
      "epoch": 0.18417621074125365,
      "grad_norm": 7.209380626678467,
      "learning_rate": 9.214304376106167e-05,
      "loss": 0.2234,
      "step": 4555
    },
    {
      "epoch": 0.18421664459652068,
      "grad_norm": 3.9611012935638428,
      "learning_rate": 9.21396115842774e-05,
      "loss": 0.1252,
      "step": 4556
    },
    {
      "epoch": 0.18425707845178768,
      "grad_norm": 7.62563943862915,
      "learning_rate": 9.213617872196355e-05,
      "loss": 0.2753,
      "step": 4557
    },
    {
      "epoch": 0.18429751230705468,
      "grad_norm": 6.4009246826171875,
      "learning_rate": 9.213274517417599e-05,
      "loss": 0.1419,
      "step": 4558
    },
    {
      "epoch": 0.18433794616232171,
      "grad_norm": 7.668725967407227,
      "learning_rate": 9.212931094097059e-05,
      "loss": 0.2028,
      "step": 4559
    },
    {
      "epoch": 0.18437838001758872,
      "grad_norm": 4.277970314025879,
      "learning_rate": 9.21258760224032e-05,
      "loss": 0.1909,
      "step": 4560
    },
    {
      "epoch": 0.18441881387285575,
      "grad_norm": 4.098956108093262,
      "learning_rate": 9.21224404185297e-05,
      "loss": 0.1082,
      "step": 4561
    },
    {
      "epoch": 0.18445924772812275,
      "grad_norm": 7.367994785308838,
      "learning_rate": 9.211900412940598e-05,
      "loss": 0.2155,
      "step": 4562
    },
    {
      "epoch": 0.18449968158338978,
      "grad_norm": 4.233806610107422,
      "learning_rate": 9.211556715508796e-05,
      "loss": 0.1384,
      "step": 4563
    },
    {
      "epoch": 0.18454011543865678,
      "grad_norm": 7.973733901977539,
      "learning_rate": 9.211212949563153e-05,
      "loss": 0.3127,
      "step": 4564
    },
    {
      "epoch": 0.1845805492939238,
      "grad_norm": 7.060271263122559,
      "learning_rate": 9.210869115109262e-05,
      "loss": 0.1902,
      "step": 4565
    },
    {
      "epoch": 0.18462098314919081,
      "grad_norm": 5.4183526039123535,
      "learning_rate": 9.210525212152716e-05,
      "loss": 0.1868,
      "step": 4566
    },
    {
      "epoch": 0.18466141700445785,
      "grad_norm": 3.387714385986328,
      "learning_rate": 9.210181240699112e-05,
      "loss": 0.3053,
      "step": 4567
    },
    {
      "epoch": 0.18470185085972485,
      "grad_norm": 5.232081890106201,
      "learning_rate": 9.209837200754044e-05,
      "loss": 0.2344,
      "step": 4568
    },
    {
      "epoch": 0.18474228471499185,
      "grad_norm": 1.9767042398452759,
      "learning_rate": 9.209493092323108e-05,
      "loss": 0.0721,
      "step": 4569
    },
    {
      "epoch": 0.18478271857025888,
      "grad_norm": 5.834506988525391,
      "learning_rate": 9.209148915411905e-05,
      "loss": 0.1741,
      "step": 4570
    },
    {
      "epoch": 0.18482315242552588,
      "grad_norm": 5.504905700683594,
      "learning_rate": 9.20880467002603e-05,
      "loss": 0.1939,
      "step": 4571
    },
    {
      "epoch": 0.1848635862807929,
      "grad_norm": 6.289308547973633,
      "learning_rate": 9.208460356171087e-05,
      "loss": 0.1631,
      "step": 4572
    },
    {
      "epoch": 0.18490402013605992,
      "grad_norm": 7.8194990158081055,
      "learning_rate": 9.208115973852673e-05,
      "loss": 0.2335,
      "step": 4573
    },
    {
      "epoch": 0.18494445399132695,
      "grad_norm": 5.219015598297119,
      "learning_rate": 9.207771523076396e-05,
      "loss": 0.293,
      "step": 4574
    },
    {
      "epoch": 0.18498488784659395,
      "grad_norm": 5.754721164703369,
      "learning_rate": 9.207427003847854e-05,
      "loss": 0.1446,
      "step": 4575
    },
    {
      "epoch": 0.18502532170186098,
      "grad_norm": 2.7232019901275635,
      "learning_rate": 9.207082416172657e-05,
      "loss": 0.2373,
      "step": 4576
    },
    {
      "epoch": 0.18506575555712798,
      "grad_norm": 2.5786547660827637,
      "learning_rate": 9.206737760056407e-05,
      "loss": 0.1352,
      "step": 4577
    },
    {
      "epoch": 0.185106189412395,
      "grad_norm": 4.983080863952637,
      "learning_rate": 9.206393035504711e-05,
      "loss": 0.1864,
      "step": 4578
    },
    {
      "epoch": 0.185146623267662,
      "grad_norm": 4.074652671813965,
      "learning_rate": 9.206048242523178e-05,
      "loss": 0.1572,
      "step": 4579
    },
    {
      "epoch": 0.18518705712292902,
      "grad_norm": 6.227214813232422,
      "learning_rate": 9.205703381117419e-05,
      "loss": 0.2294,
      "step": 4580
    },
    {
      "epoch": 0.18522749097819605,
      "grad_norm": 4.419712543487549,
      "learning_rate": 9.20535845129304e-05,
      "loss": 0.1132,
      "step": 4581
    },
    {
      "epoch": 0.18526792483346305,
      "grad_norm": 5.612473011016846,
      "learning_rate": 9.205013453055655e-05,
      "loss": 0.1251,
      "step": 4582
    },
    {
      "epoch": 0.18530835868873008,
      "grad_norm": 2.775691270828247,
      "learning_rate": 9.204668386410878e-05,
      "loss": 0.188,
      "step": 4583
    },
    {
      "epoch": 0.18534879254399708,
      "grad_norm": 3.5505058765411377,
      "learning_rate": 9.204323251364317e-05,
      "loss": 0.1835,
      "step": 4584
    },
    {
      "epoch": 0.1853892263992641,
      "grad_norm": 5.83310604095459,
      "learning_rate": 9.203978047921593e-05,
      "loss": 0.1319,
      "step": 4585
    },
    {
      "epoch": 0.18542966025453111,
      "grad_norm": 2.6768722534179688,
      "learning_rate": 9.203632776088318e-05,
      "loss": 0.1085,
      "step": 4586
    },
    {
      "epoch": 0.18547009410979814,
      "grad_norm": 4.479236602783203,
      "learning_rate": 9.203287435870111e-05,
      "loss": 0.1831,
      "step": 4587
    },
    {
      "epoch": 0.18551052796506515,
      "grad_norm": 6.296863079071045,
      "learning_rate": 9.202942027272589e-05,
      "loss": 0.1629,
      "step": 4588
    },
    {
      "epoch": 0.18555096182033218,
      "grad_norm": 2.846381902694702,
      "learning_rate": 9.20259655030137e-05,
      "loss": 0.1109,
      "step": 4589
    },
    {
      "epoch": 0.18559139567559918,
      "grad_norm": 3.2022156715393066,
      "learning_rate": 9.202251004962076e-05,
      "loss": 0.1653,
      "step": 4590
    },
    {
      "epoch": 0.18563182953086618,
      "grad_norm": 1.627022624015808,
      "learning_rate": 9.201905391260328e-05,
      "loss": 0.0701,
      "step": 4591
    },
    {
      "epoch": 0.1856722633861332,
      "grad_norm": 4.857787609100342,
      "learning_rate": 9.201559709201746e-05,
      "loss": 0.1398,
      "step": 4592
    },
    {
      "epoch": 0.18571269724140022,
      "grad_norm": 2.14768123626709,
      "learning_rate": 9.201213958791959e-05,
      "loss": 0.1195,
      "step": 4593
    },
    {
      "epoch": 0.18575313109666725,
      "grad_norm": 3.3980259895324707,
      "learning_rate": 9.200868140036586e-05,
      "loss": 0.161,
      "step": 4594
    },
    {
      "epoch": 0.18579356495193425,
      "grad_norm": 4.222605228424072,
      "learning_rate": 9.200522252941256e-05,
      "loss": 0.1524,
      "step": 4595
    },
    {
      "epoch": 0.18583399880720128,
      "grad_norm": 5.2335991859436035,
      "learning_rate": 9.200176297511596e-05,
      "loss": 0.2796,
      "step": 4596
    },
    {
      "epoch": 0.18587443266246828,
      "grad_norm": 5.923531532287598,
      "learning_rate": 9.199830273753232e-05,
      "loss": 0.4262,
      "step": 4597
    },
    {
      "epoch": 0.1859148665177353,
      "grad_norm": 5.154194355010986,
      "learning_rate": 9.199484181671794e-05,
      "loss": 0.4344,
      "step": 4598
    },
    {
      "epoch": 0.1859553003730023,
      "grad_norm": 7.00412654876709,
      "learning_rate": 9.199138021272913e-05,
      "loss": 0.1387,
      "step": 4599
    },
    {
      "epoch": 0.18599573422826934,
      "grad_norm": 2.7589285373687744,
      "learning_rate": 9.19879179256222e-05,
      "loss": 0.143,
      "step": 4600
    },
    {
      "epoch": 0.18603616808353635,
      "grad_norm": 5.244170665740967,
      "learning_rate": 9.198445495545348e-05,
      "loss": 0.2852,
      "step": 4601
    },
    {
      "epoch": 0.18607660193880335,
      "grad_norm": 5.720447540283203,
      "learning_rate": 9.19809913022793e-05,
      "loss": 0.175,
      "step": 4602
    },
    {
      "epoch": 0.18611703579407038,
      "grad_norm": 7.26362943649292,
      "learning_rate": 9.1977526966156e-05,
      "loss": 0.2331,
      "step": 4603
    },
    {
      "epoch": 0.18615746964933738,
      "grad_norm": 5.362763404846191,
      "learning_rate": 9.197406194713993e-05,
      "loss": 0.1382,
      "step": 4604
    },
    {
      "epoch": 0.1861979035046044,
      "grad_norm": 3.9796645641326904,
      "learning_rate": 9.197059624528749e-05,
      "loss": 0.2783,
      "step": 4605
    },
    {
      "epoch": 0.18623833735987141,
      "grad_norm": 2.6818222999572754,
      "learning_rate": 9.196712986065505e-05,
      "loss": 0.2065,
      "step": 4606
    },
    {
      "epoch": 0.18627877121513844,
      "grad_norm": 4.149332046508789,
      "learning_rate": 9.196366279329898e-05,
      "loss": 0.1266,
      "step": 4607
    },
    {
      "epoch": 0.18631920507040545,
      "grad_norm": 3.2332754135131836,
      "learning_rate": 9.196019504327571e-05,
      "loss": 0.1246,
      "step": 4608
    },
    {
      "epoch": 0.18635963892567248,
      "grad_norm": 7.564556121826172,
      "learning_rate": 9.195672661064163e-05,
      "loss": 0.2664,
      "step": 4609
    },
    {
      "epoch": 0.18640007278093948,
      "grad_norm": 6.0207929611206055,
      "learning_rate": 9.195325749545319e-05,
      "loss": 0.2735,
      "step": 4610
    },
    {
      "epoch": 0.18644050663620648,
      "grad_norm": 7.2770562171936035,
      "learning_rate": 9.194978769776679e-05,
      "loss": 0.1748,
      "step": 4611
    },
    {
      "epoch": 0.1864809404914735,
      "grad_norm": 4.216729164123535,
      "learning_rate": 9.19463172176389e-05,
      "loss": 0.2118,
      "step": 4612
    },
    {
      "epoch": 0.18652137434674051,
      "grad_norm": 2.464998483657837,
      "learning_rate": 9.194284605512599e-05,
      "loss": 0.1354,
      "step": 4613
    },
    {
      "epoch": 0.18656180820200755,
      "grad_norm": 5.366413116455078,
      "learning_rate": 9.19393742102845e-05,
      "loss": 0.2743,
      "step": 4614
    },
    {
      "epoch": 0.18660224205727455,
      "grad_norm": 3.067521572113037,
      "learning_rate": 9.193590168317095e-05,
      "loss": 0.2495,
      "step": 4615
    },
    {
      "epoch": 0.18664267591254158,
      "grad_norm": 5.871858596801758,
      "learning_rate": 9.19324284738418e-05,
      "loss": 0.2675,
      "step": 4616
    },
    {
      "epoch": 0.18668310976780858,
      "grad_norm": 5.901576042175293,
      "learning_rate": 9.192895458235356e-05,
      "loss": 0.326,
      "step": 4617
    },
    {
      "epoch": 0.1867235436230756,
      "grad_norm": 5.371984481811523,
      "learning_rate": 9.192548000876273e-05,
      "loss": 0.1778,
      "step": 4618
    },
    {
      "epoch": 0.1867639774783426,
      "grad_norm": 4.044281482696533,
      "learning_rate": 9.192200475312585e-05,
      "loss": 0.2294,
      "step": 4619
    },
    {
      "epoch": 0.18680441133360964,
      "grad_norm": 2.7497456073760986,
      "learning_rate": 9.191852881549944e-05,
      "loss": 0.2013,
      "step": 4620
    },
    {
      "epoch": 0.18684484518887665,
      "grad_norm": 9.951253890991211,
      "learning_rate": 9.19150521959401e-05,
      "loss": 0.3231,
      "step": 4621
    },
    {
      "epoch": 0.18688527904414365,
      "grad_norm": 2.8104376792907715,
      "learning_rate": 9.191157489450431e-05,
      "loss": 0.1684,
      "step": 4622
    },
    {
      "epoch": 0.18692571289941068,
      "grad_norm": 2.471323251724243,
      "learning_rate": 9.190809691124868e-05,
      "loss": 0.0933,
      "step": 4623
    },
    {
      "epoch": 0.18696614675467768,
      "grad_norm": 8.212456703186035,
      "learning_rate": 9.19046182462298e-05,
      "loss": 0.1362,
      "step": 4624
    },
    {
      "epoch": 0.1870065806099447,
      "grad_norm": 5.484219551086426,
      "learning_rate": 9.190113889950424e-05,
      "loss": 0.1825,
      "step": 4625
    },
    {
      "epoch": 0.18704701446521171,
      "grad_norm": 5.478992938995361,
      "learning_rate": 9.189765887112863e-05,
      "loss": 0.1584,
      "step": 4626
    },
    {
      "epoch": 0.18708744832047874,
      "grad_norm": 5.583497047424316,
      "learning_rate": 9.189417816115953e-05,
      "loss": 0.1922,
      "step": 4627
    },
    {
      "epoch": 0.18712788217574575,
      "grad_norm": 3.3540847301483154,
      "learning_rate": 9.189069676965361e-05,
      "loss": 0.1347,
      "step": 4628
    },
    {
      "epoch": 0.18716831603101278,
      "grad_norm": 6.825102806091309,
      "learning_rate": 9.18872146966675e-05,
      "loss": 0.1178,
      "step": 4629
    },
    {
      "epoch": 0.18720874988627978,
      "grad_norm": 5.160303592681885,
      "learning_rate": 9.188373194225784e-05,
      "loss": 0.3342,
      "step": 4630
    },
    {
      "epoch": 0.1872491837415468,
      "grad_norm": 7.485423564910889,
      "learning_rate": 9.188024850648127e-05,
      "loss": 0.2576,
      "step": 4631
    },
    {
      "epoch": 0.1872896175968138,
      "grad_norm": 2.7125022411346436,
      "learning_rate": 9.18767643893945e-05,
      "loss": 0.129,
      "step": 4632
    },
    {
      "epoch": 0.18733005145208081,
      "grad_norm": 4.157063007354736,
      "learning_rate": 9.187327959105416e-05,
      "loss": 0.2251,
      "step": 4633
    },
    {
      "epoch": 0.18737048530734784,
      "grad_norm": 4.714603424072266,
      "learning_rate": 9.186979411151698e-05,
      "loss": 0.2997,
      "step": 4634
    },
    {
      "epoch": 0.18741091916261485,
      "grad_norm": 5.750052452087402,
      "learning_rate": 9.186630795083966e-05,
      "loss": 0.161,
      "step": 4635
    },
    {
      "epoch": 0.18745135301788188,
      "grad_norm": 7.260510444641113,
      "learning_rate": 9.186282110907888e-05,
      "loss": 0.1859,
      "step": 4636
    },
    {
      "epoch": 0.18749178687314888,
      "grad_norm": 4.588505268096924,
      "learning_rate": 9.185933358629139e-05,
      "loss": 0.1468,
      "step": 4637
    },
    {
      "epoch": 0.1875322207284159,
      "grad_norm": 4.599420070648193,
      "learning_rate": 9.185584538253394e-05,
      "loss": 0.2108,
      "step": 4638
    },
    {
      "epoch": 0.1875726545836829,
      "grad_norm": 5.705533981323242,
      "learning_rate": 9.185235649786324e-05,
      "loss": 0.1363,
      "step": 4639
    },
    {
      "epoch": 0.18761308843894994,
      "grad_norm": 5.66126823425293,
      "learning_rate": 9.184886693233605e-05,
      "loss": 0.1653,
      "step": 4640
    },
    {
      "epoch": 0.18765352229421695,
      "grad_norm": 3.0097427368164062,
      "learning_rate": 9.184537668600918e-05,
      "loss": 0.0906,
      "step": 4641
    },
    {
      "epoch": 0.18769395614948398,
      "grad_norm": 3.797877073287964,
      "learning_rate": 9.184188575893937e-05,
      "loss": 0.1916,
      "step": 4642
    },
    {
      "epoch": 0.18773439000475098,
      "grad_norm": 3.9525673389434814,
      "learning_rate": 9.183839415118342e-05,
      "loss": 0.1621,
      "step": 4643
    },
    {
      "epoch": 0.18777482386001798,
      "grad_norm": 2.698883056640625,
      "learning_rate": 9.183490186279813e-05,
      "loss": 0.1252,
      "step": 4644
    },
    {
      "epoch": 0.187815257715285,
      "grad_norm": 3.584003210067749,
      "learning_rate": 9.183140889384035e-05,
      "loss": 0.1239,
      "step": 4645
    },
    {
      "epoch": 0.187855691570552,
      "grad_norm": 6.152257919311523,
      "learning_rate": 9.182791524436684e-05,
      "loss": 0.1847,
      "step": 4646
    },
    {
      "epoch": 0.18789612542581904,
      "grad_norm": 5.768726348876953,
      "learning_rate": 9.182442091443447e-05,
      "loss": 0.195,
      "step": 4647
    },
    {
      "epoch": 0.18793655928108605,
      "grad_norm": 3.286801815032959,
      "learning_rate": 9.18209259041001e-05,
      "loss": 0.1247,
      "step": 4648
    },
    {
      "epoch": 0.18797699313635308,
      "grad_norm": 6.596505165100098,
      "learning_rate": 9.181743021342056e-05,
      "loss": 0.2516,
      "step": 4649
    },
    {
      "epoch": 0.18801742699162008,
      "grad_norm": 5.247797012329102,
      "learning_rate": 9.18139338424527e-05,
      "loss": 0.1389,
      "step": 4650
    },
    {
      "epoch": 0.1880578608468871,
      "grad_norm": 4.079026699066162,
      "learning_rate": 9.181043679125345e-05,
      "loss": 0.1849,
      "step": 4651
    },
    {
      "epoch": 0.1880982947021541,
      "grad_norm": 7.057977676391602,
      "learning_rate": 9.180693905987968e-05,
      "loss": 0.2593,
      "step": 4652
    },
    {
      "epoch": 0.18813872855742114,
      "grad_norm": 3.144850015640259,
      "learning_rate": 9.180344064838829e-05,
      "loss": 0.1469,
      "step": 4653
    },
    {
      "epoch": 0.18817916241268814,
      "grad_norm": 7.766226768493652,
      "learning_rate": 9.179994155683618e-05,
      "loss": 0.2749,
      "step": 4654
    },
    {
      "epoch": 0.18821959626795515,
      "grad_norm": 6.198915958404541,
      "learning_rate": 9.179644178528029e-05,
      "loss": 0.1895,
      "step": 4655
    },
    {
      "epoch": 0.18826003012322218,
      "grad_norm": 7.816349983215332,
      "learning_rate": 9.179294133377755e-05,
      "loss": 0.176,
      "step": 4656
    },
    {
      "epoch": 0.18830046397848918,
      "grad_norm": 6.112473011016846,
      "learning_rate": 9.178944020238489e-05,
      "loss": 0.1718,
      "step": 4657
    },
    {
      "epoch": 0.1883408978337562,
      "grad_norm": 6.571780204772949,
      "learning_rate": 9.17859383911593e-05,
      "loss": 0.1272,
      "step": 4658
    },
    {
      "epoch": 0.1883813316890232,
      "grad_norm": 3.6545677185058594,
      "learning_rate": 9.178243590015771e-05,
      "loss": 0.155,
      "step": 4659
    },
    {
      "epoch": 0.18842176554429024,
      "grad_norm": 2.4122633934020996,
      "learning_rate": 9.177893272943713e-05,
      "loss": 0.1044,
      "step": 4660
    },
    {
      "epoch": 0.18846219939955725,
      "grad_norm": 7.017280101776123,
      "learning_rate": 9.177542887905453e-05,
      "loss": 0.3566,
      "step": 4661
    },
    {
      "epoch": 0.18850263325482428,
      "grad_norm": 5.352670192718506,
      "learning_rate": 9.177192434906692e-05,
      "loss": 0.2232,
      "step": 4662
    },
    {
      "epoch": 0.18854306711009128,
      "grad_norm": 5.5989556312561035,
      "learning_rate": 9.17684191395313e-05,
      "loss": 0.265,
      "step": 4663
    },
    {
      "epoch": 0.1885835009653583,
      "grad_norm": 4.652342796325684,
      "learning_rate": 9.176491325050472e-05,
      "loss": 0.2091,
      "step": 4664
    },
    {
      "epoch": 0.1886239348206253,
      "grad_norm": 4.629335880279541,
      "learning_rate": 9.176140668204418e-05,
      "loss": 0.1479,
      "step": 4665
    },
    {
      "epoch": 0.1886643686758923,
      "grad_norm": 2.7875304222106934,
      "learning_rate": 9.175789943420674e-05,
      "loss": 0.1325,
      "step": 4666
    },
    {
      "epoch": 0.18870480253115934,
      "grad_norm": 2.496776819229126,
      "learning_rate": 9.175439150704948e-05,
      "loss": 0.1174,
      "step": 4667
    },
    {
      "epoch": 0.18874523638642635,
      "grad_norm": 5.965044021606445,
      "learning_rate": 9.175088290062942e-05,
      "loss": 0.2351,
      "step": 4668
    },
    {
      "epoch": 0.18878567024169338,
      "grad_norm": 7.272237300872803,
      "learning_rate": 9.174737361500367e-05,
      "loss": 0.1405,
      "step": 4669
    },
    {
      "epoch": 0.18882610409696038,
      "grad_norm": 3.358297109603882,
      "learning_rate": 9.174386365022931e-05,
      "loss": 0.2281,
      "step": 4670
    },
    {
      "epoch": 0.1888665379522274,
      "grad_norm": 5.162563800811768,
      "learning_rate": 9.174035300636345e-05,
      "loss": 0.1084,
      "step": 4671
    },
    {
      "epoch": 0.1889069718074944,
      "grad_norm": 3.886901378631592,
      "learning_rate": 9.173684168346319e-05,
      "loss": 0.1331,
      "step": 4672
    },
    {
      "epoch": 0.18894740566276144,
      "grad_norm": 6.3150715827941895,
      "learning_rate": 9.173332968158566e-05,
      "loss": 0.3174,
      "step": 4673
    },
    {
      "epoch": 0.18898783951802844,
      "grad_norm": 3.852797269821167,
      "learning_rate": 9.172981700078797e-05,
      "loss": 0.171,
      "step": 4674
    },
    {
      "epoch": 0.18902827337329545,
      "grad_norm": 6.688466548919678,
      "learning_rate": 9.17263036411273e-05,
      "loss": 0.1967,
      "step": 4675
    },
    {
      "epoch": 0.18906870722856248,
      "grad_norm": 3.723714828491211,
      "learning_rate": 9.172278960266079e-05,
      "loss": 0.133,
      "step": 4676
    },
    {
      "epoch": 0.18910914108382948,
      "grad_norm": 8.443169593811035,
      "learning_rate": 9.17192748854456e-05,
      "loss": 0.2881,
      "step": 4677
    },
    {
      "epoch": 0.1891495749390965,
      "grad_norm": 7.1569342613220215,
      "learning_rate": 9.171575948953893e-05,
      "loss": 0.1783,
      "step": 4678
    },
    {
      "epoch": 0.1891900087943635,
      "grad_norm": 6.1377034187316895,
      "learning_rate": 9.171224341499793e-05,
      "loss": 0.2061,
      "step": 4679
    },
    {
      "epoch": 0.18923044264963054,
      "grad_norm": 7.8028998374938965,
      "learning_rate": 9.170872666187984e-05,
      "loss": 0.2101,
      "step": 4680
    },
    {
      "epoch": 0.18927087650489755,
      "grad_norm": 4.363171577453613,
      "learning_rate": 9.170520923024185e-05,
      "loss": 0.1265,
      "step": 4681
    },
    {
      "epoch": 0.18931131036016458,
      "grad_norm": 10.088348388671875,
      "learning_rate": 9.170169112014116e-05,
      "loss": 0.3016,
      "step": 4682
    },
    {
      "epoch": 0.18935174421543158,
      "grad_norm": 4.284903049468994,
      "learning_rate": 9.169817233163505e-05,
      "loss": 0.1547,
      "step": 4683
    },
    {
      "epoch": 0.1893921780706986,
      "grad_norm": 5.171776294708252,
      "learning_rate": 9.169465286478074e-05,
      "loss": 0.1182,
      "step": 4684
    },
    {
      "epoch": 0.1894326119259656,
      "grad_norm": 2.444493055343628,
      "learning_rate": 9.169113271963548e-05,
      "loss": 0.1291,
      "step": 4685
    },
    {
      "epoch": 0.1894730457812326,
      "grad_norm": 7.796009063720703,
      "learning_rate": 9.168761189625654e-05,
      "loss": 0.1879,
      "step": 4686
    },
    {
      "epoch": 0.18951347963649964,
      "grad_norm": 6.232384204864502,
      "learning_rate": 9.16840903947012e-05,
      "loss": 0.2141,
      "step": 4687
    },
    {
      "epoch": 0.18955391349176665,
      "grad_norm": 4.467564582824707,
      "learning_rate": 9.168056821502675e-05,
      "loss": 0.2438,
      "step": 4688
    },
    {
      "epoch": 0.18959434734703368,
      "grad_norm": 7.255736827850342,
      "learning_rate": 9.16770453572905e-05,
      "loss": 0.217,
      "step": 4689
    },
    {
      "epoch": 0.18963478120230068,
      "grad_norm": 4.385347843170166,
      "learning_rate": 9.167352182154973e-05,
      "loss": 0.1378,
      "step": 4690
    },
    {
      "epoch": 0.1896752150575677,
      "grad_norm": 2.305523157119751,
      "learning_rate": 9.166999760786178e-05,
      "loss": 0.1007,
      "step": 4691
    },
    {
      "epoch": 0.1897156489128347,
      "grad_norm": 7.299305438995361,
      "learning_rate": 9.166647271628396e-05,
      "loss": 0.2066,
      "step": 4692
    },
    {
      "epoch": 0.18975608276810174,
      "grad_norm": 5.245622634887695,
      "learning_rate": 9.166294714687367e-05,
      "loss": 0.1744,
      "step": 4693
    },
    {
      "epoch": 0.18979651662336874,
      "grad_norm": 1.6835108995437622,
      "learning_rate": 9.16594208996882e-05,
      "loss": 0.0851,
      "step": 4694
    },
    {
      "epoch": 0.18983695047863577,
      "grad_norm": 3.7930374145507812,
      "learning_rate": 9.165589397478496e-05,
      "loss": 0.1711,
      "step": 4695
    },
    {
      "epoch": 0.18987738433390278,
      "grad_norm": 6.918922424316406,
      "learning_rate": 9.165236637222128e-05,
      "loss": 0.184,
      "step": 4696
    },
    {
      "epoch": 0.18991781818916978,
      "grad_norm": 2.5388808250427246,
      "learning_rate": 9.16488380920546e-05,
      "loss": 0.0729,
      "step": 4697
    },
    {
      "epoch": 0.1899582520444368,
      "grad_norm": 5.509007930755615,
      "learning_rate": 9.164530913434228e-05,
      "loss": 0.0857,
      "step": 4698
    },
    {
      "epoch": 0.1899986858997038,
      "grad_norm": 2.6846494674682617,
      "learning_rate": 9.164177949914175e-05,
      "loss": 0.0954,
      "step": 4699
    },
    {
      "epoch": 0.19003911975497084,
      "grad_norm": 2.5614163875579834,
      "learning_rate": 9.163824918651042e-05,
      "loss": 0.1717,
      "step": 4700
    },
    {
      "epoch": 0.19007955361023784,
      "grad_norm": 5.750306606292725,
      "learning_rate": 9.163471819650572e-05,
      "loss": 0.1966,
      "step": 4701
    },
    {
      "epoch": 0.19011998746550487,
      "grad_norm": 5.76956033706665,
      "learning_rate": 9.16311865291851e-05,
      "loss": 0.2172,
      "step": 4702
    },
    {
      "epoch": 0.19016042132077188,
      "grad_norm": 2.575223684310913,
      "learning_rate": 9.162765418460602e-05,
      "loss": 0.1019,
      "step": 4703
    },
    {
      "epoch": 0.1902008551760389,
      "grad_norm": 3.838634967803955,
      "learning_rate": 9.162412116282592e-05,
      "loss": 0.3522,
      "step": 4704
    },
    {
      "epoch": 0.1902412890313059,
      "grad_norm": 3.165351390838623,
      "learning_rate": 9.162058746390229e-05,
      "loss": 0.1755,
      "step": 4705
    },
    {
      "epoch": 0.19028172288657294,
      "grad_norm": 4.404272556304932,
      "learning_rate": 9.161705308789263e-05,
      "loss": 0.1207,
      "step": 4706
    },
    {
      "epoch": 0.19032215674183994,
      "grad_norm": 9.25207233428955,
      "learning_rate": 9.161351803485441e-05,
      "loss": 0.2057,
      "step": 4707
    },
    {
      "epoch": 0.19036259059710695,
      "grad_norm": 6.416543960571289,
      "learning_rate": 9.160998230484515e-05,
      "loss": 0.1931,
      "step": 4708
    },
    {
      "epoch": 0.19040302445237398,
      "grad_norm": 2.9941041469573975,
      "learning_rate": 9.160644589792238e-05,
      "loss": 0.1401,
      "step": 4709
    },
    {
      "epoch": 0.19044345830764098,
      "grad_norm": 3.5025150775909424,
      "learning_rate": 9.160290881414362e-05,
      "loss": 0.2432,
      "step": 4710
    },
    {
      "epoch": 0.190483892162908,
      "grad_norm": 9.199921607971191,
      "learning_rate": 9.159937105356641e-05,
      "loss": 0.2122,
      "step": 4711
    },
    {
      "epoch": 0.190524326018175,
      "grad_norm": 7.5927228927612305,
      "learning_rate": 9.15958326162483e-05,
      "loss": 0.2237,
      "step": 4712
    },
    {
      "epoch": 0.19056475987344204,
      "grad_norm": 3.5495147705078125,
      "learning_rate": 9.159229350224688e-05,
      "loss": 0.1351,
      "step": 4713
    },
    {
      "epoch": 0.19060519372870904,
      "grad_norm": 4.90360164642334,
      "learning_rate": 9.158875371161969e-05,
      "loss": 0.1469,
      "step": 4714
    },
    {
      "epoch": 0.19064562758397607,
      "grad_norm": 6.6075334548950195,
      "learning_rate": 9.158521324442433e-05,
      "loss": 0.2414,
      "step": 4715
    },
    {
      "epoch": 0.19068606143924308,
      "grad_norm": 4.333644390106201,
      "learning_rate": 9.15816721007184e-05,
      "loss": 0.185,
      "step": 4716
    },
    {
      "epoch": 0.1907264952945101,
      "grad_norm": 3.156139850616455,
      "learning_rate": 9.15781302805595e-05,
      "loss": 0.1157,
      "step": 4717
    },
    {
      "epoch": 0.1907669291497771,
      "grad_norm": 5.3605523109436035,
      "learning_rate": 9.157458778400525e-05,
      "loss": 0.1278,
      "step": 4718
    },
    {
      "epoch": 0.1908073630050441,
      "grad_norm": 3.0040528774261475,
      "learning_rate": 9.157104461111327e-05,
      "loss": 0.1589,
      "step": 4719
    },
    {
      "epoch": 0.19084779686031114,
      "grad_norm": 4.250317096710205,
      "learning_rate": 9.156750076194124e-05,
      "loss": 0.1608,
      "step": 4720
    },
    {
      "epoch": 0.19088823071557814,
      "grad_norm": 3.4318759441375732,
      "learning_rate": 9.156395623654676e-05,
      "loss": 0.1158,
      "step": 4721
    },
    {
      "epoch": 0.19092866457084517,
      "grad_norm": 7.284013271331787,
      "learning_rate": 9.156041103498754e-05,
      "loss": 0.2684,
      "step": 4722
    },
    {
      "epoch": 0.19096909842611218,
      "grad_norm": 4.2357330322265625,
      "learning_rate": 9.15568651573212e-05,
      "loss": 0.3704,
      "step": 4723
    },
    {
      "epoch": 0.1910095322813792,
      "grad_norm": 5.214235782623291,
      "learning_rate": 9.155331860360548e-05,
      "loss": 0.2627,
      "step": 4724
    },
    {
      "epoch": 0.1910499661366462,
      "grad_norm": 5.813052654266357,
      "learning_rate": 9.154977137389803e-05,
      "loss": 0.1459,
      "step": 4725
    },
    {
      "epoch": 0.19109039999191324,
      "grad_norm": 3.18023419380188,
      "learning_rate": 9.154622346825659e-05,
      "loss": 0.1789,
      "step": 4726
    },
    {
      "epoch": 0.19113083384718024,
      "grad_norm": 2.9963674545288086,
      "learning_rate": 9.154267488673886e-05,
      "loss": 0.1165,
      "step": 4727
    },
    {
      "epoch": 0.19117126770244727,
      "grad_norm": 3.2409894466400146,
      "learning_rate": 9.153912562940257e-05,
      "loss": 0.1912,
      "step": 4728
    },
    {
      "epoch": 0.19121170155771428,
      "grad_norm": 3.7208495140075684,
      "learning_rate": 9.153557569630547e-05,
      "loss": 0.1297,
      "step": 4729
    },
    {
      "epoch": 0.19125213541298128,
      "grad_norm": 5.700179100036621,
      "learning_rate": 9.153202508750529e-05,
      "loss": 0.3436,
      "step": 4730
    },
    {
      "epoch": 0.1912925692682483,
      "grad_norm": 4.016676902770996,
      "learning_rate": 9.15284738030598e-05,
      "loss": 0.1457,
      "step": 4731
    },
    {
      "epoch": 0.1913330031235153,
      "grad_norm": 2.5985050201416016,
      "learning_rate": 9.152492184302679e-05,
      "loss": 0.0714,
      "step": 4732
    },
    {
      "epoch": 0.19137343697878234,
      "grad_norm": 5.750226020812988,
      "learning_rate": 9.152136920746402e-05,
      "loss": 0.2466,
      "step": 4733
    },
    {
      "epoch": 0.19141387083404934,
      "grad_norm": 7.5742645263671875,
      "learning_rate": 9.151781589642929e-05,
      "loss": 0.316,
      "step": 4734
    },
    {
      "epoch": 0.19145430468931637,
      "grad_norm": 5.182015895843506,
      "learning_rate": 9.151426190998043e-05,
      "loss": 0.175,
      "step": 4735
    },
    {
      "epoch": 0.19149473854458338,
      "grad_norm": 2.7908570766448975,
      "learning_rate": 9.151070724817524e-05,
      "loss": 0.0978,
      "step": 4736
    },
    {
      "epoch": 0.1915351723998504,
      "grad_norm": 1.8210015296936035,
      "learning_rate": 9.150715191107151e-05,
      "loss": 0.0802,
      "step": 4737
    },
    {
      "epoch": 0.1915756062551174,
      "grad_norm": 4.224796772003174,
      "learning_rate": 9.150359589872713e-05,
      "loss": 0.2554,
      "step": 4738
    },
    {
      "epoch": 0.1916160401103844,
      "grad_norm": 4.822478771209717,
      "learning_rate": 9.150003921119992e-05,
      "loss": 0.1437,
      "step": 4739
    },
    {
      "epoch": 0.19165647396565144,
      "grad_norm": 5.266294002532959,
      "learning_rate": 9.149648184854778e-05,
      "loss": 0.2459,
      "step": 4740
    },
    {
      "epoch": 0.19169690782091844,
      "grad_norm": 5.672945499420166,
      "learning_rate": 9.149292381082853e-05,
      "loss": 0.3904,
      "step": 4741
    },
    {
      "epoch": 0.19173734167618547,
      "grad_norm": 4.561313152313232,
      "learning_rate": 9.148936509810009e-05,
      "loss": 0.1964,
      "step": 4742
    },
    {
      "epoch": 0.19177777553145248,
      "grad_norm": 2.8919806480407715,
      "learning_rate": 9.148580571042032e-05,
      "loss": 0.1964,
      "step": 4743
    },
    {
      "epoch": 0.1918182093867195,
      "grad_norm": 7.991351127624512,
      "learning_rate": 9.148224564784716e-05,
      "loss": 0.2658,
      "step": 4744
    },
    {
      "epoch": 0.1918586432419865,
      "grad_norm": 6.616203308105469,
      "learning_rate": 9.14786849104385e-05,
      "loss": 0.1753,
      "step": 4745
    },
    {
      "epoch": 0.19189907709725354,
      "grad_norm": 5.101494312286377,
      "learning_rate": 9.147512349825227e-05,
      "loss": 0.2633,
      "step": 4746
    },
    {
      "epoch": 0.19193951095252054,
      "grad_norm": 9.837409019470215,
      "learning_rate": 9.147156141134643e-05,
      "loss": 0.2879,
      "step": 4747
    },
    {
      "epoch": 0.19197994480778757,
      "grad_norm": 3.6389713287353516,
      "learning_rate": 9.14679986497789e-05,
      "loss": 0.1667,
      "step": 4748
    },
    {
      "epoch": 0.19202037866305458,
      "grad_norm": 4.217433452606201,
      "learning_rate": 9.146443521360764e-05,
      "loss": 0.183,
      "step": 4749
    },
    {
      "epoch": 0.19206081251832158,
      "grad_norm": 7.759328365325928,
      "learning_rate": 9.146087110289065e-05,
      "loss": 0.2933,
      "step": 4750
    },
    {
      "epoch": 0.1921012463735886,
      "grad_norm": 4.897210121154785,
      "learning_rate": 9.145730631768587e-05,
      "loss": 0.2045,
      "step": 4751
    },
    {
      "epoch": 0.1921416802288556,
      "grad_norm": 2.6591742038726807,
      "learning_rate": 9.145374085805132e-05,
      "loss": 0.154,
      "step": 4752
    },
    {
      "epoch": 0.19218211408412264,
      "grad_norm": 2.532534122467041,
      "learning_rate": 9.1450174724045e-05,
      "loss": 0.1474,
      "step": 4753
    },
    {
      "epoch": 0.19222254793938964,
      "grad_norm": 5.961962699890137,
      "learning_rate": 9.144660791572493e-05,
      "loss": 0.3005,
      "step": 4754
    },
    {
      "epoch": 0.19226298179465667,
      "grad_norm": 4.753220081329346,
      "learning_rate": 9.144304043314911e-05,
      "loss": 0.3188,
      "step": 4755
    },
    {
      "epoch": 0.19230341564992368,
      "grad_norm": 3.4251723289489746,
      "learning_rate": 9.143947227637559e-05,
      "loss": 0.2057,
      "step": 4756
    },
    {
      "epoch": 0.1923438495051907,
      "grad_norm": 5.3391432762146,
      "learning_rate": 9.143590344546242e-05,
      "loss": 0.1619,
      "step": 4757
    },
    {
      "epoch": 0.1923842833604577,
      "grad_norm": 2.343891143798828,
      "learning_rate": 9.143233394046766e-05,
      "loss": 0.2158,
      "step": 4758
    },
    {
      "epoch": 0.19242471721572474,
      "grad_norm": 4.539008617401123,
      "learning_rate": 9.142876376144938e-05,
      "loss": 0.1007,
      "step": 4759
    },
    {
      "epoch": 0.19246515107099174,
      "grad_norm": 3.9007294178009033,
      "learning_rate": 9.142519290846564e-05,
      "loss": 0.1496,
      "step": 4760
    },
    {
      "epoch": 0.19250558492625874,
      "grad_norm": 4.427524089813232,
      "learning_rate": 9.142162138157455e-05,
      "loss": 0.1901,
      "step": 4761
    },
    {
      "epoch": 0.19254601878152577,
      "grad_norm": 2.3378114700317383,
      "learning_rate": 9.141804918083422e-05,
      "loss": 0.135,
      "step": 4762
    },
    {
      "epoch": 0.19258645263679278,
      "grad_norm": 4.620315074920654,
      "learning_rate": 9.141447630630274e-05,
      "loss": 0.1338,
      "step": 4763
    },
    {
      "epoch": 0.1926268864920598,
      "grad_norm": 4.102975368499756,
      "learning_rate": 9.141090275803824e-05,
      "loss": 0.1026,
      "step": 4764
    },
    {
      "epoch": 0.1926673203473268,
      "grad_norm": 7.070711135864258,
      "learning_rate": 9.140732853609887e-05,
      "loss": 0.2243,
      "step": 4765
    },
    {
      "epoch": 0.19270775420259384,
      "grad_norm": 4.194623947143555,
      "learning_rate": 9.140375364054276e-05,
      "loss": 0.2066,
      "step": 4766
    },
    {
      "epoch": 0.19274818805786084,
      "grad_norm": 2.8417892456054688,
      "learning_rate": 9.140017807142807e-05,
      "loss": 0.3019,
      "step": 4767
    },
    {
      "epoch": 0.19278862191312787,
      "grad_norm": 4.626707077026367,
      "learning_rate": 9.139660182881296e-05,
      "loss": 0.1494,
      "step": 4768
    },
    {
      "epoch": 0.19282905576839487,
      "grad_norm": 2.1264028549194336,
      "learning_rate": 9.139302491275563e-05,
      "loss": 0.1288,
      "step": 4769
    },
    {
      "epoch": 0.1928694896236619,
      "grad_norm": 5.561042308807373,
      "learning_rate": 9.138944732331424e-05,
      "loss": 0.2875,
      "step": 4770
    },
    {
      "epoch": 0.1929099234789289,
      "grad_norm": 3.7232182025909424,
      "learning_rate": 9.138586906054703e-05,
      "loss": 0.1958,
      "step": 4771
    },
    {
      "epoch": 0.1929503573341959,
      "grad_norm": 4.378892421722412,
      "learning_rate": 9.138229012451217e-05,
      "loss": 0.2383,
      "step": 4772
    },
    {
      "epoch": 0.19299079118946294,
      "grad_norm": 4.888164043426514,
      "learning_rate": 9.137871051526791e-05,
      "loss": 0.1568,
      "step": 4773
    },
    {
      "epoch": 0.19303122504472994,
      "grad_norm": 5.357981204986572,
      "learning_rate": 9.137513023287246e-05,
      "loss": 0.1852,
      "step": 4774
    },
    {
      "epoch": 0.19307165889999697,
      "grad_norm": 8.82971477508545,
      "learning_rate": 9.137154927738409e-05,
      "loss": 0.2664,
      "step": 4775
    },
    {
      "epoch": 0.19311209275526398,
      "grad_norm": 5.355502128601074,
      "learning_rate": 9.136796764886104e-05,
      "loss": 0.1989,
      "step": 4776
    },
    {
      "epoch": 0.193152526610531,
      "grad_norm": 7.638251781463623,
      "learning_rate": 9.136438534736158e-05,
      "loss": 0.1554,
      "step": 4777
    },
    {
      "epoch": 0.193192960465798,
      "grad_norm": 5.136251449584961,
      "learning_rate": 9.1360802372944e-05,
      "loss": 0.1128,
      "step": 4778
    },
    {
      "epoch": 0.19323339432106504,
      "grad_norm": 8.641318321228027,
      "learning_rate": 9.135721872566657e-05,
      "loss": 0.2513,
      "step": 4779
    },
    {
      "epoch": 0.19327382817633204,
      "grad_norm": 7.6406049728393555,
      "learning_rate": 9.135363440558757e-05,
      "loss": 0.2252,
      "step": 4780
    },
    {
      "epoch": 0.19331426203159907,
      "grad_norm": 4.966799736022949,
      "learning_rate": 9.135004941276535e-05,
      "loss": 0.1733,
      "step": 4781
    },
    {
      "epoch": 0.19335469588686607,
      "grad_norm": 3.6035780906677246,
      "learning_rate": 9.134646374725822e-05,
      "loss": 0.1372,
      "step": 4782
    },
    {
      "epoch": 0.19339512974213308,
      "grad_norm": 11.374776840209961,
      "learning_rate": 9.134287740912449e-05,
      "loss": 0.2579,
      "step": 4783
    },
    {
      "epoch": 0.1934355635974001,
      "grad_norm": 3.9442992210388184,
      "learning_rate": 9.133929039842253e-05,
      "loss": 0.0972,
      "step": 4784
    },
    {
      "epoch": 0.1934759974526671,
      "grad_norm": 6.878450870513916,
      "learning_rate": 9.133570271521068e-05,
      "loss": 0.1457,
      "step": 4785
    },
    {
      "epoch": 0.19351643130793414,
      "grad_norm": 4.59736442565918,
      "learning_rate": 9.13321143595473e-05,
      "loss": 0.1532,
      "step": 4786
    },
    {
      "epoch": 0.19355686516320114,
      "grad_norm": 5.000040531158447,
      "learning_rate": 9.132852533149078e-05,
      "loss": 0.1644,
      "step": 4787
    },
    {
      "epoch": 0.19359729901846817,
      "grad_norm": 4.2359089851379395,
      "learning_rate": 9.13249356310995e-05,
      "loss": 0.2038,
      "step": 4788
    },
    {
      "epoch": 0.19363773287373517,
      "grad_norm": 5.093992233276367,
      "learning_rate": 9.132134525843186e-05,
      "loss": 0.2126,
      "step": 4789
    },
    {
      "epoch": 0.1936781667290022,
      "grad_norm": 3.195911407470703,
      "learning_rate": 9.131775421354626e-05,
      "loss": 0.1534,
      "step": 4790
    },
    {
      "epoch": 0.1937186005842692,
      "grad_norm": 8.351956367492676,
      "learning_rate": 9.131416249650112e-05,
      "loss": 0.2743,
      "step": 4791
    },
    {
      "epoch": 0.19375903443953624,
      "grad_norm": 2.0010266304016113,
      "learning_rate": 9.131057010735488e-05,
      "loss": 0.0797,
      "step": 4792
    },
    {
      "epoch": 0.19379946829480324,
      "grad_norm": 2.290940761566162,
      "learning_rate": 9.130697704616597e-05,
      "loss": 0.0699,
      "step": 4793
    },
    {
      "epoch": 0.19383990215007024,
      "grad_norm": 2.112722635269165,
      "learning_rate": 9.130338331299287e-05,
      "loss": 0.1052,
      "step": 4794
    },
    {
      "epoch": 0.19388033600533727,
      "grad_norm": 3.6109962463378906,
      "learning_rate": 9.129978890789399e-05,
      "loss": 0.0948,
      "step": 4795
    },
    {
      "epoch": 0.19392076986060428,
      "grad_norm": 4.641765117645264,
      "learning_rate": 9.129619383092786e-05,
      "loss": 0.3092,
      "step": 4796
    },
    {
      "epoch": 0.1939612037158713,
      "grad_norm": 4.74528694152832,
      "learning_rate": 9.129259808215293e-05,
      "loss": 0.1987,
      "step": 4797
    },
    {
      "epoch": 0.1940016375711383,
      "grad_norm": 6.122516632080078,
      "learning_rate": 9.128900166162769e-05,
      "loss": 0.1985,
      "step": 4798
    },
    {
      "epoch": 0.19404207142640534,
      "grad_norm": 5.319849491119385,
      "learning_rate": 9.128540456941068e-05,
      "loss": 0.2195,
      "step": 4799
    },
    {
      "epoch": 0.19408250528167234,
      "grad_norm": 1.581964135169983,
      "learning_rate": 9.12818068055604e-05,
      "loss": 0.0915,
      "step": 4800
    },
    {
      "epoch": 0.19412293913693937,
      "grad_norm": 6.496124744415283,
      "learning_rate": 9.127820837013537e-05,
      "loss": 0.1059,
      "step": 4801
    },
    {
      "epoch": 0.19416337299220637,
      "grad_norm": 6.480955123901367,
      "learning_rate": 9.127460926319416e-05,
      "loss": 0.1722,
      "step": 4802
    },
    {
      "epoch": 0.19420380684747338,
      "grad_norm": 7.557129383087158,
      "learning_rate": 9.127100948479527e-05,
      "loss": 0.2918,
      "step": 4803
    },
    {
      "epoch": 0.1942442407027404,
      "grad_norm": 7.284797191619873,
      "learning_rate": 9.126740903499731e-05,
      "loss": 0.2497,
      "step": 4804
    },
    {
      "epoch": 0.1942846745580074,
      "grad_norm": 3.467097520828247,
      "learning_rate": 9.126380791385882e-05,
      "loss": 0.1045,
      "step": 4805
    },
    {
      "epoch": 0.19432510841327444,
      "grad_norm": 4.393749713897705,
      "learning_rate": 9.12602061214384e-05,
      "loss": 0.16,
      "step": 4806
    },
    {
      "epoch": 0.19436554226854144,
      "grad_norm": 3.5164358615875244,
      "learning_rate": 9.125660365779464e-05,
      "loss": 0.1387,
      "step": 4807
    },
    {
      "epoch": 0.19440597612380847,
      "grad_norm": 5.829476356506348,
      "learning_rate": 9.125300052298616e-05,
      "loss": 0.3045,
      "step": 4808
    },
    {
      "epoch": 0.19444640997907547,
      "grad_norm": 7.637275218963623,
      "learning_rate": 9.124939671707154e-05,
      "loss": 0.345,
      "step": 4809
    },
    {
      "epoch": 0.1944868438343425,
      "grad_norm": 3.27085018157959,
      "learning_rate": 9.124579224010942e-05,
      "loss": 0.2632,
      "step": 4810
    },
    {
      "epoch": 0.1945272776896095,
      "grad_norm": 2.9042954444885254,
      "learning_rate": 9.124218709215848e-05,
      "loss": 0.0923,
      "step": 4811
    },
    {
      "epoch": 0.19456771154487654,
      "grad_norm": 3.868025779724121,
      "learning_rate": 9.123858127327731e-05,
      "loss": 0.124,
      "step": 4812
    },
    {
      "epoch": 0.19460814540014354,
      "grad_norm": 7.759350299835205,
      "learning_rate": 9.12349747835246e-05,
      "loss": 0.1726,
      "step": 4813
    },
    {
      "epoch": 0.19464857925541054,
      "grad_norm": 9.181324005126953,
      "learning_rate": 9.1231367622959e-05,
      "loss": 0.376,
      "step": 4814
    },
    {
      "epoch": 0.19468901311067757,
      "grad_norm": 3.262580394744873,
      "learning_rate": 9.122775979163923e-05,
      "loss": 0.066,
      "step": 4815
    },
    {
      "epoch": 0.19472944696594457,
      "grad_norm": 3.013742208480835,
      "learning_rate": 9.122415128962396e-05,
      "loss": 0.1344,
      "step": 4816
    },
    {
      "epoch": 0.1947698808212116,
      "grad_norm": 6.5224127769470215,
      "learning_rate": 9.122054211697188e-05,
      "loss": 0.21,
      "step": 4817
    },
    {
      "epoch": 0.1948103146764786,
      "grad_norm": 5.510965824127197,
      "learning_rate": 9.12169322737417e-05,
      "loss": 0.2852,
      "step": 4818
    },
    {
      "epoch": 0.19485074853174564,
      "grad_norm": 2.9954941272735596,
      "learning_rate": 9.121332175999218e-05,
      "loss": 0.2045,
      "step": 4819
    },
    {
      "epoch": 0.19489118238701264,
      "grad_norm": 8.333091735839844,
      "learning_rate": 9.120971057578205e-05,
      "loss": 0.2631,
      "step": 4820
    },
    {
      "epoch": 0.19493161624227967,
      "grad_norm": 2.94523549079895,
      "learning_rate": 9.120609872117003e-05,
      "loss": 0.0463,
      "step": 4821
    },
    {
      "epoch": 0.19497205009754667,
      "grad_norm": 4.976259708404541,
      "learning_rate": 9.120248619621489e-05,
      "loss": 0.2512,
      "step": 4822
    },
    {
      "epoch": 0.1950124839528137,
      "grad_norm": 5.397971153259277,
      "learning_rate": 9.11988730009754e-05,
      "loss": 0.197,
      "step": 4823
    },
    {
      "epoch": 0.1950529178080807,
      "grad_norm": 6.039949417114258,
      "learning_rate": 9.119525913551034e-05,
      "loss": 0.2446,
      "step": 4824
    },
    {
      "epoch": 0.1950933516633477,
      "grad_norm": 4.33865213394165,
      "learning_rate": 9.11916445998785e-05,
      "loss": 0.1433,
      "step": 4825
    },
    {
      "epoch": 0.19513378551861474,
      "grad_norm": 3.9796392917633057,
      "learning_rate": 9.118802939413869e-05,
      "loss": 0.0755,
      "step": 4826
    },
    {
      "epoch": 0.19517421937388174,
      "grad_norm": 2.660489559173584,
      "learning_rate": 9.118441351834972e-05,
      "loss": 0.1483,
      "step": 4827
    },
    {
      "epoch": 0.19521465322914877,
      "grad_norm": 2.0612642765045166,
      "learning_rate": 9.118079697257041e-05,
      "loss": 0.0468,
      "step": 4828
    },
    {
      "epoch": 0.19525508708441577,
      "grad_norm": 7.6322021484375,
      "learning_rate": 9.117717975685958e-05,
      "loss": 0.1487,
      "step": 4829
    },
    {
      "epoch": 0.1952955209396828,
      "grad_norm": 4.812276840209961,
      "learning_rate": 9.117356187127608e-05,
      "loss": 0.1287,
      "step": 4830
    },
    {
      "epoch": 0.1953359547949498,
      "grad_norm": 3.4092166423797607,
      "learning_rate": 9.116994331587879e-05,
      "loss": 0.1312,
      "step": 4831
    },
    {
      "epoch": 0.19537638865021684,
      "grad_norm": 5.153774738311768,
      "learning_rate": 9.116632409072655e-05,
      "loss": 0.0881,
      "step": 4832
    },
    {
      "epoch": 0.19541682250548384,
      "grad_norm": 4.468329429626465,
      "learning_rate": 9.116270419587825e-05,
      "loss": 0.0931,
      "step": 4833
    },
    {
      "epoch": 0.19545725636075087,
      "grad_norm": 2.909571647644043,
      "learning_rate": 9.115908363139279e-05,
      "loss": 0.1182,
      "step": 4834
    },
    {
      "epoch": 0.19549769021601787,
      "grad_norm": 6.947986602783203,
      "learning_rate": 9.115546239732903e-05,
      "loss": 0.2893,
      "step": 4835
    },
    {
      "epoch": 0.19553812407128487,
      "grad_norm": 2.1611549854278564,
      "learning_rate": 9.115184049374593e-05,
      "loss": 0.1381,
      "step": 4836
    },
    {
      "epoch": 0.1955785579265519,
      "grad_norm": 6.947922229766846,
      "learning_rate": 9.114821792070237e-05,
      "loss": 0.18,
      "step": 4837
    },
    {
      "epoch": 0.1956189917818189,
      "grad_norm": 8.234424591064453,
      "learning_rate": 9.114459467825731e-05,
      "loss": 0.2329,
      "step": 4838
    },
    {
      "epoch": 0.19565942563708594,
      "grad_norm": 2.386941909790039,
      "learning_rate": 9.114097076646968e-05,
      "loss": 0.0282,
      "step": 4839
    },
    {
      "epoch": 0.19569985949235294,
      "grad_norm": 6.333490371704102,
      "learning_rate": 9.113734618539844e-05,
      "loss": 0.1038,
      "step": 4840
    },
    {
      "epoch": 0.19574029334761997,
      "grad_norm": 5.241218090057373,
      "learning_rate": 9.113372093510255e-05,
      "loss": 0.1005,
      "step": 4841
    },
    {
      "epoch": 0.19578072720288697,
      "grad_norm": 3.2393453121185303,
      "learning_rate": 9.1130095015641e-05,
      "loss": 0.0876,
      "step": 4842
    },
    {
      "epoch": 0.195821161058154,
      "grad_norm": 5.81091833114624,
      "learning_rate": 9.112646842707275e-05,
      "loss": 0.2305,
      "step": 4843
    },
    {
      "epoch": 0.195861594913421,
      "grad_norm": 4.144407272338867,
      "learning_rate": 9.112284116945682e-05,
      "loss": 0.3845,
      "step": 4844
    },
    {
      "epoch": 0.19590202876868804,
      "grad_norm": 6.358283519744873,
      "learning_rate": 9.111921324285222e-05,
      "loss": 0.1825,
      "step": 4845
    },
    {
      "epoch": 0.19594246262395504,
      "grad_norm": 4.247817516326904,
      "learning_rate": 9.111558464731794e-05,
      "loss": 0.1921,
      "step": 4846
    },
    {
      "epoch": 0.19598289647922204,
      "grad_norm": 8.280976295471191,
      "learning_rate": 9.111195538291304e-05,
      "loss": 0.2192,
      "step": 4847
    },
    {
      "epoch": 0.19602333033448907,
      "grad_norm": 2.5544233322143555,
      "learning_rate": 9.110832544969654e-05,
      "loss": 0.0606,
      "step": 4848
    },
    {
      "epoch": 0.19606376418975607,
      "grad_norm": 13.130144119262695,
      "learning_rate": 9.110469484772752e-05,
      "loss": 0.4779,
      "step": 4849
    },
    {
      "epoch": 0.1961041980450231,
      "grad_norm": 11.152350425720215,
      "learning_rate": 9.110106357706502e-05,
      "loss": 0.2679,
      "step": 4850
    },
    {
      "epoch": 0.1961446319002901,
      "grad_norm": 11.735795974731445,
      "learning_rate": 9.109743163776812e-05,
      "loss": 0.4136,
      "step": 4851
    },
    {
      "epoch": 0.19618506575555714,
      "grad_norm": 5.436749458312988,
      "learning_rate": 9.10937990298959e-05,
      "loss": 0.2325,
      "step": 4852
    },
    {
      "epoch": 0.19622549961082414,
      "grad_norm": 7.288030624389648,
      "learning_rate": 9.109016575350746e-05,
      "loss": 0.152,
      "step": 4853
    },
    {
      "epoch": 0.19626593346609117,
      "grad_norm": 3.785259246826172,
      "learning_rate": 9.108653180866192e-05,
      "loss": 0.2324,
      "step": 4854
    },
    {
      "epoch": 0.19630636732135817,
      "grad_norm": 6.082887172698975,
      "learning_rate": 9.108289719541836e-05,
      "loss": 0.263,
      "step": 4855
    },
    {
      "epoch": 0.1963468011766252,
      "grad_norm": 2.378286600112915,
      "learning_rate": 9.107926191383594e-05,
      "loss": 0.0975,
      "step": 4856
    },
    {
      "epoch": 0.1963872350318922,
      "grad_norm": 5.142353057861328,
      "learning_rate": 9.10756259639738e-05,
      "loss": 0.3595,
      "step": 4857
    },
    {
      "epoch": 0.1964276688871592,
      "grad_norm": 2.3722715377807617,
      "learning_rate": 9.107198934589107e-05,
      "loss": 0.11,
      "step": 4858
    },
    {
      "epoch": 0.19646810274242624,
      "grad_norm": 3.513436794281006,
      "learning_rate": 9.106835205964693e-05,
      "loss": 0.235,
      "step": 4859
    },
    {
      "epoch": 0.19650853659769324,
      "grad_norm": 5.653253078460693,
      "learning_rate": 9.106471410530053e-05,
      "loss": 0.2164,
      "step": 4860
    },
    {
      "epoch": 0.19654897045296027,
      "grad_norm": 5.82438325881958,
      "learning_rate": 9.106107548291107e-05,
      "loss": 0.2352,
      "step": 4861
    },
    {
      "epoch": 0.19658940430822727,
      "grad_norm": 4.4305830001831055,
      "learning_rate": 9.105743619253774e-05,
      "loss": 0.157,
      "step": 4862
    },
    {
      "epoch": 0.1966298381634943,
      "grad_norm": 3.538705348968506,
      "learning_rate": 9.105379623423975e-05,
      "loss": 0.1813,
      "step": 4863
    },
    {
      "epoch": 0.1966702720187613,
      "grad_norm": 7.761295318603516,
      "learning_rate": 9.10501556080763e-05,
      "loss": 0.2048,
      "step": 4864
    },
    {
      "epoch": 0.19671070587402834,
      "grad_norm": 8.826926231384277,
      "learning_rate": 9.104651431410663e-05,
      "loss": 0.25,
      "step": 4865
    },
    {
      "epoch": 0.19675113972929534,
      "grad_norm": 6.5418009757995605,
      "learning_rate": 9.104287235238995e-05,
      "loss": 0.1427,
      "step": 4866
    },
    {
      "epoch": 0.19679157358456234,
      "grad_norm": 4.237258434295654,
      "learning_rate": 9.103922972298555e-05,
      "loss": 0.0762,
      "step": 4867
    },
    {
      "epoch": 0.19683200743982937,
      "grad_norm": 5.610339641571045,
      "learning_rate": 9.103558642595266e-05,
      "loss": 0.1195,
      "step": 4868
    },
    {
      "epoch": 0.19687244129509637,
      "grad_norm": 6.4565911293029785,
      "learning_rate": 9.103194246135056e-05,
      "loss": 0.2481,
      "step": 4869
    },
    {
      "epoch": 0.1969128751503634,
      "grad_norm": 6.991971492767334,
      "learning_rate": 9.102829782923852e-05,
      "loss": 0.3265,
      "step": 4870
    },
    {
      "epoch": 0.1969533090056304,
      "grad_norm": 3.5242490768432617,
      "learning_rate": 9.102465252967583e-05,
      "loss": 0.1609,
      "step": 4871
    },
    {
      "epoch": 0.19699374286089744,
      "grad_norm": 6.453104019165039,
      "learning_rate": 9.10210065627218e-05,
      "loss": 0.2216,
      "step": 4872
    },
    {
      "epoch": 0.19703417671616444,
      "grad_norm": 5.809821605682373,
      "learning_rate": 9.101735992843574e-05,
      "loss": 0.1517,
      "step": 4873
    },
    {
      "epoch": 0.19707461057143147,
      "grad_norm": 4.715822696685791,
      "learning_rate": 9.101371262687699e-05,
      "loss": 0.1776,
      "step": 4874
    },
    {
      "epoch": 0.19711504442669847,
      "grad_norm": 1.990064263343811,
      "learning_rate": 9.101006465810486e-05,
      "loss": 0.1039,
      "step": 4875
    },
    {
      "epoch": 0.1971554782819655,
      "grad_norm": 4.031190395355225,
      "learning_rate": 9.100641602217872e-05,
      "loss": 0.1523,
      "step": 4876
    },
    {
      "epoch": 0.1971959121372325,
      "grad_norm": 5.051582336425781,
      "learning_rate": 9.10027667191579e-05,
      "loss": 0.2043,
      "step": 4877
    },
    {
      "epoch": 0.1972363459924995,
      "grad_norm": 4.9946794509887695,
      "learning_rate": 9.099911674910177e-05,
      "loss": 0.2295,
      "step": 4878
    },
    {
      "epoch": 0.19727677984776654,
      "grad_norm": 4.193070411682129,
      "learning_rate": 9.099546611206972e-05,
      "loss": 0.1646,
      "step": 4879
    },
    {
      "epoch": 0.19731721370303354,
      "grad_norm": 6.715747833251953,
      "learning_rate": 9.099181480812114e-05,
      "loss": 0.1777,
      "step": 4880
    },
    {
      "epoch": 0.19735764755830057,
      "grad_norm": 4.203471660614014,
      "learning_rate": 9.098816283731543e-05,
      "loss": 0.1878,
      "step": 4881
    },
    {
      "epoch": 0.19739808141356757,
      "grad_norm": 2.0084221363067627,
      "learning_rate": 9.0984510199712e-05,
      "loss": 0.112,
      "step": 4882
    },
    {
      "epoch": 0.1974385152688346,
      "grad_norm": 2.7467615604400635,
      "learning_rate": 9.098085689537024e-05,
      "loss": 0.1166,
      "step": 4883
    },
    {
      "epoch": 0.1974789491241016,
      "grad_norm": 3.9631118774414062,
      "learning_rate": 9.097720292434964e-05,
      "loss": 0.1258,
      "step": 4884
    },
    {
      "epoch": 0.19751938297936864,
      "grad_norm": 3.443277597427368,
      "learning_rate": 9.097354828670958e-05,
      "loss": 0.2468,
      "step": 4885
    },
    {
      "epoch": 0.19755981683463564,
      "grad_norm": 6.698768138885498,
      "learning_rate": 9.096989298250957e-05,
      "loss": 0.1934,
      "step": 4886
    },
    {
      "epoch": 0.19760025068990267,
      "grad_norm": 5.485813617706299,
      "learning_rate": 9.096623701180905e-05,
      "loss": 0.1335,
      "step": 4887
    },
    {
      "epoch": 0.19764068454516967,
      "grad_norm": 1.7506382465362549,
      "learning_rate": 9.096258037466747e-05,
      "loss": 0.0445,
      "step": 4888
    },
    {
      "epoch": 0.19768111840043667,
      "grad_norm": 3.705080032348633,
      "learning_rate": 9.095892307114437e-05,
      "loss": 0.203,
      "step": 4889
    },
    {
      "epoch": 0.1977215522557037,
      "grad_norm": 4.181849956512451,
      "learning_rate": 9.09552651012992e-05,
      "loss": 0.1603,
      "step": 4890
    },
    {
      "epoch": 0.1977619861109707,
      "grad_norm": 8.130167007446289,
      "learning_rate": 9.095160646519149e-05,
      "loss": 0.2289,
      "step": 4891
    },
    {
      "epoch": 0.19780241996623774,
      "grad_norm": 6.980086326599121,
      "learning_rate": 9.094794716288077e-05,
      "loss": 0.2442,
      "step": 4892
    },
    {
      "epoch": 0.19784285382150474,
      "grad_norm": 4.033838748931885,
      "learning_rate": 9.094428719442655e-05,
      "loss": 0.1119,
      "step": 4893
    },
    {
      "epoch": 0.19788328767677177,
      "grad_norm": 6.50110387802124,
      "learning_rate": 9.094062655988837e-05,
      "loss": 0.2171,
      "step": 4894
    },
    {
      "epoch": 0.19792372153203877,
      "grad_norm": 3.654524087905884,
      "learning_rate": 9.09369652593258e-05,
      "loss": 0.1147,
      "step": 4895
    },
    {
      "epoch": 0.1979641553873058,
      "grad_norm": 4.2362236976623535,
      "learning_rate": 9.093330329279839e-05,
      "loss": 0.2729,
      "step": 4896
    },
    {
      "epoch": 0.1980045892425728,
      "grad_norm": 6.857854843139648,
      "learning_rate": 9.092964066036571e-05,
      "loss": 0.2644,
      "step": 4897
    },
    {
      "epoch": 0.19804502309783983,
      "grad_norm": 7.226253509521484,
      "learning_rate": 9.092597736208734e-05,
      "loss": 0.0938,
      "step": 4898
    },
    {
      "epoch": 0.19808545695310684,
      "grad_norm": 5.395706653594971,
      "learning_rate": 9.092231339802289e-05,
      "loss": 0.2251,
      "step": 4899
    },
    {
      "epoch": 0.19812589080837384,
      "grad_norm": 13.016413688659668,
      "learning_rate": 9.091864876823195e-05,
      "loss": 0.3196,
      "step": 4900
    },
    {
      "epoch": 0.19816632466364087,
      "grad_norm": 3.5640807151794434,
      "learning_rate": 9.091498347277416e-05,
      "loss": 0.2527,
      "step": 4901
    },
    {
      "epoch": 0.19820675851890787,
      "grad_norm": 7.692381858825684,
      "learning_rate": 9.091131751170912e-05,
      "loss": 0.1903,
      "step": 4902
    },
    {
      "epoch": 0.1982471923741749,
      "grad_norm": 4.3520050048828125,
      "learning_rate": 9.090765088509647e-05,
      "loss": 0.1645,
      "step": 4903
    },
    {
      "epoch": 0.1982876262294419,
      "grad_norm": 4.704237937927246,
      "learning_rate": 9.090398359299588e-05,
      "loss": 0.1553,
      "step": 4904
    },
    {
      "epoch": 0.19832806008470893,
      "grad_norm": 2.665510416030884,
      "learning_rate": 9.0900315635467e-05,
      "loss": 0.1958,
      "step": 4905
    },
    {
      "epoch": 0.19836849393997594,
      "grad_norm": 2.9364259243011475,
      "learning_rate": 9.08966470125695e-05,
      "loss": 0.1876,
      "step": 4906
    },
    {
      "epoch": 0.19840892779524297,
      "grad_norm": 3.0342421531677246,
      "learning_rate": 9.089297772436304e-05,
      "loss": 0.1534,
      "step": 4907
    },
    {
      "epoch": 0.19844936165050997,
      "grad_norm": 3.285545587539673,
      "learning_rate": 9.088930777090736e-05,
      "loss": 0.1475,
      "step": 4908
    },
    {
      "epoch": 0.198489795505777,
      "grad_norm": 6.026314735412598,
      "learning_rate": 9.088563715226211e-05,
      "loss": 0.2599,
      "step": 4909
    },
    {
      "epoch": 0.198530229361044,
      "grad_norm": 3.2996554374694824,
      "learning_rate": 9.088196586848705e-05,
      "loss": 0.0924,
      "step": 4910
    },
    {
      "epoch": 0.198570663216311,
      "grad_norm": 3.3963913917541504,
      "learning_rate": 9.087829391964186e-05,
      "loss": 0.1672,
      "step": 4911
    },
    {
      "epoch": 0.19861109707157804,
      "grad_norm": 5.505708694458008,
      "learning_rate": 9.087462130578632e-05,
      "loss": 0.2105,
      "step": 4912
    },
    {
      "epoch": 0.19865153092684504,
      "grad_norm": 5.822805881500244,
      "learning_rate": 9.087094802698015e-05,
      "loss": 0.187,
      "step": 4913
    },
    {
      "epoch": 0.19869196478211207,
      "grad_norm": 8.74193000793457,
      "learning_rate": 9.086727408328311e-05,
      "loss": 0.1494,
      "step": 4914
    },
    {
      "epoch": 0.19873239863737907,
      "grad_norm": 6.621281623840332,
      "learning_rate": 9.086359947475497e-05,
      "loss": 0.1452,
      "step": 4915
    },
    {
      "epoch": 0.1987728324926461,
      "grad_norm": 6.631019592285156,
      "learning_rate": 9.085992420145551e-05,
      "loss": 0.1646,
      "step": 4916
    },
    {
      "epoch": 0.1988132663479131,
      "grad_norm": 2.7964601516723633,
      "learning_rate": 9.085624826344452e-05,
      "loss": 0.1704,
      "step": 4917
    },
    {
      "epoch": 0.19885370020318013,
      "grad_norm": 5.306056976318359,
      "learning_rate": 9.08525716607818e-05,
      "loss": 0.188,
      "step": 4918
    },
    {
      "epoch": 0.19889413405844714,
      "grad_norm": 5.034843921661377,
      "learning_rate": 9.084889439352716e-05,
      "loss": 0.1892,
      "step": 4919
    },
    {
      "epoch": 0.19893456791371417,
      "grad_norm": 10.973958015441895,
      "learning_rate": 9.08452164617404e-05,
      "loss": 0.2836,
      "step": 4920
    },
    {
      "epoch": 0.19897500176898117,
      "grad_norm": 2.3860127925872803,
      "learning_rate": 9.084153786548141e-05,
      "loss": 0.0993,
      "step": 4921
    },
    {
      "epoch": 0.19901543562424817,
      "grad_norm": 8.53832721710205,
      "learning_rate": 9.083785860480998e-05,
      "loss": 0.1872,
      "step": 4922
    },
    {
      "epoch": 0.1990558694795152,
      "grad_norm": 3.5362658500671387,
      "learning_rate": 9.083417867978599e-05,
      "loss": 0.1178,
      "step": 4923
    },
    {
      "epoch": 0.1990963033347822,
      "grad_norm": 9.429120063781738,
      "learning_rate": 9.083049809046928e-05,
      "loss": 0.2077,
      "step": 4924
    },
    {
      "epoch": 0.19913673719004923,
      "grad_norm": 8.407569885253906,
      "learning_rate": 9.082681683691976e-05,
      "loss": 0.1743,
      "step": 4925
    },
    {
      "epoch": 0.19917717104531624,
      "grad_norm": 6.194862365722656,
      "learning_rate": 9.08231349191973e-05,
      "loss": 0.2779,
      "step": 4926
    },
    {
      "epoch": 0.19921760490058327,
      "grad_norm": 4.404684543609619,
      "learning_rate": 9.08194523373618e-05,
      "loss": 0.1023,
      "step": 4927
    },
    {
      "epoch": 0.19925803875585027,
      "grad_norm": 5.386590957641602,
      "learning_rate": 9.081576909147316e-05,
      "loss": 0.129,
      "step": 4928
    },
    {
      "epoch": 0.1992984726111173,
      "grad_norm": 6.154804229736328,
      "learning_rate": 9.081208518159131e-05,
      "loss": 0.2225,
      "step": 4929
    },
    {
      "epoch": 0.1993389064663843,
      "grad_norm": 6.151362895965576,
      "learning_rate": 9.080840060777617e-05,
      "loss": 0.1697,
      "step": 4930
    },
    {
      "epoch": 0.1993793403216513,
      "grad_norm": 4.244097709655762,
      "learning_rate": 9.08047153700877e-05,
      "loss": 0.1619,
      "step": 4931
    },
    {
      "epoch": 0.19941977417691834,
      "grad_norm": 7.746179580688477,
      "learning_rate": 9.080102946858584e-05,
      "loss": 0.2196,
      "step": 4932
    },
    {
      "epoch": 0.19946020803218534,
      "grad_norm": 7.487579345703125,
      "learning_rate": 9.079734290333055e-05,
      "loss": 0.2874,
      "step": 4933
    },
    {
      "epoch": 0.19950064188745237,
      "grad_norm": 4.974328994750977,
      "learning_rate": 9.07936556743818e-05,
      "loss": 0.2081,
      "step": 4934
    },
    {
      "epoch": 0.19954107574271937,
      "grad_norm": 10.30380630493164,
      "learning_rate": 9.078996778179958e-05,
      "loss": 0.3166,
      "step": 4935
    },
    {
      "epoch": 0.1995815095979864,
      "grad_norm": 6.948485851287842,
      "learning_rate": 9.078627922564386e-05,
      "loss": 0.2476,
      "step": 4936
    },
    {
      "epoch": 0.1996219434532534,
      "grad_norm": 6.672253608703613,
      "learning_rate": 9.078259000597471e-05,
      "loss": 0.1612,
      "step": 4937
    },
    {
      "epoch": 0.19966237730852043,
      "grad_norm": 6.922666072845459,
      "learning_rate": 9.077890012285209e-05,
      "loss": 0.2955,
      "step": 4938
    },
    {
      "epoch": 0.19970281116378744,
      "grad_norm": 3.9007604122161865,
      "learning_rate": 9.077520957633603e-05,
      "loss": 0.1215,
      "step": 4939
    },
    {
      "epoch": 0.19974324501905447,
      "grad_norm": 8.303632736206055,
      "learning_rate": 9.077151836648659e-05,
      "loss": 0.3062,
      "step": 4940
    },
    {
      "epoch": 0.19978367887432147,
      "grad_norm": 6.9413676261901855,
      "learning_rate": 9.076782649336381e-05,
      "loss": 0.2469,
      "step": 4941
    },
    {
      "epoch": 0.19982411272958847,
      "grad_norm": 6.944331645965576,
      "learning_rate": 9.076413395702774e-05,
      "loss": 0.1926,
      "step": 4942
    },
    {
      "epoch": 0.1998645465848555,
      "grad_norm": 4.028487205505371,
      "learning_rate": 9.076044075753848e-05,
      "loss": 0.1251,
      "step": 4943
    },
    {
      "epoch": 0.1999049804401225,
      "grad_norm": 6.917934894561768,
      "learning_rate": 9.075674689495608e-05,
      "loss": 0.2406,
      "step": 4944
    },
    {
      "epoch": 0.19994541429538953,
      "grad_norm": 3.7953028678894043,
      "learning_rate": 9.075305236934064e-05,
      "loss": 0.2448,
      "step": 4945
    },
    {
      "epoch": 0.19998584815065654,
      "grad_norm": 3.275264024734497,
      "learning_rate": 9.074935718075225e-05,
      "loss": 0.1373,
      "step": 4946
    },
    {
      "epoch": 0.20002628200592357,
      "grad_norm": 3.5540428161621094,
      "learning_rate": 9.074566132925107e-05,
      "loss": 0.2647,
      "step": 4947
    },
    {
      "epoch": 0.20006671586119057,
      "grad_norm": 7.174386024475098,
      "learning_rate": 9.074196481489718e-05,
      "loss": 0.1257,
      "step": 4948
    },
    {
      "epoch": 0.2001071497164576,
      "grad_norm": 7.346698760986328,
      "learning_rate": 9.073826763775073e-05,
      "loss": 0.2473,
      "step": 4949
    },
    {
      "epoch": 0.2001475835717246,
      "grad_norm": 7.982274532318115,
      "learning_rate": 9.073456979787187e-05,
      "loss": 0.2531,
      "step": 4950
    },
    {
      "epoch": 0.20018801742699163,
      "grad_norm": 5.113940715789795,
      "learning_rate": 9.073087129532074e-05,
      "loss": 0.1031,
      "step": 4951
    },
    {
      "epoch": 0.20022845128225863,
      "grad_norm": 3.534412145614624,
      "learning_rate": 9.072717213015751e-05,
      "loss": 0.163,
      "step": 4952
    },
    {
      "epoch": 0.20026888513752564,
      "grad_norm": 5.842316150665283,
      "learning_rate": 9.072347230244239e-05,
      "loss": 0.1623,
      "step": 4953
    },
    {
      "epoch": 0.20030931899279267,
      "grad_norm": 5.257890701293945,
      "learning_rate": 9.071977181223554e-05,
      "loss": 0.192,
      "step": 4954
    },
    {
      "epoch": 0.20034975284805967,
      "grad_norm": 6.7178874015808105,
      "learning_rate": 9.071607065959715e-05,
      "loss": 0.1028,
      "step": 4955
    },
    {
      "epoch": 0.2003901867033267,
      "grad_norm": 3.1637585163116455,
      "learning_rate": 9.071236884458745e-05,
      "loss": 0.1027,
      "step": 4956
    },
    {
      "epoch": 0.2004306205585937,
      "grad_norm": 2.497127056121826,
      "learning_rate": 9.070866636726666e-05,
      "loss": 0.1295,
      "step": 4957
    },
    {
      "epoch": 0.20047105441386073,
      "grad_norm": 5.079200744628906,
      "learning_rate": 9.070496322769501e-05,
      "loss": 0.1485,
      "step": 4958
    },
    {
      "epoch": 0.20051148826912774,
      "grad_norm": 3.202166795730591,
      "learning_rate": 9.070125942593276e-05,
      "loss": 0.1349,
      "step": 4959
    },
    {
      "epoch": 0.20055192212439477,
      "grad_norm": 7.869395732879639,
      "learning_rate": 9.069755496204014e-05,
      "loss": 0.2236,
      "step": 4960
    },
    {
      "epoch": 0.20059235597966177,
      "grad_norm": 6.012377738952637,
      "learning_rate": 9.06938498360774e-05,
      "loss": 0.1287,
      "step": 4961
    },
    {
      "epoch": 0.2006327898349288,
      "grad_norm": 7.490691661834717,
      "learning_rate": 9.069014404810485e-05,
      "loss": 0.2412,
      "step": 4962
    },
    {
      "epoch": 0.2006732236901958,
      "grad_norm": 5.440084457397461,
      "learning_rate": 9.068643759818277e-05,
      "loss": 0.1509,
      "step": 4963
    },
    {
      "epoch": 0.2007136575454628,
      "grad_norm": 3.8305227756500244,
      "learning_rate": 9.068273048637141e-05,
      "loss": 0.2274,
      "step": 4964
    },
    {
      "epoch": 0.20075409140072983,
      "grad_norm": 5.19408655166626,
      "learning_rate": 9.067902271273117e-05,
      "loss": 0.2764,
      "step": 4965
    },
    {
      "epoch": 0.20079452525599684,
      "grad_norm": 5.728510856628418,
      "learning_rate": 9.067531427732227e-05,
      "loss": 0.129,
      "step": 4966
    },
    {
      "epoch": 0.20083495911126387,
      "grad_norm": 7.341366767883301,
      "learning_rate": 9.06716051802051e-05,
      "loss": 0.2317,
      "step": 4967
    },
    {
      "epoch": 0.20087539296653087,
      "grad_norm": 6.162217617034912,
      "learning_rate": 9.066789542143999e-05,
      "loss": 0.2032,
      "step": 4968
    },
    {
      "epoch": 0.2009158268217979,
      "grad_norm": 6.625766754150391,
      "learning_rate": 9.066418500108727e-05,
      "loss": 0.1461,
      "step": 4969
    },
    {
      "epoch": 0.2009562606770649,
      "grad_norm": 2.870574951171875,
      "learning_rate": 9.066047391920733e-05,
      "loss": 0.0999,
      "step": 4970
    },
    {
      "epoch": 0.20099669453233193,
      "grad_norm": 7.335565090179443,
      "learning_rate": 9.06567621758605e-05,
      "loss": 0.2513,
      "step": 4971
    },
    {
      "epoch": 0.20103712838759893,
      "grad_norm": 3.0091712474823,
      "learning_rate": 9.065304977110723e-05,
      "loss": 0.0782,
      "step": 4972
    },
    {
      "epoch": 0.20107756224286596,
      "grad_norm": 5.470030784606934,
      "learning_rate": 9.064933670500783e-05,
      "loss": 0.203,
      "step": 4973
    },
    {
      "epoch": 0.20111799609813297,
      "grad_norm": 3.7949721813201904,
      "learning_rate": 9.064562297762277e-05,
      "loss": 0.183,
      "step": 4974
    },
    {
      "epoch": 0.20115842995339997,
      "grad_norm": 5.376256465911865,
      "learning_rate": 9.064190858901245e-05,
      "loss": 0.1466,
      "step": 4975
    },
    {
      "epoch": 0.201198863808667,
      "grad_norm": 3.395305871963501,
      "learning_rate": 9.063819353923728e-05,
      "loss": 0.1315,
      "step": 4976
    },
    {
      "epoch": 0.201239297663934,
      "grad_norm": 5.1008429527282715,
      "learning_rate": 9.06344778283577e-05,
      "loss": 0.2839,
      "step": 4977
    },
    {
      "epoch": 0.20127973151920103,
      "grad_norm": 5.054925918579102,
      "learning_rate": 9.063076145643419e-05,
      "loss": 0.3142,
      "step": 4978
    },
    {
      "epoch": 0.20132016537446804,
      "grad_norm": 3.8735311031341553,
      "learning_rate": 9.062704442352715e-05,
      "loss": 0.1768,
      "step": 4979
    },
    {
      "epoch": 0.20136059922973507,
      "grad_norm": 7.173231601715088,
      "learning_rate": 9.062332672969709e-05,
      "loss": 0.2832,
      "step": 4980
    },
    {
      "epoch": 0.20140103308500207,
      "grad_norm": 3.9031949043273926,
      "learning_rate": 9.061960837500447e-05,
      "loss": 0.0829,
      "step": 4981
    },
    {
      "epoch": 0.2014414669402691,
      "grad_norm": 5.462329864501953,
      "learning_rate": 9.06158893595098e-05,
      "loss": 0.1676,
      "step": 4982
    },
    {
      "epoch": 0.2014819007955361,
      "grad_norm": 6.397858142852783,
      "learning_rate": 9.061216968327357e-05,
      "loss": 0.1313,
      "step": 4983
    },
    {
      "epoch": 0.20152233465080313,
      "grad_norm": 4.538091659545898,
      "learning_rate": 9.06084493463563e-05,
      "loss": 0.142,
      "step": 4984
    },
    {
      "epoch": 0.20156276850607013,
      "grad_norm": 7.192776679992676,
      "learning_rate": 9.06047283488185e-05,
      "loss": 0.275,
      "step": 4985
    },
    {
      "epoch": 0.20160320236133714,
      "grad_norm": 3.53859543800354,
      "learning_rate": 9.06010066907207e-05,
      "loss": 0.1176,
      "step": 4986
    },
    {
      "epoch": 0.20164363621660417,
      "grad_norm": 2.68984055519104,
      "learning_rate": 9.059728437212346e-05,
      "loss": 0.1256,
      "step": 4987
    },
    {
      "epoch": 0.20168407007187117,
      "grad_norm": 2.1794028282165527,
      "learning_rate": 9.059356139308734e-05,
      "loss": 0.2107,
      "step": 4988
    },
    {
      "epoch": 0.2017245039271382,
      "grad_norm": 2.1445772647857666,
      "learning_rate": 9.058983775367287e-05,
      "loss": 0.1558,
      "step": 4989
    },
    {
      "epoch": 0.2017649377824052,
      "grad_norm": 2.727665662765503,
      "learning_rate": 9.058611345394066e-05,
      "loss": 0.0807,
      "step": 4990
    },
    {
      "epoch": 0.20180537163767223,
      "grad_norm": 2.7045421600341797,
      "learning_rate": 9.05823884939513e-05,
      "loss": 0.0906,
      "step": 4991
    },
    {
      "epoch": 0.20184580549293923,
      "grad_norm": 5.5430684089660645,
      "learning_rate": 9.057866287376536e-05,
      "loss": 0.3441,
      "step": 4992
    },
    {
      "epoch": 0.20188623934820626,
      "grad_norm": 5.07893705368042,
      "learning_rate": 9.057493659344347e-05,
      "loss": 0.1749,
      "step": 4993
    },
    {
      "epoch": 0.20192667320347327,
      "grad_norm": 7.406638145446777,
      "learning_rate": 9.057120965304624e-05,
      "loss": 0.2544,
      "step": 4994
    },
    {
      "epoch": 0.20196710705874027,
      "grad_norm": 4.849218368530273,
      "learning_rate": 9.05674820526343e-05,
      "loss": 0.1301,
      "step": 4995
    },
    {
      "epoch": 0.2020075409140073,
      "grad_norm": 5.435390949249268,
      "learning_rate": 9.056375379226829e-05,
      "loss": 0.1753,
      "step": 4996
    },
    {
      "epoch": 0.2020479747692743,
      "grad_norm": 4.273417949676514,
      "learning_rate": 9.056002487200889e-05,
      "loss": 0.2082,
      "step": 4997
    },
    {
      "epoch": 0.20208840862454133,
      "grad_norm": 5.250924110412598,
      "learning_rate": 9.055629529191671e-05,
      "loss": 0.1562,
      "step": 4998
    },
    {
      "epoch": 0.20212884247980833,
      "grad_norm": 6.563113689422607,
      "learning_rate": 9.055256505205247e-05,
      "loss": 0.2206,
      "step": 4999
    },
    {
      "epoch": 0.20216927633507537,
      "grad_norm": 12.311762809753418,
      "learning_rate": 9.054883415247682e-05,
      "loss": 0.2923,
      "step": 5000
    },
    {
      "epoch": 0.20220971019034237,
      "grad_norm": 5.746432304382324,
      "learning_rate": 9.054510259325047e-05,
      "loss": 0.2628,
      "step": 5001
    },
    {
      "epoch": 0.2022501440456094,
      "grad_norm": 3.9580867290496826,
      "learning_rate": 9.054137037443414e-05,
      "loss": 0.3317,
      "step": 5002
    },
    {
      "epoch": 0.2022905779008764,
      "grad_norm": 2.724975347518921,
      "learning_rate": 9.053763749608851e-05,
      "loss": 0.0964,
      "step": 5003
    },
    {
      "epoch": 0.20233101175614343,
      "grad_norm": 4.490112781524658,
      "learning_rate": 9.053390395827433e-05,
      "loss": 0.1763,
      "step": 5004
    },
    {
      "epoch": 0.20237144561141043,
      "grad_norm": 6.153409481048584,
      "learning_rate": 9.053016976105236e-05,
      "loss": 0.19,
      "step": 5005
    },
    {
      "epoch": 0.20241187946667744,
      "grad_norm": 3.0745859146118164,
      "learning_rate": 9.05264349044833e-05,
      "loss": 0.13,
      "step": 5006
    },
    {
      "epoch": 0.20245231332194447,
      "grad_norm": 4.865169048309326,
      "learning_rate": 9.052269938862792e-05,
      "loss": 0.0986,
      "step": 5007
    },
    {
      "epoch": 0.20249274717721147,
      "grad_norm": 4.215059280395508,
      "learning_rate": 9.051896321354701e-05,
      "loss": 0.1545,
      "step": 5008
    },
    {
      "epoch": 0.2025331810324785,
      "grad_norm": 5.419270992279053,
      "learning_rate": 9.051522637930134e-05,
      "loss": 0.1334,
      "step": 5009
    },
    {
      "epoch": 0.2025736148877455,
      "grad_norm": 7.398308753967285,
      "learning_rate": 9.051148888595172e-05,
      "loss": 0.1571,
      "step": 5010
    },
    {
      "epoch": 0.20261404874301253,
      "grad_norm": 6.929096698760986,
      "learning_rate": 9.050775073355892e-05,
      "loss": 0.1593,
      "step": 5011
    },
    {
      "epoch": 0.20265448259827953,
      "grad_norm": 8.072331428527832,
      "learning_rate": 9.050401192218377e-05,
      "loss": 0.2017,
      "step": 5012
    },
    {
      "epoch": 0.20269491645354656,
      "grad_norm": 5.625253200531006,
      "learning_rate": 9.050027245188708e-05,
      "loss": 0.1588,
      "step": 5013
    },
    {
      "epoch": 0.20273535030881357,
      "grad_norm": 3.8088014125823975,
      "learning_rate": 9.04965323227297e-05,
      "loss": 0.1578,
      "step": 5014
    },
    {
      "epoch": 0.2027757841640806,
      "grad_norm": 6.518123149871826,
      "learning_rate": 9.049279153477246e-05,
      "loss": 0.1368,
      "step": 5015
    },
    {
      "epoch": 0.2028162180193476,
      "grad_norm": 2.8052139282226562,
      "learning_rate": 9.048905008807623e-05,
      "loss": 0.1661,
      "step": 5016
    },
    {
      "epoch": 0.2028566518746146,
      "grad_norm": 4.19486665725708,
      "learning_rate": 9.048530798270187e-05,
      "loss": 0.119,
      "step": 5017
    },
    {
      "epoch": 0.20289708572988163,
      "grad_norm": 2.52826189994812,
      "learning_rate": 9.048156521871024e-05,
      "loss": 0.1342,
      "step": 5018
    },
    {
      "epoch": 0.20293751958514863,
      "grad_norm": 3.0745391845703125,
      "learning_rate": 9.047782179616227e-05,
      "loss": 0.1299,
      "step": 5019
    },
    {
      "epoch": 0.20297795344041566,
      "grad_norm": 3.5455760955810547,
      "learning_rate": 9.047407771511882e-05,
      "loss": 0.1105,
      "step": 5020
    },
    {
      "epoch": 0.20301838729568267,
      "grad_norm": 3.9209983348846436,
      "learning_rate": 9.047033297564083e-05,
      "loss": 0.0701,
      "step": 5021
    },
    {
      "epoch": 0.2030588211509497,
      "grad_norm": 4.815310001373291,
      "learning_rate": 9.046658757778917e-05,
      "loss": 0.1526,
      "step": 5022
    },
    {
      "epoch": 0.2030992550062167,
      "grad_norm": 6.456916809082031,
      "learning_rate": 9.04628415216248e-05,
      "loss": 0.1939,
      "step": 5023
    },
    {
      "epoch": 0.20313968886148373,
      "grad_norm": 2.3659822940826416,
      "learning_rate": 9.045909480720868e-05,
      "loss": 0.1656,
      "step": 5024
    },
    {
      "epoch": 0.20318012271675073,
      "grad_norm": 5.2312211990356445,
      "learning_rate": 9.045534743460176e-05,
      "loss": 0.27,
      "step": 5025
    },
    {
      "epoch": 0.20322055657201776,
      "grad_norm": 6.57778787612915,
      "learning_rate": 9.045159940386495e-05,
      "loss": 0.3232,
      "step": 5026
    },
    {
      "epoch": 0.20326099042728477,
      "grad_norm": 5.794137954711914,
      "learning_rate": 9.044785071505928e-05,
      "loss": 0.1224,
      "step": 5027
    },
    {
      "epoch": 0.20330142428255177,
      "grad_norm": 8.891010284423828,
      "learning_rate": 9.044410136824571e-05,
      "loss": 0.1667,
      "step": 5028
    },
    {
      "epoch": 0.2033418581378188,
      "grad_norm": 4.734241962432861,
      "learning_rate": 9.044035136348523e-05,
      "loss": 0.1689,
      "step": 5029
    },
    {
      "epoch": 0.2033822919930858,
      "grad_norm": 3.5831189155578613,
      "learning_rate": 9.043660070083886e-05,
      "loss": 0.1724,
      "step": 5030
    },
    {
      "epoch": 0.20342272584835283,
      "grad_norm": 3.9661717414855957,
      "learning_rate": 9.04328493803676e-05,
      "loss": 0.1547,
      "step": 5031
    },
    {
      "epoch": 0.20346315970361983,
      "grad_norm": 5.650935173034668,
      "learning_rate": 9.04290974021325e-05,
      "loss": 0.1527,
      "step": 5032
    },
    {
      "epoch": 0.20350359355888686,
      "grad_norm": 6.344314098358154,
      "learning_rate": 9.042534476619457e-05,
      "loss": 0.1704,
      "step": 5033
    },
    {
      "epoch": 0.20354402741415387,
      "grad_norm": 3.8565728664398193,
      "learning_rate": 9.042159147261487e-05,
      "loss": 0.1686,
      "step": 5034
    },
    {
      "epoch": 0.2035844612694209,
      "grad_norm": 3.311082124710083,
      "learning_rate": 9.041783752145448e-05,
      "loss": 0.2097,
      "step": 5035
    },
    {
      "epoch": 0.2036248951246879,
      "grad_norm": 6.729128360748291,
      "learning_rate": 9.041408291277442e-05,
      "loss": 0.3325,
      "step": 5036
    },
    {
      "epoch": 0.20366532897995493,
      "grad_norm": 4.987082004547119,
      "learning_rate": 9.041032764663581e-05,
      "loss": 0.1034,
      "step": 5037
    },
    {
      "epoch": 0.20370576283522193,
      "grad_norm": 3.453880548477173,
      "learning_rate": 9.040657172309975e-05,
      "loss": 0.1523,
      "step": 5038
    },
    {
      "epoch": 0.20374619669048893,
      "grad_norm": 6.028136253356934,
      "learning_rate": 9.040281514222731e-05,
      "loss": 0.2432,
      "step": 5039
    },
    {
      "epoch": 0.20378663054575596,
      "grad_norm": 3.392496109008789,
      "learning_rate": 9.039905790407962e-05,
      "loss": 0.2641,
      "step": 5040
    },
    {
      "epoch": 0.20382706440102297,
      "grad_norm": 3.4824423789978027,
      "learning_rate": 9.039530000871777e-05,
      "loss": 0.1395,
      "step": 5041
    },
    {
      "epoch": 0.20386749825629,
      "grad_norm": 6.907786846160889,
      "learning_rate": 9.039154145620295e-05,
      "loss": 0.2529,
      "step": 5042
    },
    {
      "epoch": 0.203907932111557,
      "grad_norm": 4.772083282470703,
      "learning_rate": 9.038778224659626e-05,
      "loss": 0.1369,
      "step": 5043
    },
    {
      "epoch": 0.20394836596682403,
      "grad_norm": 4.049684047698975,
      "learning_rate": 9.038402237995888e-05,
      "loss": 0.1195,
      "step": 5044
    },
    {
      "epoch": 0.20398879982209103,
      "grad_norm": 2.398395299911499,
      "learning_rate": 9.038026185635195e-05,
      "loss": 0.0906,
      "step": 5045
    },
    {
      "epoch": 0.20402923367735806,
      "grad_norm": 5.306074619293213,
      "learning_rate": 9.037650067583669e-05,
      "loss": 0.2588,
      "step": 5046
    },
    {
      "epoch": 0.20406966753262507,
      "grad_norm": 2.2749834060668945,
      "learning_rate": 9.037273883847424e-05,
      "loss": 0.1134,
      "step": 5047
    },
    {
      "epoch": 0.2041101013878921,
      "grad_norm": 4.601478099822998,
      "learning_rate": 9.036897634432582e-05,
      "loss": 0.2094,
      "step": 5048
    },
    {
      "epoch": 0.2041505352431591,
      "grad_norm": 5.901350975036621,
      "learning_rate": 9.036521319345263e-05,
      "loss": 0.1618,
      "step": 5049
    },
    {
      "epoch": 0.2041909690984261,
      "grad_norm": 4.524260520935059,
      "learning_rate": 9.036144938591591e-05,
      "loss": 0.165,
      "step": 5050
    },
    {
      "epoch": 0.20423140295369313,
      "grad_norm": 8.902656555175781,
      "learning_rate": 9.035768492177685e-05,
      "loss": 0.1663,
      "step": 5051
    },
    {
      "epoch": 0.20427183680896013,
      "grad_norm": 7.721915245056152,
      "learning_rate": 9.035391980109673e-05,
      "loss": 0.3691,
      "step": 5052
    },
    {
      "epoch": 0.20431227066422716,
      "grad_norm": 2.370150089263916,
      "learning_rate": 9.035015402393681e-05,
      "loss": 0.1569,
      "step": 5053
    },
    {
      "epoch": 0.20435270451949417,
      "grad_norm": 7.344154357910156,
      "learning_rate": 9.034638759035829e-05,
      "loss": 0.3358,
      "step": 5054
    },
    {
      "epoch": 0.2043931383747612,
      "grad_norm": 7.032852649688721,
      "learning_rate": 9.03426205004225e-05,
      "loss": 0.2287,
      "step": 5055
    },
    {
      "epoch": 0.2044335722300282,
      "grad_norm": 10.133667945861816,
      "learning_rate": 9.03388527541907e-05,
      "loss": 0.3416,
      "step": 5056
    },
    {
      "epoch": 0.20447400608529523,
      "grad_norm": 10.605395317077637,
      "learning_rate": 9.03350843517242e-05,
      "loss": 0.2494,
      "step": 5057
    },
    {
      "epoch": 0.20451443994056223,
      "grad_norm": 5.671846866607666,
      "learning_rate": 9.033131529308427e-05,
      "loss": 0.1467,
      "step": 5058
    },
    {
      "epoch": 0.20455487379582923,
      "grad_norm": 6.101764678955078,
      "learning_rate": 9.032754557833226e-05,
      "loss": 0.1929,
      "step": 5059
    },
    {
      "epoch": 0.20459530765109626,
      "grad_norm": 2.4870753288269043,
      "learning_rate": 9.032377520752948e-05,
      "loss": 0.1824,
      "step": 5060
    },
    {
      "epoch": 0.20463574150636327,
      "grad_norm": 4.1082234382629395,
      "learning_rate": 9.032000418073727e-05,
      "loss": 0.2037,
      "step": 5061
    },
    {
      "epoch": 0.2046761753616303,
      "grad_norm": 2.3881821632385254,
      "learning_rate": 9.0316232498017e-05,
      "loss": 0.1119,
      "step": 5062
    },
    {
      "epoch": 0.2047166092168973,
      "grad_norm": 3.7300350666046143,
      "learning_rate": 9.031246015942998e-05,
      "loss": 0.1998,
      "step": 5063
    },
    {
      "epoch": 0.20475704307216433,
      "grad_norm": 5.238885879516602,
      "learning_rate": 9.03086871650376e-05,
      "loss": 0.1318,
      "step": 5064
    },
    {
      "epoch": 0.20479747692743133,
      "grad_norm": 2.922873020172119,
      "learning_rate": 9.030491351490126e-05,
      "loss": 0.1921,
      "step": 5065
    },
    {
      "epoch": 0.20483791078269836,
      "grad_norm": 2.1167852878570557,
      "learning_rate": 9.030113920908233e-05,
      "loss": 0.0928,
      "step": 5066
    },
    {
      "epoch": 0.20487834463796536,
      "grad_norm": 8.551018714904785,
      "learning_rate": 9.029736424764221e-05,
      "loss": 0.2516,
      "step": 5067
    },
    {
      "epoch": 0.2049187784932324,
      "grad_norm": 3.249577522277832,
      "learning_rate": 9.029358863064233e-05,
      "loss": 0.2394,
      "step": 5068
    },
    {
      "epoch": 0.2049592123484994,
      "grad_norm": 8.386224746704102,
      "learning_rate": 9.028981235814407e-05,
      "loss": 0.1811,
      "step": 5069
    },
    {
      "epoch": 0.2049996462037664,
      "grad_norm": 7.546560287475586,
      "learning_rate": 9.02860354302089e-05,
      "loss": 0.3166,
      "step": 5070
    },
    {
      "epoch": 0.20504008005903343,
      "grad_norm": 4.492493629455566,
      "learning_rate": 9.028225784689827e-05,
      "loss": 0.1755,
      "step": 5071
    },
    {
      "epoch": 0.20508051391430043,
      "grad_norm": 4.1006317138671875,
      "learning_rate": 9.02784796082736e-05,
      "loss": 0.2681,
      "step": 5072
    },
    {
      "epoch": 0.20512094776956746,
      "grad_norm": 4.085662364959717,
      "learning_rate": 9.027470071439637e-05,
      "loss": 0.1499,
      "step": 5073
    },
    {
      "epoch": 0.20516138162483447,
      "grad_norm": 3.7799081802368164,
      "learning_rate": 9.027092116532807e-05,
      "loss": 0.2078,
      "step": 5074
    },
    {
      "epoch": 0.2052018154801015,
      "grad_norm": 2.188080310821533,
      "learning_rate": 9.026714096113016e-05,
      "loss": 0.0985,
      "step": 5075
    },
    {
      "epoch": 0.2052422493353685,
      "grad_norm": 2.313584327697754,
      "learning_rate": 9.026336010186415e-05,
      "loss": 0.0635,
      "step": 5076
    },
    {
      "epoch": 0.20528268319063553,
      "grad_norm": 5.306520938873291,
      "learning_rate": 9.025957858759157e-05,
      "loss": 0.2093,
      "step": 5077
    },
    {
      "epoch": 0.20532311704590253,
      "grad_norm": 3.3482720851898193,
      "learning_rate": 9.02557964183739e-05,
      "loss": 0.2814,
      "step": 5078
    },
    {
      "epoch": 0.20536355090116956,
      "grad_norm": 4.922545433044434,
      "learning_rate": 9.025201359427267e-05,
      "loss": 0.2315,
      "step": 5079
    },
    {
      "epoch": 0.20540398475643656,
      "grad_norm": 5.448150157928467,
      "learning_rate": 9.024823011534944e-05,
      "loss": 0.1968,
      "step": 5080
    },
    {
      "epoch": 0.20544441861170357,
      "grad_norm": 7.637961387634277,
      "learning_rate": 9.024444598166575e-05,
      "loss": 0.3275,
      "step": 5081
    },
    {
      "epoch": 0.2054848524669706,
      "grad_norm": 3.6939098834991455,
      "learning_rate": 9.024066119328318e-05,
      "loss": 0.1408,
      "step": 5082
    },
    {
      "epoch": 0.2055252863222376,
      "grad_norm": 2.709108591079712,
      "learning_rate": 9.023687575026328e-05,
      "loss": 0.0733,
      "step": 5083
    },
    {
      "epoch": 0.20556572017750463,
      "grad_norm": 4.589844703674316,
      "learning_rate": 9.023308965266762e-05,
      "loss": 0.1544,
      "step": 5084
    },
    {
      "epoch": 0.20560615403277163,
      "grad_norm": 4.799004554748535,
      "learning_rate": 9.02293029005578e-05,
      "loss": 0.1186,
      "step": 5085
    },
    {
      "epoch": 0.20564658788803866,
      "grad_norm": 3.187417507171631,
      "learning_rate": 9.022551549399546e-05,
      "loss": 0.1297,
      "step": 5086
    },
    {
      "epoch": 0.20568702174330566,
      "grad_norm": 6.778784275054932,
      "learning_rate": 9.022172743304216e-05,
      "loss": 0.2054,
      "step": 5087
    },
    {
      "epoch": 0.2057274555985727,
      "grad_norm": 5.762674331665039,
      "learning_rate": 9.021793871775956e-05,
      "loss": 0.2763,
      "step": 5088
    },
    {
      "epoch": 0.2057678894538397,
      "grad_norm": 4.266824722290039,
      "learning_rate": 9.021414934820928e-05,
      "loss": 0.1308,
      "step": 5089
    },
    {
      "epoch": 0.20580832330910673,
      "grad_norm": 4.970431327819824,
      "learning_rate": 9.021035932445298e-05,
      "loss": 0.1342,
      "step": 5090
    },
    {
      "epoch": 0.20584875716437373,
      "grad_norm": 3.1687402725219727,
      "learning_rate": 9.02065686465523e-05,
      "loss": 0.1681,
      "step": 5091
    },
    {
      "epoch": 0.20588919101964073,
      "grad_norm": 5.720516204833984,
      "learning_rate": 9.020277731456891e-05,
      "loss": 0.1569,
      "step": 5092
    },
    {
      "epoch": 0.20592962487490776,
      "grad_norm": 11.879281044006348,
      "learning_rate": 9.01989853285645e-05,
      "loss": 0.1294,
      "step": 5093
    },
    {
      "epoch": 0.20597005873017477,
      "grad_norm": 5.132293701171875,
      "learning_rate": 9.019519268860075e-05,
      "loss": 0.1051,
      "step": 5094
    },
    {
      "epoch": 0.2060104925854418,
      "grad_norm": 7.284663677215576,
      "learning_rate": 9.019139939473934e-05,
      "loss": 0.1752,
      "step": 5095
    },
    {
      "epoch": 0.2060509264407088,
      "grad_norm": 2.908562421798706,
      "learning_rate": 9.018760544704202e-05,
      "loss": 0.0992,
      "step": 5096
    },
    {
      "epoch": 0.20609136029597583,
      "grad_norm": 2.8539507389068604,
      "learning_rate": 9.018381084557048e-05,
      "loss": 0.1524,
      "step": 5097
    },
    {
      "epoch": 0.20613179415124283,
      "grad_norm": 7.99936056137085,
      "learning_rate": 9.018001559038645e-05,
      "loss": 0.169,
      "step": 5098
    },
    {
      "epoch": 0.20617222800650986,
      "grad_norm": 4.569436073303223,
      "learning_rate": 9.017621968155169e-05,
      "loss": 0.4754,
      "step": 5099
    },
    {
      "epoch": 0.20621266186177686,
      "grad_norm": 5.3552069664001465,
      "learning_rate": 9.017242311912794e-05,
      "loss": 0.1759,
      "step": 5100
    },
    {
      "epoch": 0.2062530957170439,
      "grad_norm": 5.977463245391846,
      "learning_rate": 9.016862590317697e-05,
      "loss": 0.2521,
      "step": 5101
    },
    {
      "epoch": 0.2062935295723109,
      "grad_norm": 4.793139457702637,
      "learning_rate": 9.016482803376054e-05,
      "loss": 0.1169,
      "step": 5102
    },
    {
      "epoch": 0.2063339634275779,
      "grad_norm": 3.2142550945281982,
      "learning_rate": 9.016102951094044e-05,
      "loss": 0.2642,
      "step": 5103
    },
    {
      "epoch": 0.20637439728284493,
      "grad_norm": 6.640443325042725,
      "learning_rate": 9.015723033477846e-05,
      "loss": 0.2444,
      "step": 5104
    },
    {
      "epoch": 0.20641483113811193,
      "grad_norm": 8.405375480651855,
      "learning_rate": 9.015343050533642e-05,
      "loss": 0.1328,
      "step": 5105
    },
    {
      "epoch": 0.20645526499337896,
      "grad_norm": 6.172577381134033,
      "learning_rate": 9.014963002267614e-05,
      "loss": 0.2984,
      "step": 5106
    },
    {
      "epoch": 0.20649569884864596,
      "grad_norm": 5.718466281890869,
      "learning_rate": 9.01458288868594e-05,
      "loss": 0.0963,
      "step": 5107
    },
    {
      "epoch": 0.206536132703913,
      "grad_norm": 5.303503036499023,
      "learning_rate": 9.01420270979481e-05,
      "loss": 0.1873,
      "step": 5108
    },
    {
      "epoch": 0.20657656655918,
      "grad_norm": 4.5789794921875,
      "learning_rate": 9.013822465600406e-05,
      "loss": 0.2892,
      "step": 5109
    },
    {
      "epoch": 0.20661700041444703,
      "grad_norm": 3.5041208267211914,
      "learning_rate": 9.013442156108913e-05,
      "loss": 0.1878,
      "step": 5110
    },
    {
      "epoch": 0.20665743426971403,
      "grad_norm": 11.777400970458984,
      "learning_rate": 9.013061781326517e-05,
      "loss": 0.2946,
      "step": 5111
    },
    {
      "epoch": 0.20669786812498106,
      "grad_norm": 3.5302114486694336,
      "learning_rate": 9.012681341259409e-05,
      "loss": 0.1813,
      "step": 5112
    },
    {
      "epoch": 0.20673830198024806,
      "grad_norm": 4.845082759857178,
      "learning_rate": 9.012300835913777e-05,
      "loss": 0.2648,
      "step": 5113
    },
    {
      "epoch": 0.20677873583551507,
      "grad_norm": 3.5293214321136475,
      "learning_rate": 9.01192026529581e-05,
      "loss": 0.2747,
      "step": 5114
    },
    {
      "epoch": 0.2068191696907821,
      "grad_norm": 6.99151611328125,
      "learning_rate": 9.0115396294117e-05,
      "loss": 0.2589,
      "step": 5115
    },
    {
      "epoch": 0.2068596035460491,
      "grad_norm": 6.153543949127197,
      "learning_rate": 9.01115892826764e-05,
      "loss": 0.314,
      "step": 5116
    },
    {
      "epoch": 0.20690003740131613,
      "grad_norm": 3.7543251514434814,
      "learning_rate": 9.010778161869821e-05,
      "loss": 0.1974,
      "step": 5117
    },
    {
      "epoch": 0.20694047125658313,
      "grad_norm": 5.8487443923950195,
      "learning_rate": 9.010397330224438e-05,
      "loss": 0.2032,
      "step": 5118
    },
    {
      "epoch": 0.20698090511185016,
      "grad_norm": 4.388674259185791,
      "learning_rate": 9.010016433337688e-05,
      "loss": 0.2109,
      "step": 5119
    },
    {
      "epoch": 0.20702133896711716,
      "grad_norm": 7.17492151260376,
      "learning_rate": 9.009635471215768e-05,
      "loss": 0.2934,
      "step": 5120
    },
    {
      "epoch": 0.2070617728223842,
      "grad_norm": 9.107199668884277,
      "learning_rate": 9.009254443864872e-05,
      "loss": 0.2698,
      "step": 5121
    },
    {
      "epoch": 0.2071022066776512,
      "grad_norm": 7.7554731369018555,
      "learning_rate": 9.0088733512912e-05,
      "loss": 0.289,
      "step": 5122
    },
    {
      "epoch": 0.2071426405329182,
      "grad_norm": 4.801868438720703,
      "learning_rate": 9.008492193500955e-05,
      "loss": 0.1833,
      "step": 5123
    },
    {
      "epoch": 0.20718307438818523,
      "grad_norm": 7.1920485496521,
      "learning_rate": 9.008110970500332e-05,
      "loss": 0.1766,
      "step": 5124
    },
    {
      "epoch": 0.20722350824345223,
      "grad_norm": 2.556995153427124,
      "learning_rate": 9.007729682295536e-05,
      "loss": 0.1584,
      "step": 5125
    },
    {
      "epoch": 0.20726394209871926,
      "grad_norm": 3.341778039932251,
      "learning_rate": 9.00734832889277e-05,
      "loss": 0.1564,
      "step": 5126
    },
    {
      "epoch": 0.20730437595398626,
      "grad_norm": 4.8506269454956055,
      "learning_rate": 9.006966910298238e-05,
      "loss": 0.1732,
      "step": 5127
    },
    {
      "epoch": 0.2073448098092533,
      "grad_norm": 2.6603708267211914,
      "learning_rate": 9.006585426518143e-05,
      "loss": 0.175,
      "step": 5128
    },
    {
      "epoch": 0.2073852436645203,
      "grad_norm": 8.639480590820312,
      "learning_rate": 9.006203877558694e-05,
      "loss": 0.2031,
      "step": 5129
    },
    {
      "epoch": 0.20742567751978733,
      "grad_norm": 6.980950355529785,
      "learning_rate": 9.005822263426096e-05,
      "loss": 0.2248,
      "step": 5130
    },
    {
      "epoch": 0.20746611137505433,
      "grad_norm": 5.03322172164917,
      "learning_rate": 9.005440584126557e-05,
      "loss": 0.182,
      "step": 5131
    },
    {
      "epoch": 0.20750654523032136,
      "grad_norm": 4.124065399169922,
      "learning_rate": 9.005058839666285e-05,
      "loss": 0.1743,
      "step": 5132
    },
    {
      "epoch": 0.20754697908558836,
      "grad_norm": 6.076673984527588,
      "learning_rate": 9.004677030051494e-05,
      "loss": 0.2066,
      "step": 5133
    },
    {
      "epoch": 0.20758741294085536,
      "grad_norm": 3.6425371170043945,
      "learning_rate": 9.004295155288391e-05,
      "loss": 0.2126,
      "step": 5134
    },
    {
      "epoch": 0.2076278467961224,
      "grad_norm": 6.658566951751709,
      "learning_rate": 9.003913215383193e-05,
      "loss": 0.1551,
      "step": 5135
    },
    {
      "epoch": 0.2076682806513894,
      "grad_norm": 6.420670509338379,
      "learning_rate": 9.00353121034211e-05,
      "loss": 0.2589,
      "step": 5136
    },
    {
      "epoch": 0.20770871450665643,
      "grad_norm": 6.843458652496338,
      "learning_rate": 9.003149140171357e-05,
      "loss": 0.1803,
      "step": 5137
    },
    {
      "epoch": 0.20774914836192343,
      "grad_norm": 2.000040292739868,
      "learning_rate": 9.002767004877151e-05,
      "loss": 0.1211,
      "step": 5138
    },
    {
      "epoch": 0.20778958221719046,
      "grad_norm": 11.140373229980469,
      "learning_rate": 9.002384804465707e-05,
      "loss": 0.2686,
      "step": 5139
    },
    {
      "epoch": 0.20783001607245746,
      "grad_norm": 6.015961170196533,
      "learning_rate": 9.002002538943241e-05,
      "loss": 0.2017,
      "step": 5140
    },
    {
      "epoch": 0.2078704499277245,
      "grad_norm": 3.812528371810913,
      "learning_rate": 9.001620208315978e-05,
      "loss": 0.1705,
      "step": 5141
    },
    {
      "epoch": 0.2079108837829915,
      "grad_norm": 3.242722749710083,
      "learning_rate": 9.001237812590131e-05,
      "loss": 0.0976,
      "step": 5142
    },
    {
      "epoch": 0.20795131763825853,
      "grad_norm": 5.380507946014404,
      "learning_rate": 9.000855351771925e-05,
      "loss": 0.1077,
      "step": 5143
    },
    {
      "epoch": 0.20799175149352553,
      "grad_norm": 4.541932106018066,
      "learning_rate": 9.000472825867579e-05,
      "loss": 0.148,
      "step": 5144
    },
    {
      "epoch": 0.20803218534879253,
      "grad_norm": 4.148437023162842,
      "learning_rate": 9.000090234883318e-05,
      "loss": 0.219,
      "step": 5145
    },
    {
      "epoch": 0.20807261920405956,
      "grad_norm": 3.47517728805542,
      "learning_rate": 8.999707578825366e-05,
      "loss": 0.0578,
      "step": 5146
    },
    {
      "epoch": 0.20811305305932656,
      "grad_norm": 7.506556987762451,
      "learning_rate": 8.999324857699947e-05,
      "loss": 0.2278,
      "step": 5147
    },
    {
      "epoch": 0.2081534869145936,
      "grad_norm": 3.4022116661071777,
      "learning_rate": 8.998942071513288e-05,
      "loss": 0.1141,
      "step": 5148
    },
    {
      "epoch": 0.2081939207698606,
      "grad_norm": 2.472102642059326,
      "learning_rate": 8.998559220271615e-05,
      "loss": 0.081,
      "step": 5149
    },
    {
      "epoch": 0.20823435462512763,
      "grad_norm": 2.7043418884277344,
      "learning_rate": 8.998176303981157e-05,
      "loss": 0.1522,
      "step": 5150
    },
    {
      "epoch": 0.20827478848039463,
      "grad_norm": 4.737821578979492,
      "learning_rate": 8.997793322648145e-05,
      "loss": 0.2024,
      "step": 5151
    },
    {
      "epoch": 0.20831522233566166,
      "grad_norm": 3.910008668899536,
      "learning_rate": 8.997410276278806e-05,
      "loss": 0.2587,
      "step": 5152
    },
    {
      "epoch": 0.20835565619092866,
      "grad_norm": 4.677521705627441,
      "learning_rate": 8.997027164879373e-05,
      "loss": 0.1538,
      "step": 5153
    },
    {
      "epoch": 0.2083960900461957,
      "grad_norm": 6.011166572570801,
      "learning_rate": 8.99664398845608e-05,
      "loss": 0.1429,
      "step": 5154
    },
    {
      "epoch": 0.2084365239014627,
      "grad_norm": 10.45859146118164,
      "learning_rate": 8.996260747015158e-05,
      "loss": 0.3313,
      "step": 5155
    },
    {
      "epoch": 0.2084769577567297,
      "grad_norm": 3.7005436420440674,
      "learning_rate": 8.995877440562843e-05,
      "loss": 0.2323,
      "step": 5156
    },
    {
      "epoch": 0.20851739161199673,
      "grad_norm": 6.663586616516113,
      "learning_rate": 8.99549406910537e-05,
      "loss": 0.2523,
      "step": 5157
    },
    {
      "epoch": 0.20855782546726373,
      "grad_norm": 10.638587951660156,
      "learning_rate": 8.995110632648977e-05,
      "loss": 0.2876,
      "step": 5158
    },
    {
      "epoch": 0.20859825932253076,
      "grad_norm": 2.165581226348877,
      "learning_rate": 8.9947271311999e-05,
      "loss": 0.0896,
      "step": 5159
    },
    {
      "epoch": 0.20863869317779776,
      "grad_norm": 7.431328296661377,
      "learning_rate": 8.99434356476438e-05,
      "loss": 0.2924,
      "step": 5160
    },
    {
      "epoch": 0.2086791270330648,
      "grad_norm": 4.559417724609375,
      "learning_rate": 8.993959933348655e-05,
      "loss": 0.1269,
      "step": 5161
    },
    {
      "epoch": 0.2087195608883318,
      "grad_norm": 7.832269191741943,
      "learning_rate": 8.993576236958966e-05,
      "loss": 0.3183,
      "step": 5162
    },
    {
      "epoch": 0.20875999474359883,
      "grad_norm": 3.582444906234741,
      "learning_rate": 8.993192475601555e-05,
      "loss": 0.1784,
      "step": 5163
    },
    {
      "epoch": 0.20880042859886583,
      "grad_norm": 3.882697343826294,
      "learning_rate": 8.992808649282666e-05,
      "loss": 0.21,
      "step": 5164
    },
    {
      "epoch": 0.20884086245413286,
      "grad_norm": 1.9535560607910156,
      "learning_rate": 8.992424758008544e-05,
      "loss": 0.0962,
      "step": 5165
    },
    {
      "epoch": 0.20888129630939986,
      "grad_norm": 5.63818883895874,
      "learning_rate": 8.99204080178543e-05,
      "loss": 0.2362,
      "step": 5166
    },
    {
      "epoch": 0.20892173016466686,
      "grad_norm": 5.827169418334961,
      "learning_rate": 8.991656780619575e-05,
      "loss": 0.1764,
      "step": 5167
    },
    {
      "epoch": 0.2089621640199339,
      "grad_norm": 3.5099830627441406,
      "learning_rate": 8.991272694517226e-05,
      "loss": 0.1995,
      "step": 5168
    },
    {
      "epoch": 0.2090025978752009,
      "grad_norm": 2.489595890045166,
      "learning_rate": 8.990888543484626e-05,
      "loss": 0.073,
      "step": 5169
    },
    {
      "epoch": 0.20904303173046793,
      "grad_norm": 3.4847640991210938,
      "learning_rate": 8.99050432752803e-05,
      "loss": 0.3353,
      "step": 5170
    },
    {
      "epoch": 0.20908346558573493,
      "grad_norm": 8.066113471984863,
      "learning_rate": 8.990120046653686e-05,
      "loss": 0.2139,
      "step": 5171
    },
    {
      "epoch": 0.20912389944100196,
      "grad_norm": 3.075775623321533,
      "learning_rate": 8.989735700867846e-05,
      "loss": 0.1183,
      "step": 5172
    },
    {
      "epoch": 0.20916433329626896,
      "grad_norm": 6.771798610687256,
      "learning_rate": 8.989351290176764e-05,
      "loss": 0.2669,
      "step": 5173
    },
    {
      "epoch": 0.209204767151536,
      "grad_norm": 11.143489837646484,
      "learning_rate": 8.98896681458669e-05,
      "loss": 0.2558,
      "step": 5174
    },
    {
      "epoch": 0.209245201006803,
      "grad_norm": 8.185620307922363,
      "learning_rate": 8.988582274103881e-05,
      "loss": 0.1775,
      "step": 5175
    },
    {
      "epoch": 0.20928563486207002,
      "grad_norm": 2.2666239738464355,
      "learning_rate": 8.988197668734595e-05,
      "loss": 0.1209,
      "step": 5176
    },
    {
      "epoch": 0.20932606871733703,
      "grad_norm": 6.4340949058532715,
      "learning_rate": 8.987812998485083e-05,
      "loss": 0.1871,
      "step": 5177
    },
    {
      "epoch": 0.20936650257260403,
      "grad_norm": 4.126743793487549,
      "learning_rate": 8.987428263361608e-05,
      "loss": 0.1681,
      "step": 5178
    },
    {
      "epoch": 0.20940693642787106,
      "grad_norm": 4.540679931640625,
      "learning_rate": 8.987043463370428e-05,
      "loss": 0.1404,
      "step": 5179
    },
    {
      "epoch": 0.20944737028313806,
      "grad_norm": 2.98079252243042,
      "learning_rate": 8.986658598517801e-05,
      "loss": 0.1328,
      "step": 5180
    },
    {
      "epoch": 0.2094878041384051,
      "grad_norm": 0.8027064800262451,
      "learning_rate": 8.986273668809987e-05,
      "loss": 0.0276,
      "step": 5181
    },
    {
      "epoch": 0.2095282379936721,
      "grad_norm": 2.9929373264312744,
      "learning_rate": 8.985888674253253e-05,
      "loss": 0.1635,
      "step": 5182
    },
    {
      "epoch": 0.20956867184893913,
      "grad_norm": 4.276904106140137,
      "learning_rate": 8.985503614853859e-05,
      "loss": 0.1499,
      "step": 5183
    },
    {
      "epoch": 0.20960910570420613,
      "grad_norm": 3.216872453689575,
      "learning_rate": 8.985118490618069e-05,
      "loss": 0.0845,
      "step": 5184
    },
    {
      "epoch": 0.20964953955947316,
      "grad_norm": 3.1103603839874268,
      "learning_rate": 8.984733301552147e-05,
      "loss": 0.0835,
      "step": 5185
    },
    {
      "epoch": 0.20968997341474016,
      "grad_norm": 5.571695804595947,
      "learning_rate": 8.984348047662362e-05,
      "loss": 0.0952,
      "step": 5186
    },
    {
      "epoch": 0.2097304072700072,
      "grad_norm": 5.296355724334717,
      "learning_rate": 8.98396272895498e-05,
      "loss": 0.1476,
      "step": 5187
    },
    {
      "epoch": 0.2097708411252742,
      "grad_norm": 4.0923967361450195,
      "learning_rate": 8.983577345436269e-05,
      "loss": 0.101,
      "step": 5188
    },
    {
      "epoch": 0.2098112749805412,
      "grad_norm": 12.425406455993652,
      "learning_rate": 8.983191897112501e-05,
      "loss": 0.3184,
      "step": 5189
    },
    {
      "epoch": 0.20985170883580823,
      "grad_norm": 9.016788482666016,
      "learning_rate": 8.982806383989944e-05,
      "loss": 0.2363,
      "step": 5190
    },
    {
      "epoch": 0.20989214269107523,
      "grad_norm": 3.5091428756713867,
      "learning_rate": 8.98242080607487e-05,
      "loss": 0.1367,
      "step": 5191
    },
    {
      "epoch": 0.20993257654634226,
      "grad_norm": 4.968768119812012,
      "learning_rate": 8.98203516337355e-05,
      "loss": 0.0707,
      "step": 5192
    },
    {
      "epoch": 0.20997301040160926,
      "grad_norm": 6.524477481842041,
      "learning_rate": 8.98164945589226e-05,
      "loss": 0.1821,
      "step": 5193
    },
    {
      "epoch": 0.2100134442568763,
      "grad_norm": 5.993956565856934,
      "learning_rate": 8.981263683637274e-05,
      "loss": 0.2766,
      "step": 5194
    },
    {
      "epoch": 0.2100538781121433,
      "grad_norm": 4.1908135414123535,
      "learning_rate": 8.980877846614868e-05,
      "loss": 0.1851,
      "step": 5195
    },
    {
      "epoch": 0.21009431196741032,
      "grad_norm": 4.483226776123047,
      "learning_rate": 8.980491944831319e-05,
      "loss": 0.1936,
      "step": 5196
    },
    {
      "epoch": 0.21013474582267733,
      "grad_norm": 13.27910327911377,
      "learning_rate": 8.980105978292904e-05,
      "loss": 0.2522,
      "step": 5197
    },
    {
      "epoch": 0.21017517967794433,
      "grad_norm": 6.164882659912109,
      "learning_rate": 8.979719947005904e-05,
      "loss": 0.2122,
      "step": 5198
    },
    {
      "epoch": 0.21021561353321136,
      "grad_norm": 8.04697036743164,
      "learning_rate": 8.979333850976594e-05,
      "loss": 0.1757,
      "step": 5199
    },
    {
      "epoch": 0.21025604738847836,
      "grad_norm": 6.468683242797852,
      "learning_rate": 8.978947690211261e-05,
      "loss": 0.1756,
      "step": 5200
    },
    {
      "epoch": 0.2102964812437454,
      "grad_norm": 9.545432090759277,
      "learning_rate": 8.978561464716184e-05,
      "loss": 0.2057,
      "step": 5201
    },
    {
      "epoch": 0.2103369150990124,
      "grad_norm": 7.7294464111328125,
      "learning_rate": 8.978175174497648e-05,
      "loss": 0.2922,
      "step": 5202
    },
    {
      "epoch": 0.21037734895427943,
      "grad_norm": 3.7587108612060547,
      "learning_rate": 8.977788819561934e-05,
      "loss": 0.3849,
      "step": 5203
    },
    {
      "epoch": 0.21041778280954643,
      "grad_norm": 12.01777458190918,
      "learning_rate": 8.977402399915329e-05,
      "loss": 0.2329,
      "step": 5204
    },
    {
      "epoch": 0.21045821666481346,
      "grad_norm": 2.4164958000183105,
      "learning_rate": 8.97701591556412e-05,
      "loss": 0.0937,
      "step": 5205
    },
    {
      "epoch": 0.21049865052008046,
      "grad_norm": 6.585228443145752,
      "learning_rate": 8.976629366514595e-05,
      "loss": 0.1799,
      "step": 5206
    },
    {
      "epoch": 0.2105390843753475,
      "grad_norm": 6.241184711456299,
      "learning_rate": 8.976242752773038e-05,
      "loss": 0.2108,
      "step": 5207
    },
    {
      "epoch": 0.2105795182306145,
      "grad_norm": 2.501450777053833,
      "learning_rate": 8.975856074345745e-05,
      "loss": 0.2406,
      "step": 5208
    },
    {
      "epoch": 0.2106199520858815,
      "grad_norm": 5.597461700439453,
      "learning_rate": 8.975469331239001e-05,
      "loss": 0.1775,
      "step": 5209
    },
    {
      "epoch": 0.21066038594114853,
      "grad_norm": 5.8287034034729,
      "learning_rate": 8.9750825234591e-05,
      "loss": 0.1646,
      "step": 5210
    },
    {
      "epoch": 0.21070081979641553,
      "grad_norm": 5.557160377502441,
      "learning_rate": 8.974695651012335e-05,
      "loss": 0.2148,
      "step": 5211
    },
    {
      "epoch": 0.21074125365168256,
      "grad_norm": 4.819445610046387,
      "learning_rate": 8.974308713904997e-05,
      "loss": 0.1094,
      "step": 5212
    },
    {
      "epoch": 0.21078168750694956,
      "grad_norm": 3.3494904041290283,
      "learning_rate": 8.973921712143384e-05,
      "loss": 0.1305,
      "step": 5213
    },
    {
      "epoch": 0.2108221213622166,
      "grad_norm": 4.766417503356934,
      "learning_rate": 8.97353464573379e-05,
      "loss": 0.1365,
      "step": 5214
    },
    {
      "epoch": 0.2108625552174836,
      "grad_norm": 3.8469276428222656,
      "learning_rate": 8.973147514682514e-05,
      "loss": 0.1186,
      "step": 5215
    },
    {
      "epoch": 0.21090298907275062,
      "grad_norm": 9.107309341430664,
      "learning_rate": 8.97276031899585e-05,
      "loss": 0.2389,
      "step": 5216
    },
    {
      "epoch": 0.21094342292801763,
      "grad_norm": 7.933743000030518,
      "learning_rate": 8.972373058680098e-05,
      "loss": 0.2108,
      "step": 5217
    },
    {
      "epoch": 0.21098385678328466,
      "grad_norm": 8.10784912109375,
      "learning_rate": 8.971985733741562e-05,
      "loss": 0.3578,
      "step": 5218
    },
    {
      "epoch": 0.21102429063855166,
      "grad_norm": 4.4215288162231445,
      "learning_rate": 8.971598344186538e-05,
      "loss": 0.1219,
      "step": 5219
    },
    {
      "epoch": 0.21106472449381866,
      "grad_norm": 3.477217674255371,
      "learning_rate": 8.971210890021331e-05,
      "loss": 0.103,
      "step": 5220
    },
    {
      "epoch": 0.2111051583490857,
      "grad_norm": 4.460144519805908,
      "learning_rate": 8.970823371252244e-05,
      "loss": 0.1873,
      "step": 5221
    },
    {
      "epoch": 0.2111455922043527,
      "grad_norm": 3.7075352668762207,
      "learning_rate": 8.970435787885579e-05,
      "loss": 0.2276,
      "step": 5222
    },
    {
      "epoch": 0.21118602605961972,
      "grad_norm": 4.991286754608154,
      "learning_rate": 8.970048139927642e-05,
      "loss": 0.1679,
      "step": 5223
    },
    {
      "epoch": 0.21122645991488673,
      "grad_norm": 4.710934638977051,
      "learning_rate": 8.969660427384741e-05,
      "loss": 0.1443,
      "step": 5224
    },
    {
      "epoch": 0.21126689377015376,
      "grad_norm": 3.398624897003174,
      "learning_rate": 8.969272650263182e-05,
      "loss": 0.1872,
      "step": 5225
    },
    {
      "epoch": 0.21130732762542076,
      "grad_norm": 3.380337715148926,
      "learning_rate": 8.968884808569274e-05,
      "loss": 0.1068,
      "step": 5226
    },
    {
      "epoch": 0.2113477614806878,
      "grad_norm": 5.449750900268555,
      "learning_rate": 8.968496902309326e-05,
      "loss": 0.1435,
      "step": 5227
    },
    {
      "epoch": 0.2113881953359548,
      "grad_norm": 6.225983619689941,
      "learning_rate": 8.968108931489647e-05,
      "loss": 0.2717,
      "step": 5228
    },
    {
      "epoch": 0.21142862919122182,
      "grad_norm": 2.028402090072632,
      "learning_rate": 8.967720896116551e-05,
      "loss": 0.1846,
      "step": 5229
    },
    {
      "epoch": 0.21146906304648883,
      "grad_norm": 2.840054988861084,
      "learning_rate": 8.96733279619635e-05,
      "loss": 0.2629,
      "step": 5230
    },
    {
      "epoch": 0.21150949690175583,
      "grad_norm": 6.428151607513428,
      "learning_rate": 8.966944631735358e-05,
      "loss": 0.2948,
      "step": 5231
    },
    {
      "epoch": 0.21154993075702286,
      "grad_norm": 4.264814376831055,
      "learning_rate": 8.966556402739887e-05,
      "loss": 0.1937,
      "step": 5232
    },
    {
      "epoch": 0.21159036461228986,
      "grad_norm": 2.605544090270996,
      "learning_rate": 8.966168109216256e-05,
      "loss": 0.1415,
      "step": 5233
    },
    {
      "epoch": 0.2116307984675569,
      "grad_norm": 6.737301349639893,
      "learning_rate": 8.96577975117078e-05,
      "loss": 0.2122,
      "step": 5234
    },
    {
      "epoch": 0.2116712323228239,
      "grad_norm": 5.901518821716309,
      "learning_rate": 8.965391328609778e-05,
      "loss": 0.2991,
      "step": 5235
    },
    {
      "epoch": 0.21171166617809092,
      "grad_norm": 4.37439489364624,
      "learning_rate": 8.965002841539567e-05,
      "loss": 0.1306,
      "step": 5236
    },
    {
      "epoch": 0.21175210003335793,
      "grad_norm": 4.95607852935791,
      "learning_rate": 8.964614289966469e-05,
      "loss": 0.2159,
      "step": 5237
    },
    {
      "epoch": 0.21179253388862496,
      "grad_norm": 4.989381313323975,
      "learning_rate": 8.964225673896805e-05,
      "loss": 0.1503,
      "step": 5238
    },
    {
      "epoch": 0.21183296774389196,
      "grad_norm": 3.362720251083374,
      "learning_rate": 8.963836993336894e-05,
      "loss": 0.0962,
      "step": 5239
    },
    {
      "epoch": 0.211873401599159,
      "grad_norm": 5.2487311363220215,
      "learning_rate": 8.963448248293062e-05,
      "loss": 0.1395,
      "step": 5240
    },
    {
      "epoch": 0.211913835454426,
      "grad_norm": 4.447917938232422,
      "learning_rate": 8.963059438771631e-05,
      "loss": 0.1593,
      "step": 5241
    },
    {
      "epoch": 0.211954269309693,
      "grad_norm": 2.784092426300049,
      "learning_rate": 8.96267056477893e-05,
      "loss": 0.0982,
      "step": 5242
    },
    {
      "epoch": 0.21199470316496002,
      "grad_norm": 3.051164388656616,
      "learning_rate": 8.96228162632128e-05,
      "loss": 0.3067,
      "step": 5243
    },
    {
      "epoch": 0.21203513702022703,
      "grad_norm": 5.630580902099609,
      "learning_rate": 8.961892623405013e-05,
      "loss": 0.123,
      "step": 5244
    },
    {
      "epoch": 0.21207557087549406,
      "grad_norm": 3.8375375270843506,
      "learning_rate": 8.961503556036455e-05,
      "loss": 0.2362,
      "step": 5245
    },
    {
      "epoch": 0.21211600473076106,
      "grad_norm": 3.2671802043914795,
      "learning_rate": 8.961114424221934e-05,
      "loss": 0.1274,
      "step": 5246
    },
    {
      "epoch": 0.2121564385860281,
      "grad_norm": 3.9283242225646973,
      "learning_rate": 8.960725227967785e-05,
      "loss": 0.1511,
      "step": 5247
    },
    {
      "epoch": 0.2121968724412951,
      "grad_norm": 2.3747446537017822,
      "learning_rate": 8.960335967280334e-05,
      "loss": 0.1053,
      "step": 5248
    },
    {
      "epoch": 0.21223730629656212,
      "grad_norm": 4.26353120803833,
      "learning_rate": 8.959946642165917e-05,
      "loss": 0.0928,
      "step": 5249
    },
    {
      "epoch": 0.21227774015182913,
      "grad_norm": 5.287972450256348,
      "learning_rate": 8.959557252630868e-05,
      "loss": 0.174,
      "step": 5250
    },
    {
      "epoch": 0.21231817400709616,
      "grad_norm": 3.993265151977539,
      "learning_rate": 8.959167798681518e-05,
      "loss": 0.2339,
      "step": 5251
    },
    {
      "epoch": 0.21235860786236316,
      "grad_norm": 5.608341693878174,
      "learning_rate": 8.958778280324206e-05,
      "loss": 0.2186,
      "step": 5252
    },
    {
      "epoch": 0.21239904171763016,
      "grad_norm": 6.075004577636719,
      "learning_rate": 8.958388697565267e-05,
      "loss": 0.2438,
      "step": 5253
    },
    {
      "epoch": 0.2124394755728972,
      "grad_norm": 3.992189407348633,
      "learning_rate": 8.95799905041104e-05,
      "loss": 0.1429,
      "step": 5254
    },
    {
      "epoch": 0.2124799094281642,
      "grad_norm": 5.325757026672363,
      "learning_rate": 8.957609338867862e-05,
      "loss": 0.2601,
      "step": 5255
    },
    {
      "epoch": 0.21252034328343122,
      "grad_norm": 8.411439895629883,
      "learning_rate": 8.957219562942075e-05,
      "loss": 0.2211,
      "step": 5256
    },
    {
      "epoch": 0.21256077713869823,
      "grad_norm": 6.867001056671143,
      "learning_rate": 8.956829722640018e-05,
      "loss": 0.2761,
      "step": 5257
    },
    {
      "epoch": 0.21260121099396526,
      "grad_norm": 4.641340255737305,
      "learning_rate": 8.956439817968036e-05,
      "loss": 0.1403,
      "step": 5258
    },
    {
      "epoch": 0.21264164484923226,
      "grad_norm": 5.525411605834961,
      "learning_rate": 8.956049848932469e-05,
      "loss": 0.1844,
      "step": 5259
    },
    {
      "epoch": 0.2126820787044993,
      "grad_norm": 4.0200653076171875,
      "learning_rate": 8.955659815539661e-05,
      "loss": 0.2445,
      "step": 5260
    },
    {
      "epoch": 0.2127225125597663,
      "grad_norm": 4.237215042114258,
      "learning_rate": 8.955269717795958e-05,
      "loss": 0.2054,
      "step": 5261
    },
    {
      "epoch": 0.2127629464150333,
      "grad_norm": 6.039429187774658,
      "learning_rate": 8.954879555707706e-05,
      "loss": 0.194,
      "step": 5262
    },
    {
      "epoch": 0.21280338027030032,
      "grad_norm": 3.153888463973999,
      "learning_rate": 8.954489329281253e-05,
      "loss": 0.161,
      "step": 5263
    },
    {
      "epoch": 0.21284381412556733,
      "grad_norm": 6.430368423461914,
      "learning_rate": 8.954099038522945e-05,
      "loss": 0.3095,
      "step": 5264
    },
    {
      "epoch": 0.21288424798083436,
      "grad_norm": 6.979921817779541,
      "learning_rate": 8.953708683439134e-05,
      "loss": 0.1484,
      "step": 5265
    },
    {
      "epoch": 0.21292468183610136,
      "grad_norm": 5.314713478088379,
      "learning_rate": 8.95331826403617e-05,
      "loss": 0.1605,
      "step": 5266
    },
    {
      "epoch": 0.2129651156913684,
      "grad_norm": 6.317442893981934,
      "learning_rate": 8.952927780320401e-05,
      "loss": 0.2129,
      "step": 5267
    },
    {
      "epoch": 0.2130055495466354,
      "grad_norm": 8.074735641479492,
      "learning_rate": 8.952537232298183e-05,
      "loss": 0.2552,
      "step": 5268
    },
    {
      "epoch": 0.21304598340190242,
      "grad_norm": 3.2725777626037598,
      "learning_rate": 8.95214661997587e-05,
      "loss": 0.1135,
      "step": 5269
    },
    {
      "epoch": 0.21308641725716942,
      "grad_norm": 5.315301895141602,
      "learning_rate": 8.951755943359812e-05,
      "loss": 0.1669,
      "step": 5270
    },
    {
      "epoch": 0.21312685111243646,
      "grad_norm": 3.867809534072876,
      "learning_rate": 8.95136520245637e-05,
      "loss": 0.1571,
      "step": 5271
    },
    {
      "epoch": 0.21316728496770346,
      "grad_norm": 7.316400051116943,
      "learning_rate": 8.950974397271896e-05,
      "loss": 0.2379,
      "step": 5272
    },
    {
      "epoch": 0.21320771882297046,
      "grad_norm": 6.606378078460693,
      "learning_rate": 8.95058352781275e-05,
      "loss": 0.2106,
      "step": 5273
    },
    {
      "epoch": 0.2132481526782375,
      "grad_norm": 7.51435661315918,
      "learning_rate": 8.950192594085292e-05,
      "loss": 0.2826,
      "step": 5274
    },
    {
      "epoch": 0.2132885865335045,
      "grad_norm": 6.360867977142334,
      "learning_rate": 8.949801596095877e-05,
      "loss": 0.1989,
      "step": 5275
    },
    {
      "epoch": 0.21332902038877152,
      "grad_norm": 2.8132429122924805,
      "learning_rate": 8.94941053385087e-05,
      "loss": 0.1937,
      "step": 5276
    },
    {
      "epoch": 0.21336945424403853,
      "grad_norm": 4.482876777648926,
      "learning_rate": 8.949019407356634e-05,
      "loss": 0.1742,
      "step": 5277
    },
    {
      "epoch": 0.21340988809930556,
      "grad_norm": 11.14736270904541,
      "learning_rate": 8.948628216619527e-05,
      "loss": 0.3037,
      "step": 5278
    },
    {
      "epoch": 0.21345032195457256,
      "grad_norm": 5.27900505065918,
      "learning_rate": 8.948236961645915e-05,
      "loss": 0.2249,
      "step": 5279
    },
    {
      "epoch": 0.2134907558098396,
      "grad_norm": 2.3742032051086426,
      "learning_rate": 8.947845642442165e-05,
      "loss": 0.1296,
      "step": 5280
    },
    {
      "epoch": 0.2135311896651066,
      "grad_norm": 3.5166144371032715,
      "learning_rate": 8.94745425901464e-05,
      "loss": 0.2313,
      "step": 5281
    },
    {
      "epoch": 0.21357162352037362,
      "grad_norm": 1.5057975053787231,
      "learning_rate": 8.947062811369709e-05,
      "loss": 0.0815,
      "step": 5282
    },
    {
      "epoch": 0.21361205737564062,
      "grad_norm": 3.8054449558258057,
      "learning_rate": 8.94667129951374e-05,
      "loss": 0.0962,
      "step": 5283
    },
    {
      "epoch": 0.21365249123090763,
      "grad_norm": 6.141009330749512,
      "learning_rate": 8.946279723453101e-05,
      "loss": 0.1164,
      "step": 5284
    },
    {
      "epoch": 0.21369292508617466,
      "grad_norm": 7.349905014038086,
      "learning_rate": 8.945888083194163e-05,
      "loss": 0.2674,
      "step": 5285
    },
    {
      "epoch": 0.21373335894144166,
      "grad_norm": 7.2498884201049805,
      "learning_rate": 8.945496378743296e-05,
      "loss": 0.2114,
      "step": 5286
    },
    {
      "epoch": 0.2137737927967087,
      "grad_norm": 5.15765905380249,
      "learning_rate": 8.945104610106876e-05,
      "loss": 0.1345,
      "step": 5287
    },
    {
      "epoch": 0.2138142266519757,
      "grad_norm": 2.0153439044952393,
      "learning_rate": 8.94471277729127e-05,
      "loss": 0.1421,
      "step": 5288
    },
    {
      "epoch": 0.21385466050724272,
      "grad_norm": 1.9693635702133179,
      "learning_rate": 8.944320880302858e-05,
      "loss": 0.1099,
      "step": 5289
    },
    {
      "epoch": 0.21389509436250972,
      "grad_norm": 9.653358459472656,
      "learning_rate": 8.943928919148015e-05,
      "loss": 0.2293,
      "step": 5290
    },
    {
      "epoch": 0.21393552821777675,
      "grad_norm": 4.676810264587402,
      "learning_rate": 8.943536893833114e-05,
      "loss": 0.2132,
      "step": 5291
    },
    {
      "epoch": 0.21397596207304376,
      "grad_norm": 5.3209309577941895,
      "learning_rate": 8.943144804364533e-05,
      "loss": 0.1448,
      "step": 5292
    },
    {
      "epoch": 0.2140163959283108,
      "grad_norm": 3.974835157394409,
      "learning_rate": 8.942752650748653e-05,
      "loss": 0.1758,
      "step": 5293
    },
    {
      "epoch": 0.2140568297835778,
      "grad_norm": 3.626441717147827,
      "learning_rate": 8.942360432991855e-05,
      "loss": 0.1024,
      "step": 5294
    },
    {
      "epoch": 0.2140972636388448,
      "grad_norm": 3.6673710346221924,
      "learning_rate": 8.941968151100513e-05,
      "loss": 0.1633,
      "step": 5295
    },
    {
      "epoch": 0.21413769749411182,
      "grad_norm": 6.518669128417969,
      "learning_rate": 8.941575805081015e-05,
      "loss": 0.2016,
      "step": 5296
    },
    {
      "epoch": 0.21417813134937883,
      "grad_norm": 4.320281982421875,
      "learning_rate": 8.94118339493974e-05,
      "loss": 0.1695,
      "step": 5297
    },
    {
      "epoch": 0.21421856520464586,
      "grad_norm": 4.134695053100586,
      "learning_rate": 8.940790920683075e-05,
      "loss": 0.1157,
      "step": 5298
    },
    {
      "epoch": 0.21425899905991286,
      "grad_norm": 6.717288017272949,
      "learning_rate": 8.940398382317403e-05,
      "loss": 0.1665,
      "step": 5299
    },
    {
      "epoch": 0.2142994329151799,
      "grad_norm": 3.9557902812957764,
      "learning_rate": 8.940005779849108e-05,
      "loss": 0.1505,
      "step": 5300
    },
    {
      "epoch": 0.2143398667704469,
      "grad_norm": 1.7402886152267456,
      "learning_rate": 8.939613113284579e-05,
      "loss": 0.1015,
      "step": 5301
    },
    {
      "epoch": 0.21438030062571392,
      "grad_norm": 2.8747642040252686,
      "learning_rate": 8.939220382630204e-05,
      "loss": 0.2081,
      "step": 5302
    },
    {
      "epoch": 0.21442073448098092,
      "grad_norm": 8.565159797668457,
      "learning_rate": 8.938827587892373e-05,
      "loss": 0.2183,
      "step": 5303
    },
    {
      "epoch": 0.21446116833624795,
      "grad_norm": 5.532517910003662,
      "learning_rate": 8.938434729077473e-05,
      "loss": 0.2167,
      "step": 5304
    },
    {
      "epoch": 0.21450160219151496,
      "grad_norm": 3.1216375827789307,
      "learning_rate": 8.938041806191897e-05,
      "loss": 0.1055,
      "step": 5305
    },
    {
      "epoch": 0.21454203604678196,
      "grad_norm": 5.682171821594238,
      "learning_rate": 8.937648819242037e-05,
      "loss": 0.2556,
      "step": 5306
    },
    {
      "epoch": 0.214582469902049,
      "grad_norm": 3.2987968921661377,
      "learning_rate": 8.937255768234285e-05,
      "loss": 0.1634,
      "step": 5307
    },
    {
      "epoch": 0.214622903757316,
      "grad_norm": 4.429350852966309,
      "learning_rate": 8.936862653175038e-05,
      "loss": 0.1486,
      "step": 5308
    },
    {
      "epoch": 0.21466333761258302,
      "grad_norm": 5.226406097412109,
      "learning_rate": 8.936469474070688e-05,
      "loss": 0.3069,
      "step": 5309
    },
    {
      "epoch": 0.21470377146785002,
      "grad_norm": 4.73019552230835,
      "learning_rate": 8.936076230927632e-05,
      "loss": 0.2028,
      "step": 5310
    },
    {
      "epoch": 0.21474420532311705,
      "grad_norm": 4.59009313583374,
      "learning_rate": 8.935682923752267e-05,
      "loss": 0.1555,
      "step": 5311
    },
    {
      "epoch": 0.21478463917838406,
      "grad_norm": 4.458110332489014,
      "learning_rate": 8.935289552550994e-05,
      "loss": 0.2422,
      "step": 5312
    },
    {
      "epoch": 0.2148250730336511,
      "grad_norm": 6.435807228088379,
      "learning_rate": 8.93489611733021e-05,
      "loss": 0.2341,
      "step": 5313
    },
    {
      "epoch": 0.2148655068889181,
      "grad_norm": 3.6873395442962646,
      "learning_rate": 8.934502618096315e-05,
      "loss": 0.1603,
      "step": 5314
    },
    {
      "epoch": 0.21490594074418512,
      "grad_norm": 4.494266033172607,
      "learning_rate": 8.934109054855712e-05,
      "loss": 0.1147,
      "step": 5315
    },
    {
      "epoch": 0.21494637459945212,
      "grad_norm": 5.987549304962158,
      "learning_rate": 8.933715427614804e-05,
      "loss": 0.1938,
      "step": 5316
    },
    {
      "epoch": 0.21498680845471912,
      "grad_norm": 4.163389682769775,
      "learning_rate": 8.933321736379993e-05,
      "loss": 0.2582,
      "step": 5317
    },
    {
      "epoch": 0.21502724230998616,
      "grad_norm": 4.100283622741699,
      "learning_rate": 8.932927981157683e-05,
      "loss": 0.1926,
      "step": 5318
    },
    {
      "epoch": 0.21506767616525316,
      "grad_norm": 3.2852871417999268,
      "learning_rate": 8.932534161954281e-05,
      "loss": 0.0949,
      "step": 5319
    },
    {
      "epoch": 0.2151081100205202,
      "grad_norm": 3.5416741371154785,
      "learning_rate": 8.932140278776193e-05,
      "loss": 0.198,
      "step": 5320
    },
    {
      "epoch": 0.2151485438757872,
      "grad_norm": 3.1169376373291016,
      "learning_rate": 8.931746331629829e-05,
      "loss": 0.1829,
      "step": 5321
    },
    {
      "epoch": 0.21518897773105422,
      "grad_norm": 5.630311965942383,
      "learning_rate": 8.931352320521596e-05,
      "loss": 0.1415,
      "step": 5322
    },
    {
      "epoch": 0.21522941158632122,
      "grad_norm": 6.727334499359131,
      "learning_rate": 8.930958245457902e-05,
      "loss": 0.2628,
      "step": 5323
    },
    {
      "epoch": 0.21526984544158825,
      "grad_norm": 2.736785411834717,
      "learning_rate": 8.930564106445159e-05,
      "loss": 0.1554,
      "step": 5324
    },
    {
      "epoch": 0.21531027929685526,
      "grad_norm": 3.9724087715148926,
      "learning_rate": 8.93016990348978e-05,
      "loss": 0.1083,
      "step": 5325
    },
    {
      "epoch": 0.21535071315212226,
      "grad_norm": 3.782036304473877,
      "learning_rate": 8.929775636598176e-05,
      "loss": 0.1194,
      "step": 5326
    },
    {
      "epoch": 0.2153911470073893,
      "grad_norm": 6.085121154785156,
      "learning_rate": 8.929381305776765e-05,
      "loss": 0.1576,
      "step": 5327
    },
    {
      "epoch": 0.2154315808626563,
      "grad_norm": 9.962357521057129,
      "learning_rate": 8.928986911031958e-05,
      "loss": 0.261,
      "step": 5328
    },
    {
      "epoch": 0.21547201471792332,
      "grad_norm": 6.082397937774658,
      "learning_rate": 8.928592452370172e-05,
      "loss": 0.1674,
      "step": 5329
    },
    {
      "epoch": 0.21551244857319032,
      "grad_norm": 4.147129058837891,
      "learning_rate": 8.928197929797824e-05,
      "loss": 0.1733,
      "step": 5330
    },
    {
      "epoch": 0.21555288242845735,
      "grad_norm": 4.682778358459473,
      "learning_rate": 8.927803343321333e-05,
      "loss": 0.2043,
      "step": 5331
    },
    {
      "epoch": 0.21559331628372436,
      "grad_norm": 3.9567208290100098,
      "learning_rate": 8.927408692947118e-05,
      "loss": 0.1459,
      "step": 5332
    },
    {
      "epoch": 0.2156337501389914,
      "grad_norm": 2.5460264682769775,
      "learning_rate": 8.927013978681597e-05,
      "loss": 0.2285,
      "step": 5333
    },
    {
      "epoch": 0.2156741839942584,
      "grad_norm": 4.895883560180664,
      "learning_rate": 8.926619200531197e-05,
      "loss": 0.1631,
      "step": 5334
    },
    {
      "epoch": 0.21571461784952542,
      "grad_norm": 3.620026111602783,
      "learning_rate": 8.926224358502333e-05,
      "loss": 0.1251,
      "step": 5335
    },
    {
      "epoch": 0.21575505170479242,
      "grad_norm": 6.0602617263793945,
      "learning_rate": 8.925829452601431e-05,
      "loss": 0.2021,
      "step": 5336
    },
    {
      "epoch": 0.21579548556005942,
      "grad_norm": 5.825416088104248,
      "learning_rate": 8.925434482834918e-05,
      "loss": 0.2048,
      "step": 5337
    },
    {
      "epoch": 0.21583591941532645,
      "grad_norm": 6.444488525390625,
      "learning_rate": 8.925039449209216e-05,
      "loss": 0.2163,
      "step": 5338
    },
    {
      "epoch": 0.21587635327059346,
      "grad_norm": 4.4959259033203125,
      "learning_rate": 8.924644351730753e-05,
      "loss": 0.2323,
      "step": 5339
    },
    {
      "epoch": 0.2159167871258605,
      "grad_norm": 7.836727142333984,
      "learning_rate": 8.924249190405958e-05,
      "loss": 0.1697,
      "step": 5340
    },
    {
      "epoch": 0.2159572209811275,
      "grad_norm": 8.155659675598145,
      "learning_rate": 8.923853965241256e-05,
      "loss": 0.2076,
      "step": 5341
    },
    {
      "epoch": 0.21599765483639452,
      "grad_norm": 4.484534740447998,
      "learning_rate": 8.923458676243079e-05,
      "loss": 0.1669,
      "step": 5342
    },
    {
      "epoch": 0.21603808869166152,
      "grad_norm": 2.623894691467285,
      "learning_rate": 8.923063323417856e-05,
      "loss": 0.1812,
      "step": 5343
    },
    {
      "epoch": 0.21607852254692855,
      "grad_norm": 1.8793420791625977,
      "learning_rate": 8.92266790677202e-05,
      "loss": 0.1193,
      "step": 5344
    },
    {
      "epoch": 0.21611895640219556,
      "grad_norm": 4.65909481048584,
      "learning_rate": 8.922272426312004e-05,
      "loss": 0.1446,
      "step": 5345
    },
    {
      "epoch": 0.21615939025746259,
      "grad_norm": 4.16987943649292,
      "learning_rate": 8.92187688204424e-05,
      "loss": 0.1233,
      "step": 5346
    },
    {
      "epoch": 0.2161998241127296,
      "grad_norm": 1.9518911838531494,
      "learning_rate": 8.921481273975163e-05,
      "loss": 0.0811,
      "step": 5347
    },
    {
      "epoch": 0.2162402579679966,
      "grad_norm": 6.627835750579834,
      "learning_rate": 8.921085602111212e-05,
      "loss": 0.313,
      "step": 5348
    },
    {
      "epoch": 0.21628069182326362,
      "grad_norm": 5.323002338409424,
      "learning_rate": 8.920689866458818e-05,
      "loss": 0.1959,
      "step": 5349
    },
    {
      "epoch": 0.21632112567853062,
      "grad_norm": 4.926192283630371,
      "learning_rate": 8.920294067024424e-05,
      "loss": 0.244,
      "step": 5350
    },
    {
      "epoch": 0.21636155953379765,
      "grad_norm": 8.576013565063477,
      "learning_rate": 8.919898203814465e-05,
      "loss": 0.2944,
      "step": 5351
    },
    {
      "epoch": 0.21640199338906466,
      "grad_norm": 5.3044562339782715,
      "learning_rate": 8.919502276835386e-05,
      "loss": 0.2038,
      "step": 5352
    },
    {
      "epoch": 0.2164424272443317,
      "grad_norm": 4.1832594871521,
      "learning_rate": 8.919106286093622e-05,
      "loss": 0.1921,
      "step": 5353
    },
    {
      "epoch": 0.2164828610995987,
      "grad_norm": 4.153326511383057,
      "learning_rate": 8.918710231595619e-05,
      "loss": 0.1551,
      "step": 5354
    },
    {
      "epoch": 0.21652329495486572,
      "grad_norm": 2.594409465789795,
      "learning_rate": 8.918314113347818e-05,
      "loss": 0.1631,
      "step": 5355
    },
    {
      "epoch": 0.21656372881013272,
      "grad_norm": 7.255805015563965,
      "learning_rate": 8.917917931356665e-05,
      "loss": 0.1589,
      "step": 5356
    },
    {
      "epoch": 0.21660416266539975,
      "grad_norm": 3.24521541595459,
      "learning_rate": 8.917521685628603e-05,
      "loss": 0.2232,
      "step": 5357
    },
    {
      "epoch": 0.21664459652066675,
      "grad_norm": 2.537763833999634,
      "learning_rate": 8.91712537617008e-05,
      "loss": 0.2191,
      "step": 5358
    },
    {
      "epoch": 0.21668503037593376,
      "grad_norm": 5.72615385055542,
      "learning_rate": 8.916729002987542e-05,
      "loss": 0.2292,
      "step": 5359
    },
    {
      "epoch": 0.2167254642312008,
      "grad_norm": 4.312389373779297,
      "learning_rate": 8.916332566087438e-05,
      "loss": 0.1624,
      "step": 5360
    },
    {
      "epoch": 0.2167658980864678,
      "grad_norm": 7.075811862945557,
      "learning_rate": 8.915936065476216e-05,
      "loss": 0.2938,
      "step": 5361
    },
    {
      "epoch": 0.21680633194173482,
      "grad_norm": 6.624897003173828,
      "learning_rate": 8.915539501160328e-05,
      "loss": 0.227,
      "step": 5362
    },
    {
      "epoch": 0.21684676579700182,
      "grad_norm": 3.5128509998321533,
      "learning_rate": 8.915142873146225e-05,
      "loss": 0.1089,
      "step": 5363
    },
    {
      "epoch": 0.21688719965226885,
      "grad_norm": 1.5512341260910034,
      "learning_rate": 8.914746181440358e-05,
      "loss": 0.0446,
      "step": 5364
    },
    {
      "epoch": 0.21692763350753586,
      "grad_norm": 3.4936728477478027,
      "learning_rate": 8.914349426049181e-05,
      "loss": 0.1306,
      "step": 5365
    },
    {
      "epoch": 0.21696806736280289,
      "grad_norm": 5.325344562530518,
      "learning_rate": 8.913952606979149e-05,
      "loss": 0.2218,
      "step": 5366
    },
    {
      "epoch": 0.2170085012180699,
      "grad_norm": 6.527713298797607,
      "learning_rate": 8.913555724236718e-05,
      "loss": 0.2622,
      "step": 5367
    },
    {
      "epoch": 0.21704893507333692,
      "grad_norm": 5.0788679122924805,
      "learning_rate": 8.913158777828343e-05,
      "loss": 0.1532,
      "step": 5368
    },
    {
      "epoch": 0.21708936892860392,
      "grad_norm": 3.4951281547546387,
      "learning_rate": 8.91276176776048e-05,
      "loss": 0.2707,
      "step": 5369
    },
    {
      "epoch": 0.21712980278387092,
      "grad_norm": 4.559939384460449,
      "learning_rate": 8.912364694039593e-05,
      "loss": 0.2316,
      "step": 5370
    },
    {
      "epoch": 0.21717023663913795,
      "grad_norm": 2.272103786468506,
      "learning_rate": 8.911967556672138e-05,
      "loss": 0.0847,
      "step": 5371
    },
    {
      "epoch": 0.21721067049440496,
      "grad_norm": 5.19473123550415,
      "learning_rate": 8.911570355664574e-05,
      "loss": 0.216,
      "step": 5372
    },
    {
      "epoch": 0.217251104349672,
      "grad_norm": 4.884327411651611,
      "learning_rate": 8.911173091023366e-05,
      "loss": 0.1825,
      "step": 5373
    },
    {
      "epoch": 0.217291538204939,
      "grad_norm": 8.4952392578125,
      "learning_rate": 8.910775762754975e-05,
      "loss": 0.215,
      "step": 5374
    },
    {
      "epoch": 0.21733197206020602,
      "grad_norm": 6.86440372467041,
      "learning_rate": 8.910378370865866e-05,
      "loss": 0.3549,
      "step": 5375
    },
    {
      "epoch": 0.21737240591547302,
      "grad_norm": 4.392678737640381,
      "learning_rate": 8.909980915362504e-05,
      "loss": 0.1396,
      "step": 5376
    },
    {
      "epoch": 0.21741283977074005,
      "grad_norm": 5.62283182144165,
      "learning_rate": 8.909583396251352e-05,
      "loss": 0.2367,
      "step": 5377
    },
    {
      "epoch": 0.21745327362600705,
      "grad_norm": 7.4872212409973145,
      "learning_rate": 8.909185813538879e-05,
      "loss": 0.1909,
      "step": 5378
    },
    {
      "epoch": 0.21749370748127408,
      "grad_norm": 3.8919906616210938,
      "learning_rate": 8.908788167231553e-05,
      "loss": 0.2082,
      "step": 5379
    },
    {
      "epoch": 0.2175341413365411,
      "grad_norm": 6.177134990692139,
      "learning_rate": 8.908390457335843e-05,
      "loss": 0.1884,
      "step": 5380
    },
    {
      "epoch": 0.2175745751918081,
      "grad_norm": 2.8337392807006836,
      "learning_rate": 8.907992683858218e-05,
      "loss": 0.1284,
      "step": 5381
    },
    {
      "epoch": 0.21761500904707512,
      "grad_norm": 2.9820308685302734,
      "learning_rate": 8.907594846805149e-05,
      "loss": 0.1518,
      "step": 5382
    },
    {
      "epoch": 0.21765544290234212,
      "grad_norm": 3.522162437438965,
      "learning_rate": 8.90719694618311e-05,
      "loss": 0.2075,
      "step": 5383
    },
    {
      "epoch": 0.21769587675760915,
      "grad_norm": 2.7641453742980957,
      "learning_rate": 8.906798981998572e-05,
      "loss": 0.095,
      "step": 5384
    },
    {
      "epoch": 0.21773631061287615,
      "grad_norm": 6.302653789520264,
      "learning_rate": 8.906400954258009e-05,
      "loss": 0.1347,
      "step": 5385
    },
    {
      "epoch": 0.21777674446814319,
      "grad_norm": 5.479989528656006,
      "learning_rate": 8.906002862967897e-05,
      "loss": 0.0908,
      "step": 5386
    },
    {
      "epoch": 0.2178171783234102,
      "grad_norm": 1.9142744541168213,
      "learning_rate": 8.905604708134712e-05,
      "loss": 0.1341,
      "step": 5387
    },
    {
      "epoch": 0.21785761217867722,
      "grad_norm": 2.8117563724517822,
      "learning_rate": 8.905206489764932e-05,
      "loss": 0.0993,
      "step": 5388
    },
    {
      "epoch": 0.21789804603394422,
      "grad_norm": 7.664748191833496,
      "learning_rate": 8.904808207865035e-05,
      "loss": 0.2546,
      "step": 5389
    },
    {
      "epoch": 0.21793847988921122,
      "grad_norm": 2.869959592819214,
      "learning_rate": 8.904409862441499e-05,
      "loss": 0.1274,
      "step": 5390
    },
    {
      "epoch": 0.21797891374447825,
      "grad_norm": 6.664134502410889,
      "learning_rate": 8.904011453500804e-05,
      "loss": 0.2636,
      "step": 5391
    },
    {
      "epoch": 0.21801934759974526,
      "grad_norm": 4.146908760070801,
      "learning_rate": 8.903612981049434e-05,
      "loss": 0.2179,
      "step": 5392
    },
    {
      "epoch": 0.21805978145501229,
      "grad_norm": 7.9851884841918945,
      "learning_rate": 8.903214445093868e-05,
      "loss": 0.2275,
      "step": 5393
    },
    {
      "epoch": 0.2181002153102793,
      "grad_norm": 3.5592215061187744,
      "learning_rate": 8.902815845640593e-05,
      "loss": 0.0808,
      "step": 5394
    },
    {
      "epoch": 0.21814064916554632,
      "grad_norm": 6.113335132598877,
      "learning_rate": 8.90241718269609e-05,
      "loss": 0.2763,
      "step": 5395
    },
    {
      "epoch": 0.21818108302081332,
      "grad_norm": 6.257121562957764,
      "learning_rate": 8.902018456266848e-05,
      "loss": 0.1614,
      "step": 5396
    },
    {
      "epoch": 0.21822151687608035,
      "grad_norm": 3.945131540298462,
      "learning_rate": 8.90161966635935e-05,
      "loss": 0.0781,
      "step": 5397
    },
    {
      "epoch": 0.21826195073134735,
      "grad_norm": 2.5701966285705566,
      "learning_rate": 8.901220812980085e-05,
      "loss": 0.0814,
      "step": 5398
    },
    {
      "epoch": 0.21830238458661438,
      "grad_norm": 5.962611198425293,
      "learning_rate": 8.900821896135542e-05,
      "loss": 0.2118,
      "step": 5399
    },
    {
      "epoch": 0.2183428184418814,
      "grad_norm": 6.770645618438721,
      "learning_rate": 8.900422915832212e-05,
      "loss": 0.3779,
      "step": 5400
    },
    {
      "epoch": 0.2183832522971484,
      "grad_norm": 3.507009506225586,
      "learning_rate": 8.900023872076582e-05,
      "loss": 0.1268,
      "step": 5401
    },
    {
      "epoch": 0.21842368615241542,
      "grad_norm": 3.5702202320098877,
      "learning_rate": 8.899624764875145e-05,
      "loss": 0.2023,
      "step": 5402
    },
    {
      "epoch": 0.21846412000768242,
      "grad_norm": 2.3021080493927,
      "learning_rate": 8.899225594234397e-05,
      "loss": 0.083,
      "step": 5403
    },
    {
      "epoch": 0.21850455386294945,
      "grad_norm": 9.345029830932617,
      "learning_rate": 8.898826360160827e-05,
      "loss": 0.2568,
      "step": 5404
    },
    {
      "epoch": 0.21854498771821645,
      "grad_norm": 3.9948971271514893,
      "learning_rate": 8.898427062660933e-05,
      "loss": 0.2401,
      "step": 5405
    },
    {
      "epoch": 0.21858542157348348,
      "grad_norm": 4.810365676879883,
      "learning_rate": 8.898027701741207e-05,
      "loss": 0.2786,
      "step": 5406
    },
    {
      "epoch": 0.2186258554287505,
      "grad_norm": 2.5818088054656982,
      "learning_rate": 8.89762827740815e-05,
      "loss": 0.1357,
      "step": 5407
    },
    {
      "epoch": 0.21866628928401752,
      "grad_norm": 4.816763401031494,
      "learning_rate": 8.89722878966826e-05,
      "loss": 0.1982,
      "step": 5408
    },
    {
      "epoch": 0.21870672313928452,
      "grad_norm": 10.73133659362793,
      "learning_rate": 8.896829238528033e-05,
      "loss": 0.3465,
      "step": 5409
    },
    {
      "epoch": 0.21874715699455155,
      "grad_norm": 5.2296648025512695,
      "learning_rate": 8.89642962399397e-05,
      "loss": 0.2255,
      "step": 5410
    },
    {
      "epoch": 0.21878759084981855,
      "grad_norm": 3.645705461502075,
      "learning_rate": 8.896029946072571e-05,
      "loss": 0.2112,
      "step": 5411
    },
    {
      "epoch": 0.21882802470508556,
      "grad_norm": 4.58478307723999,
      "learning_rate": 8.89563020477034e-05,
      "loss": 0.1256,
      "step": 5412
    },
    {
      "epoch": 0.21886845856035259,
      "grad_norm": 6.75217866897583,
      "learning_rate": 8.895230400093778e-05,
      "loss": 0.1666,
      "step": 5413
    },
    {
      "epoch": 0.2189088924156196,
      "grad_norm": 4.598151206970215,
      "learning_rate": 8.894830532049392e-05,
      "loss": 0.1898,
      "step": 5414
    },
    {
      "epoch": 0.21894932627088662,
      "grad_norm": 4.371875286102295,
      "learning_rate": 8.894430600643685e-05,
      "loss": 0.1595,
      "step": 5415
    },
    {
      "epoch": 0.21898976012615362,
      "grad_norm": 3.2133636474609375,
      "learning_rate": 8.89403060588316e-05,
      "loss": 0.1834,
      "step": 5416
    },
    {
      "epoch": 0.21903019398142065,
      "grad_norm": 4.028210639953613,
      "learning_rate": 8.89363054777433e-05,
      "loss": 0.1871,
      "step": 5417
    },
    {
      "epoch": 0.21907062783668765,
      "grad_norm": 3.158087730407715,
      "learning_rate": 8.8932304263237e-05,
      "loss": 0.1127,
      "step": 5418
    },
    {
      "epoch": 0.21911106169195468,
      "grad_norm": 3.547004461288452,
      "learning_rate": 8.89283024153778e-05,
      "loss": 0.2285,
      "step": 5419
    },
    {
      "epoch": 0.2191514955472217,
      "grad_norm": 5.807936191558838,
      "learning_rate": 8.892429993423079e-05,
      "loss": 0.169,
      "step": 5420
    },
    {
      "epoch": 0.21919192940248872,
      "grad_norm": 5.850796699523926,
      "learning_rate": 8.89202968198611e-05,
      "loss": 0.1834,
      "step": 5421
    },
    {
      "epoch": 0.21923236325775572,
      "grad_norm": 4.228263854980469,
      "learning_rate": 8.891629307233386e-05,
      "loss": 0.15,
      "step": 5422
    },
    {
      "epoch": 0.21927279711302272,
      "grad_norm": 3.8631021976470947,
      "learning_rate": 8.891228869171418e-05,
      "loss": 0.1958,
      "step": 5423
    },
    {
      "epoch": 0.21931323096828975,
      "grad_norm": 3.0031585693359375,
      "learning_rate": 8.89082836780672e-05,
      "loss": 0.1492,
      "step": 5424
    },
    {
      "epoch": 0.21935366482355675,
      "grad_norm": 6.691739559173584,
      "learning_rate": 8.890427803145809e-05,
      "loss": 0.2356,
      "step": 5425
    },
    {
      "epoch": 0.21939409867882378,
      "grad_norm": 5.347770690917969,
      "learning_rate": 8.890027175195202e-05,
      "loss": 0.1595,
      "step": 5426
    },
    {
      "epoch": 0.2194345325340908,
      "grad_norm": 6.295507431030273,
      "learning_rate": 8.889626483961414e-05,
      "loss": 0.1692,
      "step": 5427
    },
    {
      "epoch": 0.21947496638935782,
      "grad_norm": 3.0147950649261475,
      "learning_rate": 8.889225729450965e-05,
      "loss": 0.1457,
      "step": 5428
    },
    {
      "epoch": 0.21951540024462482,
      "grad_norm": 3.0334832668304443,
      "learning_rate": 8.888824911670375e-05,
      "loss": 0.171,
      "step": 5429
    },
    {
      "epoch": 0.21955583409989185,
      "grad_norm": 6.353768825531006,
      "learning_rate": 8.888424030626164e-05,
      "loss": 0.1839,
      "step": 5430
    },
    {
      "epoch": 0.21959626795515885,
      "grad_norm": 3.200857162475586,
      "learning_rate": 8.888023086324852e-05,
      "loss": 0.0935,
      "step": 5431
    },
    {
      "epoch": 0.21963670181042588,
      "grad_norm": 5.292572498321533,
      "learning_rate": 8.887622078772966e-05,
      "loss": 0.1475,
      "step": 5432
    },
    {
      "epoch": 0.21967713566569289,
      "grad_norm": 3.6951591968536377,
      "learning_rate": 8.887221007977024e-05,
      "loss": 0.1498,
      "step": 5433
    },
    {
      "epoch": 0.2197175695209599,
      "grad_norm": 9.611342430114746,
      "learning_rate": 8.886819873943554e-05,
      "loss": 0.28,
      "step": 5434
    },
    {
      "epoch": 0.21975800337622692,
      "grad_norm": 4.252012729644775,
      "learning_rate": 8.886418676679083e-05,
      "loss": 0.1844,
      "step": 5435
    },
    {
      "epoch": 0.21979843723149392,
      "grad_norm": 5.0980634689331055,
      "learning_rate": 8.886017416190133e-05,
      "loss": 0.139,
      "step": 5436
    },
    {
      "epoch": 0.21983887108676095,
      "grad_norm": 3.603874444961548,
      "learning_rate": 8.885616092483236e-05,
      "loss": 0.1452,
      "step": 5437
    },
    {
      "epoch": 0.21987930494202795,
      "grad_norm": 3.671657085418701,
      "learning_rate": 8.885214705564919e-05,
      "loss": 0.1613,
      "step": 5438
    },
    {
      "epoch": 0.21991973879729498,
      "grad_norm": 4.875899791717529,
      "learning_rate": 8.884813255441711e-05,
      "loss": 0.1779,
      "step": 5439
    },
    {
      "epoch": 0.21996017265256199,
      "grad_norm": 5.025248050689697,
      "learning_rate": 8.884411742120143e-05,
      "loss": 0.1081,
      "step": 5440
    },
    {
      "epoch": 0.22000060650782902,
      "grad_norm": 8.046497344970703,
      "learning_rate": 8.884010165606748e-05,
      "loss": 0.1622,
      "step": 5441
    },
    {
      "epoch": 0.22004104036309602,
      "grad_norm": 3.0712976455688477,
      "learning_rate": 8.88360852590806e-05,
      "loss": 0.1732,
      "step": 5442
    },
    {
      "epoch": 0.22008147421836305,
      "grad_norm": 4.406735420227051,
      "learning_rate": 8.883206823030611e-05,
      "loss": 0.0834,
      "step": 5443
    },
    {
      "epoch": 0.22012190807363005,
      "grad_norm": 4.224616050720215,
      "learning_rate": 8.882805056980936e-05,
      "loss": 0.1436,
      "step": 5444
    },
    {
      "epoch": 0.22016234192889705,
      "grad_norm": 4.070415019989014,
      "learning_rate": 8.882403227765572e-05,
      "loss": 0.1587,
      "step": 5445
    },
    {
      "epoch": 0.22020277578416408,
      "grad_norm": 1.0926448106765747,
      "learning_rate": 8.882001335391054e-05,
      "loss": 0.0685,
      "step": 5446
    },
    {
      "epoch": 0.2202432096394311,
      "grad_norm": 3.4513583183288574,
      "learning_rate": 8.88159937986392e-05,
      "loss": 0.1034,
      "step": 5447
    },
    {
      "epoch": 0.22028364349469812,
      "grad_norm": 3.9889495372772217,
      "learning_rate": 8.881197361190713e-05,
      "loss": 0.1205,
      "step": 5448
    },
    {
      "epoch": 0.22032407734996512,
      "grad_norm": 3.175482749938965,
      "learning_rate": 8.88079527937797e-05,
      "loss": 0.116,
      "step": 5449
    },
    {
      "epoch": 0.22036451120523215,
      "grad_norm": 7.408778667449951,
      "learning_rate": 8.88039313443223e-05,
      "loss": 0.2086,
      "step": 5450
    },
    {
      "epoch": 0.22040494506049915,
      "grad_norm": 5.875043869018555,
      "learning_rate": 8.87999092636004e-05,
      "loss": 0.2007,
      "step": 5451
    },
    {
      "epoch": 0.22044537891576618,
      "grad_norm": 2.4618122577667236,
      "learning_rate": 8.879588655167939e-05,
      "loss": 0.1535,
      "step": 5452
    },
    {
      "epoch": 0.22048581277103318,
      "grad_norm": 4.515756607055664,
      "learning_rate": 8.879186320862473e-05,
      "loss": 0.1567,
      "step": 5453
    },
    {
      "epoch": 0.2205262466263002,
      "grad_norm": 5.129782199859619,
      "learning_rate": 8.87878392345019e-05,
      "loss": 0.3444,
      "step": 5454
    },
    {
      "epoch": 0.22056668048156722,
      "grad_norm": 4.513271331787109,
      "learning_rate": 8.87838146293763e-05,
      "loss": 0.1937,
      "step": 5455
    },
    {
      "epoch": 0.22060711433683422,
      "grad_norm": 1.9758552312850952,
      "learning_rate": 8.877978939331344e-05,
      "loss": 0.1089,
      "step": 5456
    },
    {
      "epoch": 0.22064754819210125,
      "grad_norm": 2.834087610244751,
      "learning_rate": 8.87757635263788e-05,
      "loss": 0.2156,
      "step": 5457
    },
    {
      "epoch": 0.22068798204736825,
      "grad_norm": 6.753626346588135,
      "learning_rate": 8.877173702863788e-05,
      "loss": 0.2938,
      "step": 5458
    },
    {
      "epoch": 0.22072841590263528,
      "grad_norm": 3.759791612625122,
      "learning_rate": 8.876770990015617e-05,
      "loss": 0.1165,
      "step": 5459
    },
    {
      "epoch": 0.22076884975790229,
      "grad_norm": 6.144514083862305,
      "learning_rate": 8.87636821409992e-05,
      "loss": 0.118,
      "step": 5460
    },
    {
      "epoch": 0.22080928361316932,
      "grad_norm": 6.986365795135498,
      "learning_rate": 8.875965375123245e-05,
      "loss": 0.211,
      "step": 5461
    },
    {
      "epoch": 0.22084971746843632,
      "grad_norm": 7.443093299865723,
      "learning_rate": 8.875562473092151e-05,
      "loss": 0.2576,
      "step": 5462
    },
    {
      "epoch": 0.22089015132370335,
      "grad_norm": 5.78293514251709,
      "learning_rate": 8.875159508013189e-05,
      "loss": 0.2737,
      "step": 5463
    },
    {
      "epoch": 0.22093058517897035,
      "grad_norm": 4.415812969207764,
      "learning_rate": 8.874756479892917e-05,
      "loss": 0.1095,
      "step": 5464
    },
    {
      "epoch": 0.22097101903423735,
      "grad_norm": 5.064813613891602,
      "learning_rate": 8.874353388737889e-05,
      "loss": 0.1293,
      "step": 5465
    },
    {
      "epoch": 0.22101145288950438,
      "grad_norm": 7.3318986892700195,
      "learning_rate": 8.873950234554662e-05,
      "loss": 0.2246,
      "step": 5466
    },
    {
      "epoch": 0.2210518867447714,
      "grad_norm": 4.837169170379639,
      "learning_rate": 8.873547017349796e-05,
      "loss": 0.1347,
      "step": 5467
    },
    {
      "epoch": 0.22109232060003842,
      "grad_norm": 2.2200138568878174,
      "learning_rate": 8.873143737129852e-05,
      "loss": 0.1094,
      "step": 5468
    },
    {
      "epoch": 0.22113275445530542,
      "grad_norm": 3.650245189666748,
      "learning_rate": 8.872740393901389e-05,
      "loss": 0.1157,
      "step": 5469
    },
    {
      "epoch": 0.22117318831057245,
      "grad_norm": 3.5192058086395264,
      "learning_rate": 8.872336987670969e-05,
      "loss": 0.1129,
      "step": 5470
    },
    {
      "epoch": 0.22121362216583945,
      "grad_norm": 7.238174915313721,
      "learning_rate": 8.871933518445154e-05,
      "loss": 0.2309,
      "step": 5471
    },
    {
      "epoch": 0.22125405602110648,
      "grad_norm": 8.418448448181152,
      "learning_rate": 8.871529986230508e-05,
      "loss": 0.4156,
      "step": 5472
    },
    {
      "epoch": 0.22129448987637348,
      "grad_norm": 4.049168109893799,
      "learning_rate": 8.871126391033595e-05,
      "loss": 0.1335,
      "step": 5473
    },
    {
      "epoch": 0.22133492373164051,
      "grad_norm": 5.120578765869141,
      "learning_rate": 8.870722732860984e-05,
      "loss": 0.11,
      "step": 5474
    },
    {
      "epoch": 0.22137535758690752,
      "grad_norm": 6.028683662414551,
      "learning_rate": 8.870319011719236e-05,
      "loss": 0.2396,
      "step": 5475
    },
    {
      "epoch": 0.22141579144217452,
      "grad_norm": 9.519779205322266,
      "learning_rate": 8.869915227614924e-05,
      "loss": 0.1915,
      "step": 5476
    },
    {
      "epoch": 0.22145622529744155,
      "grad_norm": 2.5778841972351074,
      "learning_rate": 8.869511380554613e-05,
      "loss": 0.1184,
      "step": 5477
    },
    {
      "epoch": 0.22149665915270855,
      "grad_norm": 4.181035995483398,
      "learning_rate": 8.869107470544874e-05,
      "loss": 0.124,
      "step": 5478
    },
    {
      "epoch": 0.22153709300797558,
      "grad_norm": 2.5753984451293945,
      "learning_rate": 8.86870349759228e-05,
      "loss": 0.1235,
      "step": 5479
    },
    {
      "epoch": 0.22157752686324259,
      "grad_norm": 2.5421254634857178,
      "learning_rate": 8.868299461703403e-05,
      "loss": 0.0734,
      "step": 5480
    },
    {
      "epoch": 0.22161796071850962,
      "grad_norm": 4.001224994659424,
      "learning_rate": 8.867895362884811e-05,
      "loss": 0.125,
      "step": 5481
    },
    {
      "epoch": 0.22165839457377662,
      "grad_norm": 4.332638740539551,
      "learning_rate": 8.867491201143082e-05,
      "loss": 0.1328,
      "step": 5482
    },
    {
      "epoch": 0.22169882842904365,
      "grad_norm": 12.000165939331055,
      "learning_rate": 8.86708697648479e-05,
      "loss": 0.2451,
      "step": 5483
    },
    {
      "epoch": 0.22173926228431065,
      "grad_norm": 3.8164045810699463,
      "learning_rate": 8.866682688916511e-05,
      "loss": 0.1248,
      "step": 5484
    },
    {
      "epoch": 0.22177969613957768,
      "grad_norm": 5.635714054107666,
      "learning_rate": 8.866278338444821e-05,
      "loss": 0.1718,
      "step": 5485
    },
    {
      "epoch": 0.22182012999484468,
      "grad_norm": 4.019015312194824,
      "learning_rate": 8.8658739250763e-05,
      "loss": 0.1307,
      "step": 5486
    },
    {
      "epoch": 0.2218605638501117,
      "grad_norm": 3.732320785522461,
      "learning_rate": 8.865469448817526e-05,
      "loss": 0.1803,
      "step": 5487
    },
    {
      "epoch": 0.22190099770537872,
      "grad_norm": 3.5771353244781494,
      "learning_rate": 8.865064909675077e-05,
      "loss": 0.1687,
      "step": 5488
    },
    {
      "epoch": 0.22194143156064572,
      "grad_norm": 4.400257587432861,
      "learning_rate": 8.864660307655537e-05,
      "loss": 0.1414,
      "step": 5489
    },
    {
      "epoch": 0.22198186541591275,
      "grad_norm": 1.5285699367523193,
      "learning_rate": 8.864255642765487e-05,
      "loss": 0.0325,
      "step": 5490
    },
    {
      "epoch": 0.22202229927117975,
      "grad_norm": 3.830777883529663,
      "learning_rate": 8.863850915011511e-05,
      "loss": 0.148,
      "step": 5491
    },
    {
      "epoch": 0.22206273312644678,
      "grad_norm": 3.936600923538208,
      "learning_rate": 8.863446124400191e-05,
      "loss": 0.1237,
      "step": 5492
    },
    {
      "epoch": 0.22210316698171378,
      "grad_norm": 3.7365736961364746,
      "learning_rate": 8.863041270938115e-05,
      "loss": 0.1893,
      "step": 5493
    },
    {
      "epoch": 0.22214360083698081,
      "grad_norm": 2.185232162475586,
      "learning_rate": 8.862636354631866e-05,
      "loss": 0.1257,
      "step": 5494
    },
    {
      "epoch": 0.22218403469224782,
      "grad_norm": 3.416292667388916,
      "learning_rate": 8.862231375488033e-05,
      "loss": 0.0721,
      "step": 5495
    },
    {
      "epoch": 0.22222446854751485,
      "grad_norm": 2.414379835128784,
      "learning_rate": 8.861826333513206e-05,
      "loss": 0.1184,
      "step": 5496
    },
    {
      "epoch": 0.22226490240278185,
      "grad_norm": 6.208682537078857,
      "learning_rate": 8.861421228713969e-05,
      "loss": 0.1817,
      "step": 5497
    },
    {
      "epoch": 0.22230533625804885,
      "grad_norm": 4.2622599601745605,
      "learning_rate": 8.861016061096918e-05,
      "loss": 0.1942,
      "step": 5498
    },
    {
      "epoch": 0.22234577011331588,
      "grad_norm": 3.1867921352386475,
      "learning_rate": 8.860610830668641e-05,
      "loss": 0.0584,
      "step": 5499
    },
    {
      "epoch": 0.22238620396858289,
      "grad_norm": 1.0862969160079956,
      "learning_rate": 8.860205537435732e-05,
      "loss": 0.0489,
      "step": 5500
    },
    {
      "epoch": 0.22242663782384992,
      "grad_norm": 5.063252925872803,
      "learning_rate": 8.859800181404782e-05,
      "loss": 0.2788,
      "step": 5501
    },
    {
      "epoch": 0.22246707167911692,
      "grad_norm": 3.8565986156463623,
      "learning_rate": 8.859394762582388e-05,
      "loss": 0.1934,
      "step": 5502
    },
    {
      "epoch": 0.22250750553438395,
      "grad_norm": 3.0188145637512207,
      "learning_rate": 8.858989280975144e-05,
      "loss": 0.1391,
      "step": 5503
    },
    {
      "epoch": 0.22254793938965095,
      "grad_norm": 8.409645080566406,
      "learning_rate": 8.858583736589648e-05,
      "loss": 0.2834,
      "step": 5504
    },
    {
      "epoch": 0.22258837324491798,
      "grad_norm": 7.941526412963867,
      "learning_rate": 8.858178129432493e-05,
      "loss": 0.1926,
      "step": 5505
    },
    {
      "epoch": 0.22262880710018498,
      "grad_norm": 10.991561889648438,
      "learning_rate": 8.857772459510282e-05,
      "loss": 0.2699,
      "step": 5506
    },
    {
      "epoch": 0.222669240955452,
      "grad_norm": 6.304360389709473,
      "learning_rate": 8.857366726829613e-05,
      "loss": 0.1054,
      "step": 5507
    },
    {
      "epoch": 0.22270967481071902,
      "grad_norm": 7.175690174102783,
      "learning_rate": 8.856960931397087e-05,
      "loss": 0.19,
      "step": 5508
    },
    {
      "epoch": 0.22275010866598602,
      "grad_norm": 3.104278326034546,
      "learning_rate": 8.856555073219303e-05,
      "loss": 0.0953,
      "step": 5509
    },
    {
      "epoch": 0.22279054252125305,
      "grad_norm": 3.33054518699646,
      "learning_rate": 8.856149152302867e-05,
      "loss": 0.1467,
      "step": 5510
    },
    {
      "epoch": 0.22283097637652005,
      "grad_norm": 7.779376983642578,
      "learning_rate": 8.855743168654381e-05,
      "loss": 0.3426,
      "step": 5511
    },
    {
      "epoch": 0.22287141023178708,
      "grad_norm": 7.885648727416992,
      "learning_rate": 8.85533712228045e-05,
      "loss": 0.2963,
      "step": 5512
    },
    {
      "epoch": 0.22291184408705408,
      "grad_norm": 11.896480560302734,
      "learning_rate": 8.854931013187677e-05,
      "loss": 0.4704,
      "step": 5513
    },
    {
      "epoch": 0.22295227794232111,
      "grad_norm": 4.120419025421143,
      "learning_rate": 8.854524841382673e-05,
      "loss": 0.1253,
      "step": 5514
    },
    {
      "epoch": 0.22299271179758812,
      "grad_norm": 5.346358776092529,
      "learning_rate": 8.854118606872041e-05,
      "loss": 0.1534,
      "step": 5515
    },
    {
      "epoch": 0.22303314565285515,
      "grad_norm": 3.8812921047210693,
      "learning_rate": 8.853712309662394e-05,
      "loss": 0.2462,
      "step": 5516
    },
    {
      "epoch": 0.22307357950812215,
      "grad_norm": 6.470430850982666,
      "learning_rate": 8.85330594976034e-05,
      "loss": 0.225,
      "step": 5517
    },
    {
      "epoch": 0.22311401336338915,
      "grad_norm": 4.764116287231445,
      "learning_rate": 8.852899527172489e-05,
      "loss": 0.1089,
      "step": 5518
    },
    {
      "epoch": 0.22315444721865618,
      "grad_norm": 7.5875630378723145,
      "learning_rate": 8.852493041905451e-05,
      "loss": 0.2462,
      "step": 5519
    },
    {
      "epoch": 0.22319488107392318,
      "grad_norm": 9.999929428100586,
      "learning_rate": 8.852086493965843e-05,
      "loss": 0.3527,
      "step": 5520
    },
    {
      "epoch": 0.22323531492919022,
      "grad_norm": 11.076635360717773,
      "learning_rate": 8.851679883360276e-05,
      "loss": 0.3427,
      "step": 5521
    },
    {
      "epoch": 0.22327574878445722,
      "grad_norm": 3.043954610824585,
      "learning_rate": 8.851273210095365e-05,
      "loss": 0.1019,
      "step": 5522
    },
    {
      "epoch": 0.22331618263972425,
      "grad_norm": 6.286318778991699,
      "learning_rate": 8.850866474177728e-05,
      "loss": 0.2068,
      "step": 5523
    },
    {
      "epoch": 0.22335661649499125,
      "grad_norm": 5.649576187133789,
      "learning_rate": 8.850459675613977e-05,
      "loss": 0.1782,
      "step": 5524
    },
    {
      "epoch": 0.22339705035025828,
      "grad_norm": 1.9968878030776978,
      "learning_rate": 8.850052814410734e-05,
      "loss": 0.1763,
      "step": 5525
    },
    {
      "epoch": 0.22343748420552528,
      "grad_norm": 8.372543334960938,
      "learning_rate": 8.849645890574617e-05,
      "loss": 0.3333,
      "step": 5526
    },
    {
      "epoch": 0.2234779180607923,
      "grad_norm": 2.5873026847839355,
      "learning_rate": 8.849238904112244e-05,
      "loss": 0.1851,
      "step": 5527
    },
    {
      "epoch": 0.22351835191605932,
      "grad_norm": 2.7721164226531982,
      "learning_rate": 8.84883185503024e-05,
      "loss": 0.2194,
      "step": 5528
    },
    {
      "epoch": 0.22355878577132632,
      "grad_norm": 2.356398105621338,
      "learning_rate": 8.848424743335222e-05,
      "loss": 0.1625,
      "step": 5529
    },
    {
      "epoch": 0.22359921962659335,
      "grad_norm": 2.87585186958313,
      "learning_rate": 8.848017569033815e-05,
      "loss": 0.1097,
      "step": 5530
    },
    {
      "epoch": 0.22363965348186035,
      "grad_norm": 4.976321220397949,
      "learning_rate": 8.847610332132643e-05,
      "loss": 0.1075,
      "step": 5531
    },
    {
      "epoch": 0.22368008733712738,
      "grad_norm": 3.167246103286743,
      "learning_rate": 8.847203032638332e-05,
      "loss": 0.224,
      "step": 5532
    },
    {
      "epoch": 0.22372052119239438,
      "grad_norm": 6.384296417236328,
      "learning_rate": 8.846795670557508e-05,
      "loss": 0.1324,
      "step": 5533
    },
    {
      "epoch": 0.22376095504766141,
      "grad_norm": 5.064076900482178,
      "learning_rate": 8.846388245896796e-05,
      "loss": 0.1672,
      "step": 5534
    },
    {
      "epoch": 0.22380138890292842,
      "grad_norm": 3.6748969554901123,
      "learning_rate": 8.845980758662823e-05,
      "loss": 0.1189,
      "step": 5535
    },
    {
      "epoch": 0.22384182275819545,
      "grad_norm": 4.205347061157227,
      "learning_rate": 8.84557320886222e-05,
      "loss": 0.2459,
      "step": 5536
    },
    {
      "epoch": 0.22388225661346245,
      "grad_norm": 2.939605236053467,
      "learning_rate": 8.84516559650162e-05,
      "loss": 0.1451,
      "step": 5537
    },
    {
      "epoch": 0.22392269046872948,
      "grad_norm": 2.417372465133667,
      "learning_rate": 8.84475792158765e-05,
      "loss": 0.1486,
      "step": 5538
    },
    {
      "epoch": 0.22396312432399648,
      "grad_norm": 3.650648593902588,
      "learning_rate": 8.844350184126942e-05,
      "loss": 0.1512,
      "step": 5539
    },
    {
      "epoch": 0.22400355817926348,
      "grad_norm": 2.423804998397827,
      "learning_rate": 8.843942384126132e-05,
      "loss": 0.1402,
      "step": 5540
    },
    {
      "epoch": 0.22404399203453051,
      "grad_norm": 2.870243549346924,
      "learning_rate": 8.84353452159185e-05,
      "loss": 0.096,
      "step": 5541
    },
    {
      "epoch": 0.22408442588979752,
      "grad_norm": 3.101025104522705,
      "learning_rate": 8.843126596530735e-05,
      "loss": 0.1208,
      "step": 5542
    },
    {
      "epoch": 0.22412485974506455,
      "grad_norm": 4.065934658050537,
      "learning_rate": 8.842718608949422e-05,
      "loss": 0.1605,
      "step": 5543
    },
    {
      "epoch": 0.22416529360033155,
      "grad_norm": 4.258211135864258,
      "learning_rate": 8.842310558854548e-05,
      "loss": 0.1291,
      "step": 5544
    },
    {
      "epoch": 0.22420572745559858,
      "grad_norm": 4.689475059509277,
      "learning_rate": 8.84190244625275e-05,
      "loss": 0.1332,
      "step": 5545
    },
    {
      "epoch": 0.22424616131086558,
      "grad_norm": 4.3837480545043945,
      "learning_rate": 8.841494271150667e-05,
      "loss": 0.2159,
      "step": 5546
    },
    {
      "epoch": 0.2242865951661326,
      "grad_norm": 9.329647064208984,
      "learning_rate": 8.841086033554944e-05,
      "loss": 0.2654,
      "step": 5547
    },
    {
      "epoch": 0.22432702902139962,
      "grad_norm": 5.497567176818848,
      "learning_rate": 8.840677733472215e-05,
      "loss": 0.2377,
      "step": 5548
    },
    {
      "epoch": 0.22436746287666665,
      "grad_norm": 7.371769428253174,
      "learning_rate": 8.84026937090913e-05,
      "loss": 0.2328,
      "step": 5549
    },
    {
      "epoch": 0.22440789673193365,
      "grad_norm": 5.3459014892578125,
      "learning_rate": 8.839860945872324e-05,
      "loss": 0.2435,
      "step": 5550
    },
    {
      "epoch": 0.22444833058720065,
      "grad_norm": 14.5158109664917,
      "learning_rate": 8.839452458368447e-05,
      "loss": 0.3203,
      "step": 5551
    },
    {
      "epoch": 0.22448876444246768,
      "grad_norm": 9.902429580688477,
      "learning_rate": 8.839043908404144e-05,
      "loss": 0.3551,
      "step": 5552
    },
    {
      "epoch": 0.22452919829773468,
      "grad_norm": 6.617310523986816,
      "learning_rate": 8.838635295986058e-05,
      "loss": 0.1588,
      "step": 5553
    },
    {
      "epoch": 0.2245696321530017,
      "grad_norm": 13.468600273132324,
      "learning_rate": 8.838226621120839e-05,
      "loss": 0.3369,
      "step": 5554
    },
    {
      "epoch": 0.22461006600826872,
      "grad_norm": 6.561675548553467,
      "learning_rate": 8.837817883815135e-05,
      "loss": 0.2608,
      "step": 5555
    },
    {
      "epoch": 0.22465049986353575,
      "grad_norm": 12.042304992675781,
      "learning_rate": 8.837409084075596e-05,
      "loss": 0.2483,
      "step": 5556
    },
    {
      "epoch": 0.22469093371880275,
      "grad_norm": 6.338623046875,
      "learning_rate": 8.83700022190887e-05,
      "loss": 0.128,
      "step": 5557
    },
    {
      "epoch": 0.22473136757406978,
      "grad_norm": 3.868267059326172,
      "learning_rate": 8.836591297321611e-05,
      "loss": 0.1756,
      "step": 5558
    },
    {
      "epoch": 0.22477180142933678,
      "grad_norm": 2.4094626903533936,
      "learning_rate": 8.836182310320468e-05,
      "loss": 0.2866,
      "step": 5559
    },
    {
      "epoch": 0.2248122352846038,
      "grad_norm": 4.931764602661133,
      "learning_rate": 8.835773260912099e-05,
      "loss": 0.2337,
      "step": 5560
    },
    {
      "epoch": 0.22485266913987081,
      "grad_norm": 5.78426456451416,
      "learning_rate": 8.835364149103156e-05,
      "loss": 0.136,
      "step": 5561
    },
    {
      "epoch": 0.22489310299513782,
      "grad_norm": 6.0288214683532715,
      "learning_rate": 8.834954974900293e-05,
      "loss": 0.3593,
      "step": 5562
    },
    {
      "epoch": 0.22493353685040485,
      "grad_norm": 4.931115627288818,
      "learning_rate": 8.83454573831017e-05,
      "loss": 0.1383,
      "step": 5563
    },
    {
      "epoch": 0.22497397070567185,
      "grad_norm": 5.857620716094971,
      "learning_rate": 8.83413643933944e-05,
      "loss": 0.1839,
      "step": 5564
    },
    {
      "epoch": 0.22501440456093888,
      "grad_norm": 2.3521385192871094,
      "learning_rate": 8.833727077994766e-05,
      "loss": 0.2118,
      "step": 5565
    },
    {
      "epoch": 0.22505483841620588,
      "grad_norm": 9.719688415527344,
      "learning_rate": 8.833317654282807e-05,
      "loss": 0.2624,
      "step": 5566
    },
    {
      "epoch": 0.2250952722714729,
      "grad_norm": 3.721994400024414,
      "learning_rate": 8.832908168210219e-05,
      "loss": 0.1095,
      "step": 5567
    },
    {
      "epoch": 0.22513570612673992,
      "grad_norm": 5.088572978973389,
      "learning_rate": 8.83249861978367e-05,
      "loss": 0.3447,
      "step": 5568
    },
    {
      "epoch": 0.22517613998200695,
      "grad_norm": 4.827566623687744,
      "learning_rate": 8.832089009009815e-05,
      "loss": 0.1607,
      "step": 5569
    },
    {
      "epoch": 0.22521657383727395,
      "grad_norm": 2.273463726043701,
      "learning_rate": 8.831679335895324e-05,
      "loss": 0.0745,
      "step": 5570
    },
    {
      "epoch": 0.22525700769254098,
      "grad_norm": 2.7915804386138916,
      "learning_rate": 8.83126960044686e-05,
      "loss": 0.0927,
      "step": 5571
    },
    {
      "epoch": 0.22529744154780798,
      "grad_norm": 8.259066581726074,
      "learning_rate": 8.830859802671087e-05,
      "loss": 0.2077,
      "step": 5572
    },
    {
      "epoch": 0.22533787540307498,
      "grad_norm": 3.6964385509490967,
      "learning_rate": 8.830449942574672e-05,
      "loss": 0.1531,
      "step": 5573
    },
    {
      "epoch": 0.225378309258342,
      "grad_norm": 3.6525774002075195,
      "learning_rate": 8.830040020164284e-05,
      "loss": 0.1374,
      "step": 5574
    },
    {
      "epoch": 0.22541874311360902,
      "grad_norm": 3.2001147270202637,
      "learning_rate": 8.82963003544659e-05,
      "loss": 0.0939,
      "step": 5575
    },
    {
      "epoch": 0.22545917696887605,
      "grad_norm": 4.361302375793457,
      "learning_rate": 8.829219988428262e-05,
      "loss": 0.2784,
      "step": 5576
    },
    {
      "epoch": 0.22549961082414305,
      "grad_norm": 4.6830244064331055,
      "learning_rate": 8.82880987911597e-05,
      "loss": 0.2087,
      "step": 5577
    },
    {
      "epoch": 0.22554004467941008,
      "grad_norm": 6.494311332702637,
      "learning_rate": 8.828399707516382e-05,
      "loss": 0.1896,
      "step": 5578
    },
    {
      "epoch": 0.22558047853467708,
      "grad_norm": 2.033996105194092,
      "learning_rate": 8.827989473636173e-05,
      "loss": 0.1225,
      "step": 5579
    },
    {
      "epoch": 0.2256209123899441,
      "grad_norm": 5.2412309646606445,
      "learning_rate": 8.827579177482019e-05,
      "loss": 0.1505,
      "step": 5580
    },
    {
      "epoch": 0.22566134624521111,
      "grad_norm": 6.4794440269470215,
      "learning_rate": 8.827168819060592e-05,
      "loss": 0.194,
      "step": 5581
    },
    {
      "epoch": 0.22570178010047812,
      "grad_norm": 4.51427698135376,
      "learning_rate": 8.82675839837857e-05,
      "loss": 0.1477,
      "step": 5582
    },
    {
      "epoch": 0.22574221395574515,
      "grad_norm": 4.01408576965332,
      "learning_rate": 8.826347915442628e-05,
      "loss": 0.1378,
      "step": 5583
    },
    {
      "epoch": 0.22578264781101215,
      "grad_norm": 5.341332912445068,
      "learning_rate": 8.825937370259443e-05,
      "loss": 0.1733,
      "step": 5584
    },
    {
      "epoch": 0.22582308166627918,
      "grad_norm": 3.4240050315856934,
      "learning_rate": 8.825526762835696e-05,
      "loss": 0.1598,
      "step": 5585
    },
    {
      "epoch": 0.22586351552154618,
      "grad_norm": 3.9128031730651855,
      "learning_rate": 8.825116093178066e-05,
      "loss": 0.1305,
      "step": 5586
    },
    {
      "epoch": 0.2259039493768132,
      "grad_norm": 3.4751100540161133,
      "learning_rate": 8.824705361293234e-05,
      "loss": 0.0922,
      "step": 5587
    },
    {
      "epoch": 0.22594438323208021,
      "grad_norm": 2.8438680171966553,
      "learning_rate": 8.82429456718788e-05,
      "loss": 0.1586,
      "step": 5588
    },
    {
      "epoch": 0.22598481708734725,
      "grad_norm": 5.316544055938721,
      "learning_rate": 8.823883710868688e-05,
      "loss": 0.2216,
      "step": 5589
    },
    {
      "epoch": 0.22602525094261425,
      "grad_norm": 3.2949109077453613,
      "learning_rate": 8.823472792342344e-05,
      "loss": 0.0969,
      "step": 5590
    },
    {
      "epoch": 0.22606568479788128,
      "grad_norm": 13.559076309204102,
      "learning_rate": 8.823061811615529e-05,
      "loss": 0.2975,
      "step": 5591
    },
    {
      "epoch": 0.22610611865314828,
      "grad_norm": 10.484284400939941,
      "learning_rate": 8.822650768694932e-05,
      "loss": 0.2537,
      "step": 5592
    },
    {
      "epoch": 0.22614655250841528,
      "grad_norm": 5.503856658935547,
      "learning_rate": 8.822239663587238e-05,
      "loss": 0.1074,
      "step": 5593
    },
    {
      "epoch": 0.2261869863636823,
      "grad_norm": 4.814883232116699,
      "learning_rate": 8.821828496299134e-05,
      "loss": 0.1354,
      "step": 5594
    },
    {
      "epoch": 0.22622742021894932,
      "grad_norm": 4.239748954772949,
      "learning_rate": 8.821417266837314e-05,
      "loss": 0.155,
      "step": 5595
    },
    {
      "epoch": 0.22626785407421635,
      "grad_norm": 3.0424327850341797,
      "learning_rate": 8.82100597520846e-05,
      "loss": 0.1141,
      "step": 5596
    },
    {
      "epoch": 0.22630828792948335,
      "grad_norm": 1.9998880624771118,
      "learning_rate": 8.820594621419269e-05,
      "loss": 0.0761,
      "step": 5597
    },
    {
      "epoch": 0.22634872178475038,
      "grad_norm": 6.920135021209717,
      "learning_rate": 8.820183205476431e-05,
      "loss": 0.2485,
      "step": 5598
    },
    {
      "epoch": 0.22638915564001738,
      "grad_norm": 6.219224452972412,
      "learning_rate": 8.819771727386638e-05,
      "loss": 0.2299,
      "step": 5599
    },
    {
      "epoch": 0.2264295894952844,
      "grad_norm": 3.9673335552215576,
      "learning_rate": 8.819360187156587e-05,
      "loss": 0.1345,
      "step": 5600
    },
    {
      "epoch": 0.2264700233505514,
      "grad_norm": 10.38260269165039,
      "learning_rate": 8.81894858479297e-05,
      "loss": 0.3287,
      "step": 5601
    },
    {
      "epoch": 0.22651045720581844,
      "grad_norm": 2.4500370025634766,
      "learning_rate": 8.818536920302484e-05,
      "loss": 0.1115,
      "step": 5602
    },
    {
      "epoch": 0.22655089106108545,
      "grad_norm": 3.2346901893615723,
      "learning_rate": 8.818125193691826e-05,
      "loss": 0.2033,
      "step": 5603
    },
    {
      "epoch": 0.22659132491635245,
      "grad_norm": 7.0145978927612305,
      "learning_rate": 8.817713404967694e-05,
      "loss": 0.2318,
      "step": 5604
    },
    {
      "epoch": 0.22663175877161948,
      "grad_norm": 9.668561935424805,
      "learning_rate": 8.817301554136786e-05,
      "loss": 0.3077,
      "step": 5605
    },
    {
      "epoch": 0.22667219262688648,
      "grad_norm": 5.305925369262695,
      "learning_rate": 8.816889641205803e-05,
      "loss": 0.1642,
      "step": 5606
    },
    {
      "epoch": 0.2267126264821535,
      "grad_norm": 4.054882049560547,
      "learning_rate": 8.816477666181447e-05,
      "loss": 0.221,
      "step": 5607
    },
    {
      "epoch": 0.22675306033742051,
      "grad_norm": 6.026086330413818,
      "learning_rate": 8.816065629070416e-05,
      "loss": 0.1445,
      "step": 5608
    },
    {
      "epoch": 0.22679349419268754,
      "grad_norm": 6.513998031616211,
      "learning_rate": 8.81565352987942e-05,
      "loss": 0.1823,
      "step": 5609
    },
    {
      "epoch": 0.22683392804795455,
      "grad_norm": 3.448760747909546,
      "learning_rate": 8.815241368615157e-05,
      "loss": 0.1293,
      "step": 5610
    },
    {
      "epoch": 0.22687436190322158,
      "grad_norm": 3.9986343383789062,
      "learning_rate": 8.814829145284333e-05,
      "loss": 0.185,
      "step": 5611
    },
    {
      "epoch": 0.22691479575848858,
      "grad_norm": 4.70590353012085,
      "learning_rate": 8.814416859893657e-05,
      "loss": 0.0826,
      "step": 5612
    },
    {
      "epoch": 0.2269552296137556,
      "grad_norm": 2.818089723587036,
      "learning_rate": 8.814004512449834e-05,
      "loss": 0.2406,
      "step": 5613
    },
    {
      "epoch": 0.2269956634690226,
      "grad_norm": 2.8246779441833496,
      "learning_rate": 8.813592102959573e-05,
      "loss": 0.0627,
      "step": 5614
    },
    {
      "epoch": 0.22703609732428962,
      "grad_norm": 9.609831809997559,
      "learning_rate": 8.813179631429581e-05,
      "loss": 0.1926,
      "step": 5615
    },
    {
      "epoch": 0.22707653117955665,
      "grad_norm": 3.8344485759735107,
      "learning_rate": 8.81276709786657e-05,
      "loss": 0.172,
      "step": 5616
    },
    {
      "epoch": 0.22711696503482365,
      "grad_norm": 4.41788911819458,
      "learning_rate": 8.812354502277252e-05,
      "loss": 0.1185,
      "step": 5617
    },
    {
      "epoch": 0.22715739889009068,
      "grad_norm": 2.3496484756469727,
      "learning_rate": 8.811941844668335e-05,
      "loss": 0.0708,
      "step": 5618
    },
    {
      "epoch": 0.22719783274535768,
      "grad_norm": 2.704084873199463,
      "learning_rate": 8.811529125046537e-05,
      "loss": 0.063,
      "step": 5619
    },
    {
      "epoch": 0.2272382666006247,
      "grad_norm": 3.1632213592529297,
      "learning_rate": 8.811116343418571e-05,
      "loss": 0.1273,
      "step": 5620
    },
    {
      "epoch": 0.2272787004558917,
      "grad_norm": 5.178030014038086,
      "learning_rate": 8.810703499791149e-05,
      "loss": 0.144,
      "step": 5621
    },
    {
      "epoch": 0.22731913431115874,
      "grad_norm": 4.727217197418213,
      "learning_rate": 8.81029059417099e-05,
      "loss": 0.2651,
      "step": 5622
    },
    {
      "epoch": 0.22735956816642575,
      "grad_norm": 5.951588153839111,
      "learning_rate": 8.809877626564811e-05,
      "loss": 0.3089,
      "step": 5623
    },
    {
      "epoch": 0.22740000202169278,
      "grad_norm": 5.231448650360107,
      "learning_rate": 8.809464596979328e-05,
      "loss": 0.1849,
      "step": 5624
    },
    {
      "epoch": 0.22744043587695978,
      "grad_norm": 3.884702205657959,
      "learning_rate": 8.809051505421265e-05,
      "loss": 0.1365,
      "step": 5625
    },
    {
      "epoch": 0.22748086973222678,
      "grad_norm": 4.465366363525391,
      "learning_rate": 8.808638351897338e-05,
      "loss": 0.1019,
      "step": 5626
    },
    {
      "epoch": 0.2275213035874938,
      "grad_norm": 3.602226734161377,
      "learning_rate": 8.808225136414269e-05,
      "loss": 0.142,
      "step": 5627
    },
    {
      "epoch": 0.22756173744276081,
      "grad_norm": 3.752310276031494,
      "learning_rate": 8.80781185897878e-05,
      "loss": 0.1634,
      "step": 5628
    },
    {
      "epoch": 0.22760217129802784,
      "grad_norm": 2.9510955810546875,
      "learning_rate": 8.807398519597596e-05,
      "loss": 0.1995,
      "step": 5629
    },
    {
      "epoch": 0.22764260515329485,
      "grad_norm": 3.5611281394958496,
      "learning_rate": 8.806985118277438e-05,
      "loss": 0.2171,
      "step": 5630
    },
    {
      "epoch": 0.22768303900856188,
      "grad_norm": 2.823993444442749,
      "learning_rate": 8.806571655025036e-05,
      "loss": 0.1431,
      "step": 5631
    },
    {
      "epoch": 0.22772347286382888,
      "grad_norm": 3.2524046897888184,
      "learning_rate": 8.806158129847112e-05,
      "loss": 0.1755,
      "step": 5632
    },
    {
      "epoch": 0.2277639067190959,
      "grad_norm": 7.895308017730713,
      "learning_rate": 8.805744542750395e-05,
      "loss": 0.196,
      "step": 5633
    },
    {
      "epoch": 0.2278043405743629,
      "grad_norm": 3.4701883792877197,
      "learning_rate": 8.805330893741613e-05,
      "loss": 0.1093,
      "step": 5634
    },
    {
      "epoch": 0.22784477442962994,
      "grad_norm": 5.96124792098999,
      "learning_rate": 8.804917182827494e-05,
      "loss": 0.2199,
      "step": 5635
    },
    {
      "epoch": 0.22788520828489695,
      "grad_norm": 2.6076436042785645,
      "learning_rate": 8.80450341001477e-05,
      "loss": 0.1297,
      "step": 5636
    },
    {
      "epoch": 0.22792564214016395,
      "grad_norm": 6.441141128540039,
      "learning_rate": 8.804089575310174e-05,
      "loss": 0.319,
      "step": 5637
    },
    {
      "epoch": 0.22796607599543098,
      "grad_norm": 2.1003963947296143,
      "learning_rate": 8.803675678720434e-05,
      "loss": 0.213,
      "step": 5638
    },
    {
      "epoch": 0.22800650985069798,
      "grad_norm": 4.632483959197998,
      "learning_rate": 8.803261720252285e-05,
      "loss": 0.2464,
      "step": 5639
    },
    {
      "epoch": 0.228046943705965,
      "grad_norm": 1.3712785243988037,
      "learning_rate": 8.802847699912464e-05,
      "loss": 0.0512,
      "step": 5640
    },
    {
      "epoch": 0.228087377561232,
      "grad_norm": 4.605969429016113,
      "learning_rate": 8.802433617707703e-05,
      "loss": 0.1352,
      "step": 5641
    },
    {
      "epoch": 0.22812781141649904,
      "grad_norm": 9.304512977600098,
      "learning_rate": 8.802019473644739e-05,
      "loss": 0.2339,
      "step": 5642
    },
    {
      "epoch": 0.22816824527176605,
      "grad_norm": 8.498616218566895,
      "learning_rate": 8.801605267730309e-05,
      "loss": 0.252,
      "step": 5643
    },
    {
      "epoch": 0.22820867912703308,
      "grad_norm": 10.316244125366211,
      "learning_rate": 8.801190999971153e-05,
      "loss": 0.2521,
      "step": 5644
    },
    {
      "epoch": 0.22824911298230008,
      "grad_norm": 5.377568244934082,
      "learning_rate": 8.800776670374007e-05,
      "loss": 0.1238,
      "step": 5645
    },
    {
      "epoch": 0.22828954683756708,
      "grad_norm": 4.451637268066406,
      "learning_rate": 8.800362278945616e-05,
      "loss": 0.3562,
      "step": 5646
    },
    {
      "epoch": 0.2283299806928341,
      "grad_norm": 2.483294725418091,
      "learning_rate": 8.799947825692718e-05,
      "loss": 0.137,
      "step": 5647
    },
    {
      "epoch": 0.2283704145481011,
      "grad_norm": 7.848944187164307,
      "learning_rate": 8.799533310622057e-05,
      "loss": 0.2175,
      "step": 5648
    },
    {
      "epoch": 0.22841084840336814,
      "grad_norm": 16.201627731323242,
      "learning_rate": 8.799118733740377e-05,
      "loss": 0.4497,
      "step": 5649
    },
    {
      "epoch": 0.22845128225863515,
      "grad_norm": 5.268087387084961,
      "learning_rate": 8.798704095054418e-05,
      "loss": 0.2038,
      "step": 5650
    },
    {
      "epoch": 0.22849171611390218,
      "grad_norm": 3.106041431427002,
      "learning_rate": 8.798289394570931e-05,
      "loss": 0.2137,
      "step": 5651
    },
    {
      "epoch": 0.22853214996916918,
      "grad_norm": 3.4316821098327637,
      "learning_rate": 8.797874632296658e-05,
      "loss": 0.1966,
      "step": 5652
    },
    {
      "epoch": 0.2285725838244362,
      "grad_norm": 7.747198104858398,
      "learning_rate": 8.797459808238349e-05,
      "loss": 0.1783,
      "step": 5653
    },
    {
      "epoch": 0.2286130176797032,
      "grad_norm": 3.277883529663086,
      "learning_rate": 8.797044922402751e-05,
      "loss": 0.218,
      "step": 5654
    },
    {
      "epoch": 0.22865345153497024,
      "grad_norm": 3.436851739883423,
      "learning_rate": 8.796629974796614e-05,
      "loss": 0.0916,
      "step": 5655
    },
    {
      "epoch": 0.22869388539023724,
      "grad_norm": 2.523634672164917,
      "learning_rate": 8.796214965426688e-05,
      "loss": 0.1815,
      "step": 5656
    },
    {
      "epoch": 0.22873431924550425,
      "grad_norm": 7.130584716796875,
      "learning_rate": 8.795799894299725e-05,
      "loss": 0.1477,
      "step": 5657
    },
    {
      "epoch": 0.22877475310077128,
      "grad_norm": 3.3015730381011963,
      "learning_rate": 8.795384761422478e-05,
      "loss": 0.1821,
      "step": 5658
    },
    {
      "epoch": 0.22881518695603828,
      "grad_norm": 2.5596835613250732,
      "learning_rate": 8.794969566801699e-05,
      "loss": 0.2046,
      "step": 5659
    },
    {
      "epoch": 0.2288556208113053,
      "grad_norm": 3.6378467082977295,
      "learning_rate": 8.794554310444142e-05,
      "loss": 0.3056,
      "step": 5660
    },
    {
      "epoch": 0.2288960546665723,
      "grad_norm": 6.256381511688232,
      "learning_rate": 8.794138992356563e-05,
      "loss": 0.3074,
      "step": 5661
    },
    {
      "epoch": 0.22893648852183934,
      "grad_norm": 5.03485107421875,
      "learning_rate": 8.79372361254572e-05,
      "loss": 0.172,
      "step": 5662
    },
    {
      "epoch": 0.22897692237710635,
      "grad_norm": 4.5250115394592285,
      "learning_rate": 8.793308171018369e-05,
      "loss": 0.2311,
      "step": 5663
    },
    {
      "epoch": 0.22901735623237338,
      "grad_norm": 3.8390610218048096,
      "learning_rate": 8.792892667781268e-05,
      "loss": 0.1276,
      "step": 5664
    },
    {
      "epoch": 0.22905779008764038,
      "grad_norm": 4.40824031829834,
      "learning_rate": 8.792477102841178e-05,
      "loss": 0.1903,
      "step": 5665
    },
    {
      "epoch": 0.2290982239429074,
      "grad_norm": 7.8341593742370605,
      "learning_rate": 8.792061476204858e-05,
      "loss": 0.2277,
      "step": 5666
    },
    {
      "epoch": 0.2291386577981744,
      "grad_norm": 2.2590179443359375,
      "learning_rate": 8.791645787879069e-05,
      "loss": 0.0797,
      "step": 5667
    },
    {
      "epoch": 0.2291790916534414,
      "grad_norm": 2.4610371589660645,
      "learning_rate": 8.791230037870575e-05,
      "loss": 0.125,
      "step": 5668
    },
    {
      "epoch": 0.22921952550870844,
      "grad_norm": 6.41936731338501,
      "learning_rate": 8.790814226186139e-05,
      "loss": 0.2073,
      "step": 5669
    },
    {
      "epoch": 0.22925995936397545,
      "grad_norm": 3.810642957687378,
      "learning_rate": 8.790398352832524e-05,
      "loss": 0.179,
      "step": 5670
    },
    {
      "epoch": 0.22930039321924248,
      "grad_norm": 2.0104305744171143,
      "learning_rate": 8.789982417816499e-05,
      "loss": 0.1141,
      "step": 5671
    },
    {
      "epoch": 0.22934082707450948,
      "grad_norm": 4.0161566734313965,
      "learning_rate": 8.789566421144826e-05,
      "loss": 0.1432,
      "step": 5672
    },
    {
      "epoch": 0.2293812609297765,
      "grad_norm": 3.213958978652954,
      "learning_rate": 8.789150362824276e-05,
      "loss": 0.2507,
      "step": 5673
    },
    {
      "epoch": 0.2294216947850435,
      "grad_norm": 10.71965503692627,
      "learning_rate": 8.788734242861616e-05,
      "loss": 0.3903,
      "step": 5674
    },
    {
      "epoch": 0.22946212864031054,
      "grad_norm": 2.893892288208008,
      "learning_rate": 8.788318061263615e-05,
      "loss": 0.1572,
      "step": 5675
    },
    {
      "epoch": 0.22950256249557754,
      "grad_norm": 4.652428150177002,
      "learning_rate": 8.787901818037043e-05,
      "loss": 0.1766,
      "step": 5676
    },
    {
      "epoch": 0.22954299635084457,
      "grad_norm": 3.6191799640655518,
      "learning_rate": 8.787485513188673e-05,
      "loss": 0.0937,
      "step": 5677
    },
    {
      "epoch": 0.22958343020611158,
      "grad_norm": 5.515514850616455,
      "learning_rate": 8.787069146725277e-05,
      "loss": 0.136,
      "step": 5678
    },
    {
      "epoch": 0.22962386406137858,
      "grad_norm": 3.7299797534942627,
      "learning_rate": 8.786652718653629e-05,
      "loss": 0.119,
      "step": 5679
    },
    {
      "epoch": 0.2296642979166456,
      "grad_norm": 5.593990325927734,
      "learning_rate": 8.786236228980503e-05,
      "loss": 0.1989,
      "step": 5680
    },
    {
      "epoch": 0.2297047317719126,
      "grad_norm": 2.976088523864746,
      "learning_rate": 8.785819677712675e-05,
      "loss": 0.1154,
      "step": 5681
    },
    {
      "epoch": 0.22974516562717964,
      "grad_norm": 2.38149094581604,
      "learning_rate": 8.785403064856919e-05,
      "loss": 0.1306,
      "step": 5682
    },
    {
      "epoch": 0.22978559948244665,
      "grad_norm": 7.248859405517578,
      "learning_rate": 8.784986390420016e-05,
      "loss": 0.1693,
      "step": 5683
    },
    {
      "epoch": 0.22982603333771368,
      "grad_norm": 3.3389604091644287,
      "learning_rate": 8.78456965440874e-05,
      "loss": 0.1077,
      "step": 5684
    },
    {
      "epoch": 0.22986646719298068,
      "grad_norm": 7.830263137817383,
      "learning_rate": 8.784152856829876e-05,
      "loss": 0.1742,
      "step": 5685
    },
    {
      "epoch": 0.2299069010482477,
      "grad_norm": 6.438128471374512,
      "learning_rate": 8.783735997690201e-05,
      "loss": 0.4185,
      "step": 5686
    },
    {
      "epoch": 0.2299473349035147,
      "grad_norm": 6.003082752227783,
      "learning_rate": 8.783319076996497e-05,
      "loss": 0.115,
      "step": 5687
    },
    {
      "epoch": 0.22998776875878174,
      "grad_norm": 2.30472469329834,
      "learning_rate": 8.782902094755546e-05,
      "loss": 0.1082,
      "step": 5688
    },
    {
      "epoch": 0.23002820261404874,
      "grad_norm": 3.564647912979126,
      "learning_rate": 8.782485050974133e-05,
      "loss": 0.1205,
      "step": 5689
    },
    {
      "epoch": 0.23006863646931575,
      "grad_norm": 3.6493117809295654,
      "learning_rate": 8.782067945659042e-05,
      "loss": 0.157,
      "step": 5690
    },
    {
      "epoch": 0.23010907032458278,
      "grad_norm": 5.7299346923828125,
      "learning_rate": 8.781650778817058e-05,
      "loss": 0.1704,
      "step": 5691
    },
    {
      "epoch": 0.23014950417984978,
      "grad_norm": 3.4585371017456055,
      "learning_rate": 8.781233550454967e-05,
      "loss": 0.2401,
      "step": 5692
    },
    {
      "epoch": 0.2301899380351168,
      "grad_norm": 7.273004531860352,
      "learning_rate": 8.780816260579558e-05,
      "loss": 0.1771,
      "step": 5693
    },
    {
      "epoch": 0.2302303718903838,
      "grad_norm": 5.131725788116455,
      "learning_rate": 8.780398909197618e-05,
      "loss": 0.1751,
      "step": 5694
    },
    {
      "epoch": 0.23027080574565084,
      "grad_norm": 3.7104175090789795,
      "learning_rate": 8.779981496315938e-05,
      "loss": 0.1078,
      "step": 5695
    },
    {
      "epoch": 0.23031123960091784,
      "grad_norm": 7.176881790161133,
      "learning_rate": 8.779564021941307e-05,
      "loss": 0.3424,
      "step": 5696
    },
    {
      "epoch": 0.23035167345618487,
      "grad_norm": 8.062840461730957,
      "learning_rate": 8.779146486080518e-05,
      "loss": 0.2316,
      "step": 5697
    },
    {
      "epoch": 0.23039210731145188,
      "grad_norm": 4.456916332244873,
      "learning_rate": 8.778728888740361e-05,
      "loss": 0.2626,
      "step": 5698
    },
    {
      "epoch": 0.2304325411667189,
      "grad_norm": 4.188305854797363,
      "learning_rate": 8.778311229927633e-05,
      "loss": 0.0696,
      "step": 5699
    },
    {
      "epoch": 0.2304729750219859,
      "grad_norm": 6.49384069442749,
      "learning_rate": 8.777893509649127e-05,
      "loss": 0.1889,
      "step": 5700
    },
    {
      "epoch": 0.2305134088772529,
      "grad_norm": 2.323225736618042,
      "learning_rate": 8.777475727911636e-05,
      "loss": 0.2635,
      "step": 5701
    },
    {
      "epoch": 0.23055384273251994,
      "grad_norm": 3.604560613632202,
      "learning_rate": 8.77705788472196e-05,
      "loss": 0.1108,
      "step": 5702
    },
    {
      "epoch": 0.23059427658778694,
      "grad_norm": 7.651488304138184,
      "learning_rate": 8.776639980086894e-05,
      "loss": 0.208,
      "step": 5703
    },
    {
      "epoch": 0.23063471044305398,
      "grad_norm": 1.8665341138839722,
      "learning_rate": 8.776222014013239e-05,
      "loss": 0.0529,
      "step": 5704
    },
    {
      "epoch": 0.23067514429832098,
      "grad_norm": 3.5386860370635986,
      "learning_rate": 8.775803986507792e-05,
      "loss": 0.1267,
      "step": 5705
    },
    {
      "epoch": 0.230715578153588,
      "grad_norm": 4.705434799194336,
      "learning_rate": 8.775385897577356e-05,
      "loss": 0.1693,
      "step": 5706
    },
    {
      "epoch": 0.230756012008855,
      "grad_norm": 2.4155731201171875,
      "learning_rate": 8.774967747228731e-05,
      "loss": 0.1329,
      "step": 5707
    },
    {
      "epoch": 0.23079644586412204,
      "grad_norm": 7.418895244598389,
      "learning_rate": 8.77454953546872e-05,
      "loss": 0.1512,
      "step": 5708
    },
    {
      "epoch": 0.23083687971938904,
      "grad_norm": 4.128955841064453,
      "learning_rate": 8.774131262304124e-05,
      "loss": 0.2248,
      "step": 5709
    },
    {
      "epoch": 0.23087731357465605,
      "grad_norm": 2.5535778999328613,
      "learning_rate": 8.773712927741751e-05,
      "loss": 0.1542,
      "step": 5710
    },
    {
      "epoch": 0.23091774742992308,
      "grad_norm": 3.577630043029785,
      "learning_rate": 8.773294531788404e-05,
      "loss": 0.2057,
      "step": 5711
    },
    {
      "epoch": 0.23095818128519008,
      "grad_norm": 3.247633695602417,
      "learning_rate": 8.772876074450892e-05,
      "loss": 0.1345,
      "step": 5712
    },
    {
      "epoch": 0.2309986151404571,
      "grad_norm": 2.9401376247406006,
      "learning_rate": 8.77245755573602e-05,
      "loss": 0.1956,
      "step": 5713
    },
    {
      "epoch": 0.2310390489957241,
      "grad_norm": 2.4084811210632324,
      "learning_rate": 8.772038975650598e-05,
      "loss": 0.1949,
      "step": 5714
    },
    {
      "epoch": 0.23107948285099114,
      "grad_norm": 2.4987294673919678,
      "learning_rate": 8.771620334201435e-05,
      "loss": 0.2415,
      "step": 5715
    },
    {
      "epoch": 0.23111991670625814,
      "grad_norm": 8.282980918884277,
      "learning_rate": 8.771201631395343e-05,
      "loss": 0.2865,
      "step": 5716
    },
    {
      "epoch": 0.23116035056152517,
      "grad_norm": 2.6672399044036865,
      "learning_rate": 8.77078286723913e-05,
      "loss": 0.0877,
      "step": 5717
    },
    {
      "epoch": 0.23120078441679218,
      "grad_norm": 7.097591400146484,
      "learning_rate": 8.770364041739611e-05,
      "loss": 0.1793,
      "step": 5718
    },
    {
      "epoch": 0.2312412182720592,
      "grad_norm": 1.9232172966003418,
      "learning_rate": 8.7699451549036e-05,
      "loss": 0.1131,
      "step": 5719
    },
    {
      "epoch": 0.2312816521273262,
      "grad_norm": 5.013349533081055,
      "learning_rate": 8.769526206737909e-05,
      "loss": 0.0894,
      "step": 5720
    },
    {
      "epoch": 0.2313220859825932,
      "grad_norm": 2.935312032699585,
      "learning_rate": 8.769107197249356e-05,
      "loss": 0.1264,
      "step": 5721
    },
    {
      "epoch": 0.23136251983786024,
      "grad_norm": 3.1348071098327637,
      "learning_rate": 8.768688126444755e-05,
      "loss": 0.1355,
      "step": 5722
    },
    {
      "epoch": 0.23140295369312724,
      "grad_norm": 2.349069595336914,
      "learning_rate": 8.768268994330926e-05,
      "loss": 0.1006,
      "step": 5723
    },
    {
      "epoch": 0.23144338754839427,
      "grad_norm": 5.177303791046143,
      "learning_rate": 8.767849800914686e-05,
      "loss": 0.2295,
      "step": 5724
    },
    {
      "epoch": 0.23148382140366128,
      "grad_norm": 6.543885231018066,
      "learning_rate": 8.767430546202854e-05,
      "loss": 0.2478,
      "step": 5725
    },
    {
      "epoch": 0.2315242552589283,
      "grad_norm": 6.990058898925781,
      "learning_rate": 8.767011230202253e-05,
      "loss": 0.1867,
      "step": 5726
    },
    {
      "epoch": 0.2315646891141953,
      "grad_norm": 4.587096691131592,
      "learning_rate": 8.7665918529197e-05,
      "loss": 0.2872,
      "step": 5727
    },
    {
      "epoch": 0.23160512296946234,
      "grad_norm": 6.3970046043396,
      "learning_rate": 8.766172414362024e-05,
      "loss": 0.1566,
      "step": 5728
    },
    {
      "epoch": 0.23164555682472934,
      "grad_norm": 3.2849466800689697,
      "learning_rate": 8.765752914536043e-05,
      "loss": 0.1813,
      "step": 5729
    },
    {
      "epoch": 0.23168599067999637,
      "grad_norm": 5.731575012207031,
      "learning_rate": 8.765333353448582e-05,
      "loss": 0.1412,
      "step": 5730
    },
    {
      "epoch": 0.23172642453526338,
      "grad_norm": 3.79323148727417,
      "learning_rate": 8.764913731106467e-05,
      "loss": 0.3537,
      "step": 5731
    },
    {
      "epoch": 0.23176685839053038,
      "grad_norm": 7.754724025726318,
      "learning_rate": 8.764494047516528e-05,
      "loss": 0.2335,
      "step": 5732
    },
    {
      "epoch": 0.2318072922457974,
      "grad_norm": 4.933526992797852,
      "learning_rate": 8.764074302685586e-05,
      "loss": 0.3179,
      "step": 5733
    },
    {
      "epoch": 0.2318477261010644,
      "grad_norm": 8.600529670715332,
      "learning_rate": 8.763654496620473e-05,
      "loss": 0.3224,
      "step": 5734
    },
    {
      "epoch": 0.23188815995633144,
      "grad_norm": 2.0110955238342285,
      "learning_rate": 8.76323462932802e-05,
      "loss": 0.0532,
      "step": 5735
    },
    {
      "epoch": 0.23192859381159844,
      "grad_norm": 6.674431800842285,
      "learning_rate": 8.762814700815055e-05,
      "loss": 0.1841,
      "step": 5736
    },
    {
      "epoch": 0.23196902766686547,
      "grad_norm": 3.6567726135253906,
      "learning_rate": 8.76239471108841e-05,
      "loss": 0.1751,
      "step": 5737
    },
    {
      "epoch": 0.23200946152213248,
      "grad_norm": 5.264025688171387,
      "learning_rate": 8.761974660154917e-05,
      "loss": 0.2108,
      "step": 5738
    },
    {
      "epoch": 0.2320498953773995,
      "grad_norm": 2.649386167526245,
      "learning_rate": 8.76155454802141e-05,
      "loss": 0.1209,
      "step": 5739
    },
    {
      "epoch": 0.2320903292326665,
      "grad_norm": 2.847513198852539,
      "learning_rate": 8.761134374694723e-05,
      "loss": 0.2378,
      "step": 5740
    },
    {
      "epoch": 0.23213076308793354,
      "grad_norm": 3.577235698699951,
      "learning_rate": 8.760714140181694e-05,
      "loss": 0.0858,
      "step": 5741
    },
    {
      "epoch": 0.23217119694320054,
      "grad_norm": 4.185028553009033,
      "learning_rate": 8.760293844489156e-05,
      "loss": 0.1261,
      "step": 5742
    },
    {
      "epoch": 0.23221163079846754,
      "grad_norm": 2.3881616592407227,
      "learning_rate": 8.759873487623946e-05,
      "loss": 0.1008,
      "step": 5743
    },
    {
      "epoch": 0.23225206465373457,
      "grad_norm": 4.907158851623535,
      "learning_rate": 8.759453069592904e-05,
      "loss": 0.2529,
      "step": 5744
    },
    {
      "epoch": 0.23229249850900158,
      "grad_norm": 8.864971160888672,
      "learning_rate": 8.759032590402872e-05,
      "loss": 0.2571,
      "step": 5745
    },
    {
      "epoch": 0.2323329323642686,
      "grad_norm": 8.148448944091797,
      "learning_rate": 8.758612050060685e-05,
      "loss": 0.3103,
      "step": 5746
    },
    {
      "epoch": 0.2323733662195356,
      "grad_norm": 3.8654823303222656,
      "learning_rate": 8.758191448573188e-05,
      "loss": 0.0603,
      "step": 5747
    },
    {
      "epoch": 0.23241380007480264,
      "grad_norm": 2.587003469467163,
      "learning_rate": 8.757770785947221e-05,
      "loss": 0.078,
      "step": 5748
    },
    {
      "epoch": 0.23245423393006964,
      "grad_norm": 3.87430739402771,
      "learning_rate": 8.757350062189631e-05,
      "loss": 0.0999,
      "step": 5749
    },
    {
      "epoch": 0.23249466778533667,
      "grad_norm": 13.096192359924316,
      "learning_rate": 8.756929277307259e-05,
      "loss": 0.3003,
      "step": 5750
    },
    {
      "epoch": 0.23253510164060368,
      "grad_norm": 2.8238162994384766,
      "learning_rate": 8.756508431306951e-05,
      "loss": 0.1917,
      "step": 5751
    },
    {
      "epoch": 0.2325755354958707,
      "grad_norm": 6.673470497131348,
      "learning_rate": 8.756087524195555e-05,
      "loss": 0.1887,
      "step": 5752
    },
    {
      "epoch": 0.2326159693511377,
      "grad_norm": 4.917054176330566,
      "learning_rate": 8.755666555979915e-05,
      "loss": 0.1403,
      "step": 5753
    },
    {
      "epoch": 0.2326564032064047,
      "grad_norm": 6.934584140777588,
      "learning_rate": 8.755245526666884e-05,
      "loss": 0.2271,
      "step": 5754
    },
    {
      "epoch": 0.23269683706167174,
      "grad_norm": 11.554547309875488,
      "learning_rate": 8.754824436263307e-05,
      "loss": 0.2719,
      "step": 5755
    },
    {
      "epoch": 0.23273727091693874,
      "grad_norm": 5.0187225341796875,
      "learning_rate": 8.754403284776037e-05,
      "loss": 0.1601,
      "step": 5756
    },
    {
      "epoch": 0.23277770477220577,
      "grad_norm": 4.799263954162598,
      "learning_rate": 8.753982072211924e-05,
      "loss": 0.3102,
      "step": 5757
    },
    {
      "epoch": 0.23281813862747278,
      "grad_norm": 3.329210042953491,
      "learning_rate": 8.753560798577823e-05,
      "loss": 0.1409,
      "step": 5758
    },
    {
      "epoch": 0.2328585724827398,
      "grad_norm": 6.168715953826904,
      "learning_rate": 8.753139463880582e-05,
      "loss": 0.2019,
      "step": 5759
    },
    {
      "epoch": 0.2328990063380068,
      "grad_norm": 2.401323080062866,
      "learning_rate": 8.75271806812706e-05,
      "loss": 0.1743,
      "step": 5760
    },
    {
      "epoch": 0.23293944019327384,
      "grad_norm": 5.326553821563721,
      "learning_rate": 8.75229661132411e-05,
      "loss": 0.2354,
      "step": 5761
    },
    {
      "epoch": 0.23297987404854084,
      "grad_norm": 7.545034408569336,
      "learning_rate": 8.751875093478588e-05,
      "loss": 0.2768,
      "step": 5762
    },
    {
      "epoch": 0.23302030790380787,
      "grad_norm": 6.828022003173828,
      "learning_rate": 8.751453514597354e-05,
      "loss": 0.2114,
      "step": 5763
    },
    {
      "epoch": 0.23306074175907487,
      "grad_norm": 2.339292049407959,
      "learning_rate": 8.751031874687263e-05,
      "loss": 0.1785,
      "step": 5764
    },
    {
      "epoch": 0.23310117561434188,
      "grad_norm": 4.959456920623779,
      "learning_rate": 8.750610173755177e-05,
      "loss": 0.2792,
      "step": 5765
    },
    {
      "epoch": 0.2331416094696089,
      "grad_norm": 2.2757108211517334,
      "learning_rate": 8.750188411807954e-05,
      "loss": 0.26,
      "step": 5766
    },
    {
      "epoch": 0.2331820433248759,
      "grad_norm": 1.756913661956787,
      "learning_rate": 8.749766588852457e-05,
      "loss": 0.1486,
      "step": 5767
    },
    {
      "epoch": 0.23322247718014294,
      "grad_norm": 2.4095423221588135,
      "learning_rate": 8.749344704895548e-05,
      "loss": 0.1522,
      "step": 5768
    },
    {
      "epoch": 0.23326291103540994,
      "grad_norm": 3.765080690383911,
      "learning_rate": 8.74892275994409e-05,
      "loss": 0.2167,
      "step": 5769
    },
    {
      "epoch": 0.23330334489067697,
      "grad_norm": 6.353774547576904,
      "learning_rate": 8.748500754004944e-05,
      "loss": 0.2431,
      "step": 5770
    },
    {
      "epoch": 0.23334377874594397,
      "grad_norm": 5.1684417724609375,
      "learning_rate": 8.74807868708498e-05,
      "loss": 0.137,
      "step": 5771
    },
    {
      "epoch": 0.233384212601211,
      "grad_norm": 2.4692533016204834,
      "learning_rate": 8.747656559191063e-05,
      "loss": 0.1377,
      "step": 5772
    },
    {
      "epoch": 0.233424646456478,
      "grad_norm": 3.280935525894165,
      "learning_rate": 8.747234370330058e-05,
      "loss": 0.2056,
      "step": 5773
    },
    {
      "epoch": 0.23346508031174504,
      "grad_norm": 3.835749864578247,
      "learning_rate": 8.746812120508836e-05,
      "loss": 0.11,
      "step": 5774
    },
    {
      "epoch": 0.23350551416701204,
      "grad_norm": 4.708775043487549,
      "learning_rate": 8.746389809734266e-05,
      "loss": 0.1599,
      "step": 5775
    },
    {
      "epoch": 0.23354594802227904,
      "grad_norm": 4.239015579223633,
      "learning_rate": 8.745967438013215e-05,
      "loss": 0.1408,
      "step": 5776
    },
    {
      "epoch": 0.23358638187754607,
      "grad_norm": 4.3496413230896,
      "learning_rate": 8.745545005352557e-05,
      "loss": 0.1207,
      "step": 5777
    },
    {
      "epoch": 0.23362681573281308,
      "grad_norm": 3.9284868240356445,
      "learning_rate": 8.745122511759163e-05,
      "loss": 0.1584,
      "step": 5778
    },
    {
      "epoch": 0.2336672495880801,
      "grad_norm": 5.885026931762695,
      "learning_rate": 8.744699957239908e-05,
      "loss": 0.198,
      "step": 5779
    },
    {
      "epoch": 0.2337076834433471,
      "grad_norm": 2.9485673904418945,
      "learning_rate": 8.744277341801664e-05,
      "loss": 0.2196,
      "step": 5780
    },
    {
      "epoch": 0.23374811729861414,
      "grad_norm": 7.390035629272461,
      "learning_rate": 8.743854665451305e-05,
      "loss": 0.2785,
      "step": 5781
    },
    {
      "epoch": 0.23378855115388114,
      "grad_norm": 3.329568386077881,
      "learning_rate": 8.74343192819571e-05,
      "loss": 0.2456,
      "step": 5782
    },
    {
      "epoch": 0.23382898500914817,
      "grad_norm": 3.300050735473633,
      "learning_rate": 8.743009130041757e-05,
      "loss": 0.11,
      "step": 5783
    },
    {
      "epoch": 0.23386941886441517,
      "grad_norm": 6.936415672302246,
      "learning_rate": 8.74258627099632e-05,
      "loss": 0.2326,
      "step": 5784
    },
    {
      "epoch": 0.23390985271968218,
      "grad_norm": 3.382483720779419,
      "learning_rate": 8.742163351066281e-05,
      "loss": 0.1037,
      "step": 5785
    },
    {
      "epoch": 0.2339502865749492,
      "grad_norm": 3.018651247024536,
      "learning_rate": 8.74174037025852e-05,
      "loss": 0.0982,
      "step": 5786
    },
    {
      "epoch": 0.2339907204302162,
      "grad_norm": 3.692439079284668,
      "learning_rate": 8.741317328579917e-05,
      "loss": 0.1302,
      "step": 5787
    },
    {
      "epoch": 0.23403115428548324,
      "grad_norm": 2.0556282997131348,
      "learning_rate": 8.740894226037355e-05,
      "loss": 0.0681,
      "step": 5788
    },
    {
      "epoch": 0.23407158814075024,
      "grad_norm": 4.005192756652832,
      "learning_rate": 8.740471062637717e-05,
      "loss": 0.1894,
      "step": 5789
    },
    {
      "epoch": 0.23411202199601727,
      "grad_norm": 4.588472843170166,
      "learning_rate": 8.740047838387885e-05,
      "loss": 0.2187,
      "step": 5790
    },
    {
      "epoch": 0.23415245585128427,
      "grad_norm": 4.5771803855896,
      "learning_rate": 8.739624553294746e-05,
      "loss": 0.2587,
      "step": 5791
    },
    {
      "epoch": 0.2341928897065513,
      "grad_norm": 6.830857276916504,
      "learning_rate": 8.739201207365186e-05,
      "loss": 0.2328,
      "step": 5792
    },
    {
      "epoch": 0.2342333235618183,
      "grad_norm": 4.287374019622803,
      "learning_rate": 8.738777800606092e-05,
      "loss": 0.1453,
      "step": 5793
    },
    {
      "epoch": 0.23427375741708534,
      "grad_norm": 10.03058910369873,
      "learning_rate": 8.738354333024351e-05,
      "loss": 0.2978,
      "step": 5794
    },
    {
      "epoch": 0.23431419127235234,
      "grad_norm": 6.294463634490967,
      "learning_rate": 8.737930804626852e-05,
      "loss": 0.173,
      "step": 5795
    },
    {
      "epoch": 0.23435462512761934,
      "grad_norm": 5.071019172668457,
      "learning_rate": 8.737507215420489e-05,
      "loss": 0.1651,
      "step": 5796
    },
    {
      "epoch": 0.23439505898288637,
      "grad_norm": 3.06192684173584,
      "learning_rate": 8.737083565412146e-05,
      "loss": 0.1349,
      "step": 5797
    },
    {
      "epoch": 0.23443549283815338,
      "grad_norm": 4.775763034820557,
      "learning_rate": 8.736659854608721e-05,
      "loss": 0.115,
      "step": 5798
    },
    {
      "epoch": 0.2344759266934204,
      "grad_norm": 2.225261688232422,
      "learning_rate": 8.736236083017105e-05,
      "loss": 0.1293,
      "step": 5799
    },
    {
      "epoch": 0.2345163605486874,
      "grad_norm": 4.4199347496032715,
      "learning_rate": 8.735812250644188e-05,
      "loss": 0.1832,
      "step": 5800
    },
    {
      "epoch": 0.23455679440395444,
      "grad_norm": 5.317866802215576,
      "learning_rate": 8.735388357496871e-05,
      "loss": 0.2008,
      "step": 5801
    },
    {
      "epoch": 0.23459722825922144,
      "grad_norm": 2.7359652519226074,
      "learning_rate": 8.734964403582046e-05,
      "loss": 0.2234,
      "step": 5802
    },
    {
      "epoch": 0.23463766211448847,
      "grad_norm": 8.821404457092285,
      "learning_rate": 8.734540388906614e-05,
      "loss": 0.3094,
      "step": 5803
    },
    {
      "epoch": 0.23467809596975547,
      "grad_norm": 6.769132614135742,
      "learning_rate": 8.734116313477467e-05,
      "loss": 0.1578,
      "step": 5804
    },
    {
      "epoch": 0.2347185298250225,
      "grad_norm": 6.217795372009277,
      "learning_rate": 8.733692177301509e-05,
      "loss": 0.2463,
      "step": 5805
    },
    {
      "epoch": 0.2347589636802895,
      "grad_norm": 2.9610724449157715,
      "learning_rate": 8.733267980385638e-05,
      "loss": 0.1591,
      "step": 5806
    },
    {
      "epoch": 0.2347993975355565,
      "grad_norm": 3.7270467281341553,
      "learning_rate": 8.732843722736753e-05,
      "loss": 0.2956,
      "step": 5807
    },
    {
      "epoch": 0.23483983139082354,
      "grad_norm": 4.4273271560668945,
      "learning_rate": 8.732419404361758e-05,
      "loss": 0.1765,
      "step": 5808
    },
    {
      "epoch": 0.23488026524609054,
      "grad_norm": 3.3255577087402344,
      "learning_rate": 8.731995025267556e-05,
      "loss": 0.1365,
      "step": 5809
    },
    {
      "epoch": 0.23492069910135757,
      "grad_norm": 8.559576988220215,
      "learning_rate": 8.73157058546105e-05,
      "loss": 0.1813,
      "step": 5810
    },
    {
      "epoch": 0.23496113295662457,
      "grad_norm": 4.305734157562256,
      "learning_rate": 8.731146084949147e-05,
      "loss": 0.1383,
      "step": 5811
    },
    {
      "epoch": 0.2350015668118916,
      "grad_norm": 3.887387990951538,
      "learning_rate": 8.730721523738749e-05,
      "loss": 0.1957,
      "step": 5812
    },
    {
      "epoch": 0.2350420006671586,
      "grad_norm": 2.232271194458008,
      "learning_rate": 8.730296901836765e-05,
      "loss": 0.1972,
      "step": 5813
    },
    {
      "epoch": 0.23508243452242564,
      "grad_norm": 4.348437786102295,
      "learning_rate": 8.729872219250102e-05,
      "loss": 0.1786,
      "step": 5814
    },
    {
      "epoch": 0.23512286837769264,
      "grad_norm": 2.5670852661132812,
      "learning_rate": 8.729447475985671e-05,
      "loss": 0.2047,
      "step": 5815
    },
    {
      "epoch": 0.23516330223295967,
      "grad_norm": 1.968658447265625,
      "learning_rate": 8.729022672050378e-05,
      "loss": 0.1085,
      "step": 5816
    },
    {
      "epoch": 0.23520373608822667,
      "grad_norm": 1.6469684839248657,
      "learning_rate": 8.728597807451136e-05,
      "loss": 0.0815,
      "step": 5817
    },
    {
      "epoch": 0.23524416994349368,
      "grad_norm": 1.3831133842468262,
      "learning_rate": 8.728172882194856e-05,
      "loss": 0.0523,
      "step": 5818
    },
    {
      "epoch": 0.2352846037987607,
      "grad_norm": 3.7314982414245605,
      "learning_rate": 8.727747896288452e-05,
      "loss": 0.1326,
      "step": 5819
    },
    {
      "epoch": 0.2353250376540277,
      "grad_norm": 3.7328240871429443,
      "learning_rate": 8.727322849738837e-05,
      "loss": 0.1417,
      "step": 5820
    },
    {
      "epoch": 0.23536547150929474,
      "grad_norm": 2.3268916606903076,
      "learning_rate": 8.726897742552925e-05,
      "loss": 0.12,
      "step": 5821
    },
    {
      "epoch": 0.23540590536456174,
      "grad_norm": 4.0223894119262695,
      "learning_rate": 8.726472574737632e-05,
      "loss": 0.1849,
      "step": 5822
    },
    {
      "epoch": 0.23544633921982877,
      "grad_norm": 7.2805094718933105,
      "learning_rate": 8.726047346299875e-05,
      "loss": 0.1529,
      "step": 5823
    },
    {
      "epoch": 0.23548677307509577,
      "grad_norm": 5.168961048126221,
      "learning_rate": 8.725622057246571e-05,
      "loss": 0.1119,
      "step": 5824
    },
    {
      "epoch": 0.2355272069303628,
      "grad_norm": 5.309213161468506,
      "learning_rate": 8.72519670758464e-05,
      "loss": 0.267,
      "step": 5825
    },
    {
      "epoch": 0.2355676407856298,
      "grad_norm": 6.623654365539551,
      "learning_rate": 8.724771297321e-05,
      "loss": 0.2157,
      "step": 5826
    },
    {
      "epoch": 0.23560807464089684,
      "grad_norm": 5.449578285217285,
      "learning_rate": 8.724345826462573e-05,
      "loss": 0.2572,
      "step": 5827
    },
    {
      "epoch": 0.23564850849616384,
      "grad_norm": 6.319265842437744,
      "learning_rate": 8.723920295016279e-05,
      "loss": 0.1908,
      "step": 5828
    },
    {
      "epoch": 0.23568894235143084,
      "grad_norm": 3.596027374267578,
      "learning_rate": 8.72349470298904e-05,
      "loss": 0.1538,
      "step": 5829
    },
    {
      "epoch": 0.23572937620669787,
      "grad_norm": 6.945806980133057,
      "learning_rate": 8.723069050387785e-05,
      "loss": 0.1906,
      "step": 5830
    },
    {
      "epoch": 0.23576981006196487,
      "grad_norm": 3.5267140865325928,
      "learning_rate": 8.722643337219431e-05,
      "loss": 0.151,
      "step": 5831
    },
    {
      "epoch": 0.2358102439172319,
      "grad_norm": 4.481578350067139,
      "learning_rate": 8.722217563490908e-05,
      "loss": 0.1609,
      "step": 5832
    },
    {
      "epoch": 0.2358506777724989,
      "grad_norm": 4.303544044494629,
      "learning_rate": 8.721791729209142e-05,
      "loss": 0.1594,
      "step": 5833
    },
    {
      "epoch": 0.23589111162776594,
      "grad_norm": 8.424549102783203,
      "learning_rate": 8.72136583438106e-05,
      "loss": 0.2806,
      "step": 5834
    },
    {
      "epoch": 0.23593154548303294,
      "grad_norm": 3.254166603088379,
      "learning_rate": 8.72093987901359e-05,
      "loss": 0.1041,
      "step": 5835
    },
    {
      "epoch": 0.23597197933829997,
      "grad_norm": 2.9343652725219727,
      "learning_rate": 8.720513863113663e-05,
      "loss": 0.1437,
      "step": 5836
    },
    {
      "epoch": 0.23601241319356697,
      "grad_norm": 4.241241931915283,
      "learning_rate": 8.720087786688207e-05,
      "loss": 0.1908,
      "step": 5837
    },
    {
      "epoch": 0.236052847048834,
      "grad_norm": 4.533597469329834,
      "learning_rate": 8.719661649744157e-05,
      "loss": 0.1449,
      "step": 5838
    },
    {
      "epoch": 0.236093280904101,
      "grad_norm": 6.601241111755371,
      "learning_rate": 8.719235452288441e-05,
      "loss": 0.2474,
      "step": 5839
    },
    {
      "epoch": 0.236133714759368,
      "grad_norm": 5.3936614990234375,
      "learning_rate": 8.718809194327997e-05,
      "loss": 0.1519,
      "step": 5840
    },
    {
      "epoch": 0.23617414861463504,
      "grad_norm": 6.60890531539917,
      "learning_rate": 8.718382875869756e-05,
      "loss": 0.3376,
      "step": 5841
    },
    {
      "epoch": 0.23621458246990204,
      "grad_norm": 2.936448812484741,
      "learning_rate": 8.717956496920654e-05,
      "loss": 0.1402,
      "step": 5842
    },
    {
      "epoch": 0.23625501632516907,
      "grad_norm": 5.989627361297607,
      "learning_rate": 8.717530057487629e-05,
      "loss": 0.2525,
      "step": 5843
    },
    {
      "epoch": 0.23629545018043607,
      "grad_norm": 4.293125152587891,
      "learning_rate": 8.717103557577615e-05,
      "loss": 0.1998,
      "step": 5844
    },
    {
      "epoch": 0.2363358840357031,
      "grad_norm": 6.075552940368652,
      "learning_rate": 8.716676997197555e-05,
      "loss": 0.2591,
      "step": 5845
    },
    {
      "epoch": 0.2363763178909701,
      "grad_norm": 5.427404880523682,
      "learning_rate": 8.716250376354385e-05,
      "loss": 0.0835,
      "step": 5846
    },
    {
      "epoch": 0.23641675174623714,
      "grad_norm": 6.130009174346924,
      "learning_rate": 8.715823695055046e-05,
      "loss": 0.2353,
      "step": 5847
    },
    {
      "epoch": 0.23645718560150414,
      "grad_norm": 4.348407745361328,
      "learning_rate": 8.715396953306478e-05,
      "loss": 0.267,
      "step": 5848
    },
    {
      "epoch": 0.23649761945677114,
      "grad_norm": 7.366415500640869,
      "learning_rate": 8.714970151115625e-05,
      "loss": 0.1881,
      "step": 5849
    },
    {
      "epoch": 0.23653805331203817,
      "grad_norm": 3.2398664951324463,
      "learning_rate": 8.714543288489432e-05,
      "loss": 0.1139,
      "step": 5850
    },
    {
      "epoch": 0.23657848716730517,
      "grad_norm": 5.649832248687744,
      "learning_rate": 8.714116365434839e-05,
      "loss": 0.2578,
      "step": 5851
    },
    {
      "epoch": 0.2366189210225722,
      "grad_norm": 2.6740245819091797,
      "learning_rate": 8.713689381958795e-05,
      "loss": 0.1679,
      "step": 5852
    },
    {
      "epoch": 0.2366593548778392,
      "grad_norm": 3.967548131942749,
      "learning_rate": 8.713262338068243e-05,
      "loss": 0.1872,
      "step": 5853
    },
    {
      "epoch": 0.23669978873310624,
      "grad_norm": 4.226831912994385,
      "learning_rate": 8.712835233770133e-05,
      "loss": 0.1886,
      "step": 5854
    },
    {
      "epoch": 0.23674022258837324,
      "grad_norm": 5.640081405639648,
      "learning_rate": 8.712408069071411e-05,
      "loss": 0.2116,
      "step": 5855
    },
    {
      "epoch": 0.23678065644364027,
      "grad_norm": 2.474032402038574,
      "learning_rate": 8.711980843979027e-05,
      "loss": 0.1502,
      "step": 5856
    },
    {
      "epoch": 0.23682109029890727,
      "grad_norm": 3.420659303665161,
      "learning_rate": 8.711553558499929e-05,
      "loss": 0.2218,
      "step": 5857
    },
    {
      "epoch": 0.2368615241541743,
      "grad_norm": 4.594886302947998,
      "learning_rate": 8.711126212641075e-05,
      "loss": 0.1995,
      "step": 5858
    },
    {
      "epoch": 0.2369019580094413,
      "grad_norm": 6.94998836517334,
      "learning_rate": 8.710698806409409e-05,
      "loss": 0.2613,
      "step": 5859
    },
    {
      "epoch": 0.2369423918647083,
      "grad_norm": 3.810699224472046,
      "learning_rate": 8.710271339811888e-05,
      "loss": 0.2414,
      "step": 5860
    },
    {
      "epoch": 0.23698282571997534,
      "grad_norm": 7.499741554260254,
      "learning_rate": 8.709843812855465e-05,
      "loss": 0.1937,
      "step": 5861
    },
    {
      "epoch": 0.23702325957524234,
      "grad_norm": 4.931144714355469,
      "learning_rate": 8.709416225547096e-05,
      "loss": 0.128,
      "step": 5862
    },
    {
      "epoch": 0.23706369343050937,
      "grad_norm": 2.9475507736206055,
      "learning_rate": 8.708988577893736e-05,
      "loss": 0.2139,
      "step": 5863
    },
    {
      "epoch": 0.23710412728577637,
      "grad_norm": 3.591439723968506,
      "learning_rate": 8.708560869902342e-05,
      "loss": 0.2295,
      "step": 5864
    },
    {
      "epoch": 0.2371445611410434,
      "grad_norm": 4.529487609863281,
      "learning_rate": 8.708133101579874e-05,
      "loss": 0.1935,
      "step": 5865
    },
    {
      "epoch": 0.2371849949963104,
      "grad_norm": 6.102591037750244,
      "learning_rate": 8.707705272933288e-05,
      "loss": 0.1877,
      "step": 5866
    },
    {
      "epoch": 0.23722542885157744,
      "grad_norm": 4.529348373413086,
      "learning_rate": 8.707277383969546e-05,
      "loss": 0.1789,
      "step": 5867
    },
    {
      "epoch": 0.23726586270684444,
      "grad_norm": 5.818760871887207,
      "learning_rate": 8.706849434695607e-05,
      "loss": 0.2781,
      "step": 5868
    },
    {
      "epoch": 0.23730629656211147,
      "grad_norm": 4.015359401702881,
      "learning_rate": 8.706421425118436e-05,
      "loss": 0.1195,
      "step": 5869
    },
    {
      "epoch": 0.23734673041737847,
      "grad_norm": 2.948803663253784,
      "learning_rate": 8.705993355244994e-05,
      "loss": 0.1357,
      "step": 5870
    },
    {
      "epoch": 0.23738716427264547,
      "grad_norm": 5.522337913513184,
      "learning_rate": 8.705565225082244e-05,
      "loss": 0.1231,
      "step": 5871
    },
    {
      "epoch": 0.2374275981279125,
      "grad_norm": 2.600602865219116,
      "learning_rate": 8.705137034637152e-05,
      "loss": 0.1913,
      "step": 5872
    },
    {
      "epoch": 0.2374680319831795,
      "grad_norm": 5.414295673370361,
      "learning_rate": 8.704708783916683e-05,
      "loss": 0.1661,
      "step": 5873
    },
    {
      "epoch": 0.23750846583844654,
      "grad_norm": 3.5818450450897217,
      "learning_rate": 8.704280472927804e-05,
      "loss": 0.1705,
      "step": 5874
    },
    {
      "epoch": 0.23754889969371354,
      "grad_norm": 2.1138157844543457,
      "learning_rate": 8.703852101677484e-05,
      "loss": 0.1061,
      "step": 5875
    },
    {
      "epoch": 0.23758933354898057,
      "grad_norm": 3.862276792526245,
      "learning_rate": 8.703423670172691e-05,
      "loss": 0.0831,
      "step": 5876
    },
    {
      "epoch": 0.23762976740424757,
      "grad_norm": 3.2862191200256348,
      "learning_rate": 8.702995178420395e-05,
      "loss": 0.193,
      "step": 5877
    },
    {
      "epoch": 0.2376702012595146,
      "grad_norm": 2.788506269454956,
      "learning_rate": 8.702566626427566e-05,
      "loss": 0.1478,
      "step": 5878
    },
    {
      "epoch": 0.2377106351147816,
      "grad_norm": 4.104318141937256,
      "learning_rate": 8.702138014201176e-05,
      "loss": 0.1211,
      "step": 5879
    },
    {
      "epoch": 0.23775106897004863,
      "grad_norm": 4.658287048339844,
      "learning_rate": 8.701709341748197e-05,
      "loss": 0.1015,
      "step": 5880
    },
    {
      "epoch": 0.23779150282531564,
      "grad_norm": 6.327422618865967,
      "learning_rate": 8.701280609075605e-05,
      "loss": 0.2001,
      "step": 5881
    },
    {
      "epoch": 0.23783193668058264,
      "grad_norm": 3.2962646484375,
      "learning_rate": 8.700851816190372e-05,
      "loss": 0.1955,
      "step": 5882
    },
    {
      "epoch": 0.23787237053584967,
      "grad_norm": 2.978949785232544,
      "learning_rate": 8.700422963099474e-05,
      "loss": 0.2011,
      "step": 5883
    },
    {
      "epoch": 0.23791280439111667,
      "grad_norm": 9.541155815124512,
      "learning_rate": 8.699994049809891e-05,
      "loss": 0.1872,
      "step": 5884
    },
    {
      "epoch": 0.2379532382463837,
      "grad_norm": 4.963170051574707,
      "learning_rate": 8.699565076328597e-05,
      "loss": 0.1111,
      "step": 5885
    },
    {
      "epoch": 0.2379936721016507,
      "grad_norm": 4.113926410675049,
      "learning_rate": 8.69913604266257e-05,
      "loss": 0.1011,
      "step": 5886
    },
    {
      "epoch": 0.23803410595691774,
      "grad_norm": 6.18961763381958,
      "learning_rate": 8.698706948818791e-05,
      "loss": 0.2459,
      "step": 5887
    },
    {
      "epoch": 0.23807453981218474,
      "grad_norm": 3.1595184803009033,
      "learning_rate": 8.698277794804243e-05,
      "loss": 0.1329,
      "step": 5888
    },
    {
      "epoch": 0.23811497366745177,
      "grad_norm": 2.9568212032318115,
      "learning_rate": 8.697848580625902e-05,
      "loss": 0.1943,
      "step": 5889
    },
    {
      "epoch": 0.23815540752271877,
      "grad_norm": 4.943511486053467,
      "learning_rate": 8.697419306290756e-05,
      "loss": 0.1432,
      "step": 5890
    },
    {
      "epoch": 0.2381958413779858,
      "grad_norm": 3.9663479328155518,
      "learning_rate": 8.696989971805784e-05,
      "loss": 0.2461,
      "step": 5891
    },
    {
      "epoch": 0.2382362752332528,
      "grad_norm": 3.9418280124664307,
      "learning_rate": 8.696560577177975e-05,
      "loss": 0.1536,
      "step": 5892
    },
    {
      "epoch": 0.2382767090885198,
      "grad_norm": 5.7964277267456055,
      "learning_rate": 8.69613112241431e-05,
      "loss": 0.155,
      "step": 5893
    },
    {
      "epoch": 0.23831714294378684,
      "grad_norm": 10.561478614807129,
      "learning_rate": 8.695701607521778e-05,
      "loss": 0.2146,
      "step": 5894
    },
    {
      "epoch": 0.23835757679905384,
      "grad_norm": 3.3532555103302,
      "learning_rate": 8.695272032507364e-05,
      "loss": 0.1057,
      "step": 5895
    },
    {
      "epoch": 0.23839801065432087,
      "grad_norm": 4.04283332824707,
      "learning_rate": 8.69484239737806e-05,
      "loss": 0.0644,
      "step": 5896
    },
    {
      "epoch": 0.23843844450958787,
      "grad_norm": 2.1637003421783447,
      "learning_rate": 8.694412702140853e-05,
      "loss": 0.0403,
      "step": 5897
    },
    {
      "epoch": 0.2384788783648549,
      "grad_norm": 5.118740558624268,
      "learning_rate": 8.693982946802732e-05,
      "loss": 0.1471,
      "step": 5898
    },
    {
      "epoch": 0.2385193122201219,
      "grad_norm": 3.4040768146514893,
      "learning_rate": 8.69355313137069e-05,
      "loss": 0.2177,
      "step": 5899
    },
    {
      "epoch": 0.23855974607538893,
      "grad_norm": 7.387509346008301,
      "learning_rate": 8.69312325585172e-05,
      "loss": 0.2381,
      "step": 5900
    },
    {
      "epoch": 0.23860017993065594,
      "grad_norm": 5.387016296386719,
      "learning_rate": 8.692693320252817e-05,
      "loss": 0.1852,
      "step": 5901
    },
    {
      "epoch": 0.23864061378592297,
      "grad_norm": 2.965902090072632,
      "learning_rate": 8.692263324580969e-05,
      "loss": 0.094,
      "step": 5902
    },
    {
      "epoch": 0.23868104764118997,
      "grad_norm": 3.6524593830108643,
      "learning_rate": 8.691833268843176e-05,
      "loss": 0.2512,
      "step": 5903
    },
    {
      "epoch": 0.23872148149645697,
      "grad_norm": 5.109528541564941,
      "learning_rate": 8.691403153046433e-05,
      "loss": 0.2786,
      "step": 5904
    },
    {
      "epoch": 0.238761915351724,
      "grad_norm": 5.959028720855713,
      "learning_rate": 8.690972977197736e-05,
      "loss": 0.1838,
      "step": 5905
    },
    {
      "epoch": 0.238802349206991,
      "grad_norm": 7.3199334144592285,
      "learning_rate": 8.690542741304087e-05,
      "loss": 0.2338,
      "step": 5906
    },
    {
      "epoch": 0.23884278306225804,
      "grad_norm": 5.897810935974121,
      "learning_rate": 8.690112445372479e-05,
      "loss": 0.1687,
      "step": 5907
    },
    {
      "epoch": 0.23888321691752504,
      "grad_norm": 5.121602535247803,
      "learning_rate": 8.689682089409917e-05,
      "loss": 0.1559,
      "step": 5908
    },
    {
      "epoch": 0.23892365077279207,
      "grad_norm": 3.4499974250793457,
      "learning_rate": 8.689251673423401e-05,
      "loss": 0.2596,
      "step": 5909
    },
    {
      "epoch": 0.23896408462805907,
      "grad_norm": 4.022715091705322,
      "learning_rate": 8.688821197419931e-05,
      "loss": 0.1747,
      "step": 5910
    },
    {
      "epoch": 0.2390045184833261,
      "grad_norm": 2.1420822143554688,
      "learning_rate": 8.688390661406514e-05,
      "loss": 0.1691,
      "step": 5911
    },
    {
      "epoch": 0.2390449523385931,
      "grad_norm": 3.609046459197998,
      "learning_rate": 8.687960065390149e-05,
      "loss": 0.1622,
      "step": 5912
    },
    {
      "epoch": 0.2390853861938601,
      "grad_norm": 2.8119325637817383,
      "learning_rate": 8.687529409377845e-05,
      "loss": 0.1076,
      "step": 5913
    },
    {
      "epoch": 0.23912582004912714,
      "grad_norm": 2.6061933040618896,
      "learning_rate": 8.687098693376605e-05,
      "loss": 0.1958,
      "step": 5914
    },
    {
      "epoch": 0.23916625390439414,
      "grad_norm": 2.707136392593384,
      "learning_rate": 8.686667917393438e-05,
      "loss": 0.1092,
      "step": 5915
    },
    {
      "epoch": 0.23920668775966117,
      "grad_norm": 1.820064663887024,
      "learning_rate": 8.686237081435352e-05,
      "loss": 0.104,
      "step": 5916
    },
    {
      "epoch": 0.23924712161492817,
      "grad_norm": 3.2410926818847656,
      "learning_rate": 8.685806185509356e-05,
      "loss": 0.1474,
      "step": 5917
    },
    {
      "epoch": 0.2392875554701952,
      "grad_norm": 6.133182048797607,
      "learning_rate": 8.685375229622456e-05,
      "loss": 0.1022,
      "step": 5918
    },
    {
      "epoch": 0.2393279893254622,
      "grad_norm": 3.047478437423706,
      "learning_rate": 8.684944213781667e-05,
      "loss": 0.1127,
      "step": 5919
    },
    {
      "epoch": 0.23936842318072923,
      "grad_norm": 4.528893947601318,
      "learning_rate": 8.684513137994e-05,
      "loss": 0.2314,
      "step": 5920
    },
    {
      "epoch": 0.23940885703599624,
      "grad_norm": 7.739344120025635,
      "learning_rate": 8.684082002266466e-05,
      "loss": 0.3065,
      "step": 5921
    },
    {
      "epoch": 0.23944929089126327,
      "grad_norm": 1.0010637044906616,
      "learning_rate": 8.68365080660608e-05,
      "loss": 0.0412,
      "step": 5922
    },
    {
      "epoch": 0.23948972474653027,
      "grad_norm": 2.762718439102173,
      "learning_rate": 8.683219551019859e-05,
      "loss": 0.1276,
      "step": 5923
    },
    {
      "epoch": 0.23953015860179727,
      "grad_norm": 8.688218116760254,
      "learning_rate": 8.682788235514813e-05,
      "loss": 0.194,
      "step": 5924
    },
    {
      "epoch": 0.2395705924570643,
      "grad_norm": 3.121171712875366,
      "learning_rate": 8.682356860097965e-05,
      "loss": 0.2172,
      "step": 5925
    },
    {
      "epoch": 0.2396110263123313,
      "grad_norm": 3.0537304878234863,
      "learning_rate": 8.681925424776329e-05,
      "loss": 0.1375,
      "step": 5926
    },
    {
      "epoch": 0.23965146016759833,
      "grad_norm": 2.694565773010254,
      "learning_rate": 8.681493929556923e-05,
      "loss": 0.119,
      "step": 5927
    },
    {
      "epoch": 0.23969189402286534,
      "grad_norm": 1.5173289775848389,
      "learning_rate": 8.681062374446769e-05,
      "loss": 0.0391,
      "step": 5928
    },
    {
      "epoch": 0.23973232787813237,
      "grad_norm": 5.209552764892578,
      "learning_rate": 8.680630759452888e-05,
      "loss": 0.2003,
      "step": 5929
    },
    {
      "epoch": 0.23977276173339937,
      "grad_norm": 5.338234901428223,
      "learning_rate": 8.680199084582296e-05,
      "loss": 0.2738,
      "step": 5930
    },
    {
      "epoch": 0.2398131955886664,
      "grad_norm": 3.0946736335754395,
      "learning_rate": 8.679767349842023e-05,
      "loss": 0.1803,
      "step": 5931
    },
    {
      "epoch": 0.2398536294439334,
      "grad_norm": 1.4720056056976318,
      "learning_rate": 8.679335555239088e-05,
      "loss": 0.0486,
      "step": 5932
    },
    {
      "epoch": 0.23989406329920043,
      "grad_norm": 2.2390949726104736,
      "learning_rate": 8.678903700780516e-05,
      "loss": 0.0927,
      "step": 5933
    },
    {
      "epoch": 0.23993449715446744,
      "grad_norm": 2.8669111728668213,
      "learning_rate": 8.678471786473334e-05,
      "loss": 0.1533,
      "step": 5934
    },
    {
      "epoch": 0.23997493100973444,
      "grad_norm": 3.3295137882232666,
      "learning_rate": 8.678039812324567e-05,
      "loss": 0.0797,
      "step": 5935
    },
    {
      "epoch": 0.24001536486500147,
      "grad_norm": 5.791916847229004,
      "learning_rate": 8.677607778341241e-05,
      "loss": 0.173,
      "step": 5936
    },
    {
      "epoch": 0.24005579872026847,
      "grad_norm": 5.18472146987915,
      "learning_rate": 8.677175684530389e-05,
      "loss": 0.123,
      "step": 5937
    },
    {
      "epoch": 0.2400962325755355,
      "grad_norm": 4.101205825805664,
      "learning_rate": 8.676743530899036e-05,
      "loss": 0.1604,
      "step": 5938
    },
    {
      "epoch": 0.2401366664308025,
      "grad_norm": 6.613603591918945,
      "learning_rate": 8.676311317454213e-05,
      "loss": 0.3285,
      "step": 5939
    },
    {
      "epoch": 0.24017710028606953,
      "grad_norm": 3.877138137817383,
      "learning_rate": 8.675879044202952e-05,
      "loss": 0.1483,
      "step": 5940
    },
    {
      "epoch": 0.24021753414133654,
      "grad_norm": 4.349688529968262,
      "learning_rate": 8.675446711152286e-05,
      "loss": 0.1818,
      "step": 5941
    },
    {
      "epoch": 0.24025796799660357,
      "grad_norm": 4.138794898986816,
      "learning_rate": 8.675014318309247e-05,
      "loss": 0.1098,
      "step": 5942
    },
    {
      "epoch": 0.24029840185187057,
      "grad_norm": 7.146904945373535,
      "learning_rate": 8.674581865680868e-05,
      "loss": 0.0889,
      "step": 5943
    },
    {
      "epoch": 0.2403388357071376,
      "grad_norm": 3.275465250015259,
      "learning_rate": 8.674149353274188e-05,
      "loss": 0.0711,
      "step": 5944
    },
    {
      "epoch": 0.2403792695624046,
      "grad_norm": 3.784762382507324,
      "learning_rate": 8.673716781096241e-05,
      "loss": 0.0827,
      "step": 5945
    },
    {
      "epoch": 0.2404197034176716,
      "grad_norm": 6.69539737701416,
      "learning_rate": 8.673284149154062e-05,
      "loss": 0.2771,
      "step": 5946
    },
    {
      "epoch": 0.24046013727293863,
      "grad_norm": 7.916467666625977,
      "learning_rate": 8.672851457454694e-05,
      "loss": 0.3424,
      "step": 5947
    },
    {
      "epoch": 0.24050057112820564,
      "grad_norm": 2.8775734901428223,
      "learning_rate": 8.67241870600517e-05,
      "loss": 0.2087,
      "step": 5948
    },
    {
      "epoch": 0.24054100498347267,
      "grad_norm": 2.998142719268799,
      "learning_rate": 8.671985894812535e-05,
      "loss": 0.0888,
      "step": 5949
    },
    {
      "epoch": 0.24058143883873967,
      "grad_norm": 5.161098957061768,
      "learning_rate": 8.671553023883827e-05,
      "loss": 0.1359,
      "step": 5950
    },
    {
      "epoch": 0.2406218726940067,
      "grad_norm": 7.176464080810547,
      "learning_rate": 8.67112009322609e-05,
      "loss": 0.1972,
      "step": 5951
    },
    {
      "epoch": 0.2406623065492737,
      "grad_norm": 4.090066909790039,
      "learning_rate": 8.670687102846366e-05,
      "loss": 0.1149,
      "step": 5952
    },
    {
      "epoch": 0.24070274040454073,
      "grad_norm": 5.411351680755615,
      "learning_rate": 8.670254052751699e-05,
      "loss": 0.1941,
      "step": 5953
    },
    {
      "epoch": 0.24074317425980774,
      "grad_norm": 2.3755807876586914,
      "learning_rate": 8.669820942949134e-05,
      "loss": 0.2539,
      "step": 5954
    },
    {
      "epoch": 0.24078360811507477,
      "grad_norm": 6.3790388107299805,
      "learning_rate": 8.669387773445716e-05,
      "loss": 0.1938,
      "step": 5955
    },
    {
      "epoch": 0.24082404197034177,
      "grad_norm": 4.425403118133545,
      "learning_rate": 8.668954544248493e-05,
      "loss": 0.1793,
      "step": 5956
    },
    {
      "epoch": 0.24086447582560877,
      "grad_norm": 3.932326078414917,
      "learning_rate": 8.668521255364514e-05,
      "loss": 0.2266,
      "step": 5957
    },
    {
      "epoch": 0.2409049096808758,
      "grad_norm": 6.011348724365234,
      "learning_rate": 8.668087906800824e-05,
      "loss": 0.3062,
      "step": 5958
    },
    {
      "epoch": 0.2409453435361428,
      "grad_norm": 4.63346004486084,
      "learning_rate": 8.667654498564474e-05,
      "loss": 0.1196,
      "step": 5959
    },
    {
      "epoch": 0.24098577739140983,
      "grad_norm": 2.6123921871185303,
      "learning_rate": 8.667221030662518e-05,
      "loss": 0.1411,
      "step": 5960
    },
    {
      "epoch": 0.24102621124667684,
      "grad_norm": 3.8898696899414062,
      "learning_rate": 8.666787503102004e-05,
      "loss": 0.1585,
      "step": 5961
    },
    {
      "epoch": 0.24106664510194387,
      "grad_norm": 5.681251049041748,
      "learning_rate": 8.666353915889987e-05,
      "loss": 0.1883,
      "step": 5962
    },
    {
      "epoch": 0.24110707895721087,
      "grad_norm": 2.7869033813476562,
      "learning_rate": 8.665920269033518e-05,
      "loss": 0.1424,
      "step": 5963
    },
    {
      "epoch": 0.2411475128124779,
      "grad_norm": 4.215097904205322,
      "learning_rate": 8.665486562539653e-05,
      "loss": 0.1008,
      "step": 5964
    },
    {
      "epoch": 0.2411879466677449,
      "grad_norm": 12.377710342407227,
      "learning_rate": 8.665052796415448e-05,
      "loss": 0.338,
      "step": 5965
    },
    {
      "epoch": 0.24122838052301193,
      "grad_norm": 5.801630020141602,
      "learning_rate": 8.664618970667959e-05,
      "loss": 0.1894,
      "step": 5966
    },
    {
      "epoch": 0.24126881437827893,
      "grad_norm": 2.7168922424316406,
      "learning_rate": 8.664185085304244e-05,
      "loss": 0.0998,
      "step": 5967
    },
    {
      "epoch": 0.24130924823354594,
      "grad_norm": 2.9877679347991943,
      "learning_rate": 8.663751140331362e-05,
      "loss": 0.1775,
      "step": 5968
    },
    {
      "epoch": 0.24134968208881297,
      "grad_norm": 3.547353982925415,
      "learning_rate": 8.663317135756371e-05,
      "loss": 0.1909,
      "step": 5969
    },
    {
      "epoch": 0.24139011594407997,
      "grad_norm": 2.1182587146759033,
      "learning_rate": 8.662883071586332e-05,
      "loss": 0.1777,
      "step": 5970
    },
    {
      "epoch": 0.241430549799347,
      "grad_norm": 5.092315673828125,
      "learning_rate": 8.662448947828305e-05,
      "loss": 0.239,
      "step": 5971
    },
    {
      "epoch": 0.241470983654614,
      "grad_norm": 4.349262714385986,
      "learning_rate": 8.662014764489357e-05,
      "loss": 0.2238,
      "step": 5972
    },
    {
      "epoch": 0.24151141750988103,
      "grad_norm": 4.644000053405762,
      "learning_rate": 8.661580521576545e-05,
      "loss": 0.238,
      "step": 5973
    },
    {
      "epoch": 0.24155185136514803,
      "grad_norm": 3.5591719150543213,
      "learning_rate": 8.661146219096938e-05,
      "loss": 0.1155,
      "step": 5974
    },
    {
      "epoch": 0.24159228522041507,
      "grad_norm": 2.778860092163086,
      "learning_rate": 8.6607118570576e-05,
      "loss": 0.1316,
      "step": 5975
    },
    {
      "epoch": 0.24163271907568207,
      "grad_norm": 2.853416919708252,
      "learning_rate": 8.660277435465594e-05,
      "loss": 0.1694,
      "step": 5976
    },
    {
      "epoch": 0.24167315293094907,
      "grad_norm": 5.313234329223633,
      "learning_rate": 8.659842954327992e-05,
      "loss": 0.2186,
      "step": 5977
    },
    {
      "epoch": 0.2417135867862161,
      "grad_norm": 2.550760507583618,
      "learning_rate": 8.659408413651861e-05,
      "loss": 0.094,
      "step": 5978
    },
    {
      "epoch": 0.2417540206414831,
      "grad_norm": 3.766831874847412,
      "learning_rate": 8.65897381344427e-05,
      "loss": 0.2396,
      "step": 5979
    },
    {
      "epoch": 0.24179445449675013,
      "grad_norm": 7.09255313873291,
      "learning_rate": 8.658539153712286e-05,
      "loss": 0.2572,
      "step": 5980
    },
    {
      "epoch": 0.24183488835201714,
      "grad_norm": 6.072479724884033,
      "learning_rate": 8.658104434462982e-05,
      "loss": 0.1831,
      "step": 5981
    },
    {
      "epoch": 0.24187532220728417,
      "grad_norm": 4.733118057250977,
      "learning_rate": 8.657669655703433e-05,
      "loss": 0.1759,
      "step": 5982
    },
    {
      "epoch": 0.24191575606255117,
      "grad_norm": 5.013708114624023,
      "learning_rate": 8.657234817440708e-05,
      "loss": 0.2601,
      "step": 5983
    },
    {
      "epoch": 0.2419561899178182,
      "grad_norm": 4.62626314163208,
      "learning_rate": 8.656799919681883e-05,
      "loss": 0.2134,
      "step": 5984
    },
    {
      "epoch": 0.2419966237730852,
      "grad_norm": 3.8375024795532227,
      "learning_rate": 8.656364962434033e-05,
      "loss": 0.301,
      "step": 5985
    },
    {
      "epoch": 0.24203705762835223,
      "grad_norm": 2.5020620822906494,
      "learning_rate": 8.655929945704233e-05,
      "loss": 0.1129,
      "step": 5986
    },
    {
      "epoch": 0.24207749148361923,
      "grad_norm": 3.215634822845459,
      "learning_rate": 8.655494869499558e-05,
      "loss": 0.2104,
      "step": 5987
    },
    {
      "epoch": 0.24211792533888624,
      "grad_norm": 3.1493115425109863,
      "learning_rate": 8.65505973382709e-05,
      "loss": 0.1872,
      "step": 5988
    },
    {
      "epoch": 0.24215835919415327,
      "grad_norm": 3.9308154582977295,
      "learning_rate": 8.654624538693907e-05,
      "loss": 0.2378,
      "step": 5989
    },
    {
      "epoch": 0.24219879304942027,
      "grad_norm": 2.025744915008545,
      "learning_rate": 8.654189284107088e-05,
      "loss": 0.1456,
      "step": 5990
    },
    {
      "epoch": 0.2422392269046873,
      "grad_norm": 3.391645908355713,
      "learning_rate": 8.65375397007371e-05,
      "loss": 0.1188,
      "step": 5991
    },
    {
      "epoch": 0.2422796607599543,
      "grad_norm": 7.289383411407471,
      "learning_rate": 8.653318596600862e-05,
      "loss": 0.208,
      "step": 5992
    },
    {
      "epoch": 0.24232009461522133,
      "grad_norm": 3.654409408569336,
      "learning_rate": 8.65288316369562e-05,
      "loss": 0.1813,
      "step": 5993
    },
    {
      "epoch": 0.24236052847048833,
      "grad_norm": 4.555336952209473,
      "learning_rate": 8.652447671365072e-05,
      "loss": 0.1197,
      "step": 5994
    },
    {
      "epoch": 0.24240096232575536,
      "grad_norm": 4.378109931945801,
      "learning_rate": 8.6520121196163e-05,
      "loss": 0.1918,
      "step": 5995
    },
    {
      "epoch": 0.24244139618102237,
      "grad_norm": 1.2791835069656372,
      "learning_rate": 8.651576508456392e-05,
      "loss": 0.0739,
      "step": 5996
    },
    {
      "epoch": 0.2424818300362894,
      "grad_norm": 3.9173526763916016,
      "learning_rate": 8.651140837892432e-05,
      "loss": 0.1931,
      "step": 5997
    },
    {
      "epoch": 0.2425222638915564,
      "grad_norm": 4.513161659240723,
      "learning_rate": 8.65070510793151e-05,
      "loss": 0.0788,
      "step": 5998
    },
    {
      "epoch": 0.2425626977468234,
      "grad_norm": 7.073320388793945,
      "learning_rate": 8.65026931858071e-05,
      "loss": 0.2172,
      "step": 5999
    },
    {
      "epoch": 0.24260313160209043,
      "grad_norm": 3.8800010681152344,
      "learning_rate": 8.649833469847129e-05,
      "loss": 0.1563,
      "step": 6000
    },
    {
      "epoch": 0.24264356545735744,
      "grad_norm": 2.8107495307922363,
      "learning_rate": 8.649397561737851e-05,
      "loss": 0.2392,
      "step": 6001
    },
    {
      "epoch": 0.24268399931262447,
      "grad_norm": 8.04423713684082,
      "learning_rate": 8.648961594259968e-05,
      "loss": 0.2228,
      "step": 6002
    },
    {
      "epoch": 0.24272443316789147,
      "grad_norm": 6.598309516906738,
      "learning_rate": 8.648525567420576e-05,
      "loss": 0.1908,
      "step": 6003
    },
    {
      "epoch": 0.2427648670231585,
      "grad_norm": 3.7928972244262695,
      "learning_rate": 8.648089481226764e-05,
      "loss": 0.2214,
      "step": 6004
    },
    {
      "epoch": 0.2428053008784255,
      "grad_norm": 1.895093321800232,
      "learning_rate": 8.647653335685631e-05,
      "loss": 0.1524,
      "step": 6005
    },
    {
      "epoch": 0.24284573473369253,
      "grad_norm": 4.698232650756836,
      "learning_rate": 8.647217130804268e-05,
      "loss": 0.1546,
      "step": 6006
    },
    {
      "epoch": 0.24288616858895953,
      "grad_norm": 3.976858615875244,
      "learning_rate": 8.646780866589772e-05,
      "loss": 0.2591,
      "step": 6007
    },
    {
      "epoch": 0.24292660244422656,
      "grad_norm": 5.355378150939941,
      "learning_rate": 8.646344543049243e-05,
      "loss": 0.2391,
      "step": 6008
    },
    {
      "epoch": 0.24296703629949357,
      "grad_norm": 3.624940872192383,
      "learning_rate": 8.645908160189774e-05,
      "loss": 0.1035,
      "step": 6009
    },
    {
      "epoch": 0.24300747015476057,
      "grad_norm": 7.696066379547119,
      "learning_rate": 8.64547171801847e-05,
      "loss": 0.1876,
      "step": 6010
    },
    {
      "epoch": 0.2430479040100276,
      "grad_norm": 4.4118452072143555,
      "learning_rate": 8.645035216542428e-05,
      "loss": 0.1505,
      "step": 6011
    },
    {
      "epoch": 0.2430883378652946,
      "grad_norm": 9.357704162597656,
      "learning_rate": 8.64459865576875e-05,
      "loss": 0.2712,
      "step": 6012
    },
    {
      "epoch": 0.24312877172056163,
      "grad_norm": 2.713459014892578,
      "learning_rate": 8.644162035704535e-05,
      "loss": 0.1985,
      "step": 6013
    },
    {
      "epoch": 0.24316920557582863,
      "grad_norm": 5.679785251617432,
      "learning_rate": 8.643725356356889e-05,
      "loss": 0.1645,
      "step": 6014
    },
    {
      "epoch": 0.24320963943109566,
      "grad_norm": 7.06746244430542,
      "learning_rate": 8.643288617732917e-05,
      "loss": 0.2948,
      "step": 6015
    },
    {
      "epoch": 0.24325007328636267,
      "grad_norm": 3.614020347595215,
      "learning_rate": 8.642851819839719e-05,
      "loss": 0.2815,
      "step": 6016
    },
    {
      "epoch": 0.2432905071416297,
      "grad_norm": 8.398319244384766,
      "learning_rate": 8.642414962684406e-05,
      "loss": 0.1573,
      "step": 6017
    },
    {
      "epoch": 0.2433309409968967,
      "grad_norm": 5.846600532531738,
      "learning_rate": 8.641978046274082e-05,
      "loss": 0.1309,
      "step": 6018
    },
    {
      "epoch": 0.24337137485216373,
      "grad_norm": 2.964691400527954,
      "learning_rate": 8.641541070615855e-05,
      "loss": 0.0609,
      "step": 6019
    },
    {
      "epoch": 0.24341180870743073,
      "grad_norm": 4.7332444190979,
      "learning_rate": 8.641104035716836e-05,
      "loss": 0.2204,
      "step": 6020
    },
    {
      "epoch": 0.24345224256269773,
      "grad_norm": 3.9071009159088135,
      "learning_rate": 8.640666941584132e-05,
      "loss": 0.1772,
      "step": 6021
    },
    {
      "epoch": 0.24349267641796477,
      "grad_norm": 4.52919864654541,
      "learning_rate": 8.640229788224853e-05,
      "loss": 0.1845,
      "step": 6022
    },
    {
      "epoch": 0.24353311027323177,
      "grad_norm": 6.53327751159668,
      "learning_rate": 8.639792575646115e-05,
      "loss": 0.1407,
      "step": 6023
    },
    {
      "epoch": 0.2435735441284988,
      "grad_norm": 4.435870170593262,
      "learning_rate": 8.639355303855028e-05,
      "loss": 0.0937,
      "step": 6024
    },
    {
      "epoch": 0.2436139779837658,
      "grad_norm": 6.9111480712890625,
      "learning_rate": 8.638917972858704e-05,
      "loss": 0.2429,
      "step": 6025
    },
    {
      "epoch": 0.24365441183903283,
      "grad_norm": 4.262193202972412,
      "learning_rate": 8.638480582664259e-05,
      "loss": 0.22,
      "step": 6026
    },
    {
      "epoch": 0.24369484569429983,
      "grad_norm": 3.620594024658203,
      "learning_rate": 8.638043133278809e-05,
      "loss": 0.1129,
      "step": 6027
    },
    {
      "epoch": 0.24373527954956686,
      "grad_norm": 6.571465969085693,
      "learning_rate": 8.637605624709468e-05,
      "loss": 0.1726,
      "step": 6028
    },
    {
      "epoch": 0.24377571340483387,
      "grad_norm": 2.8471567630767822,
      "learning_rate": 8.637168056963357e-05,
      "loss": 0.0976,
      "step": 6029
    },
    {
      "epoch": 0.2438161472601009,
      "grad_norm": 6.479422092437744,
      "learning_rate": 8.636730430047593e-05,
      "loss": 0.2516,
      "step": 6030
    },
    {
      "epoch": 0.2438565811153679,
      "grad_norm": 2.0430047512054443,
      "learning_rate": 8.636292743969294e-05,
      "loss": 0.1564,
      "step": 6031
    },
    {
      "epoch": 0.2438970149706349,
      "grad_norm": 3.429826498031616,
      "learning_rate": 8.635854998735582e-05,
      "loss": 0.1513,
      "step": 6032
    },
    {
      "epoch": 0.24393744882590193,
      "grad_norm": 2.052880048751831,
      "learning_rate": 8.635417194353578e-05,
      "loss": 0.0597,
      "step": 6033
    },
    {
      "epoch": 0.24397788268116893,
      "grad_norm": 8.362695693969727,
      "learning_rate": 8.634979330830404e-05,
      "loss": 0.2508,
      "step": 6034
    },
    {
      "epoch": 0.24401831653643596,
      "grad_norm": 2.54459285736084,
      "learning_rate": 8.634541408173182e-05,
      "loss": 0.0596,
      "step": 6035
    },
    {
      "epoch": 0.24405875039170297,
      "grad_norm": 4.136692523956299,
      "learning_rate": 8.634103426389037e-05,
      "loss": 0.1318,
      "step": 6036
    },
    {
      "epoch": 0.24409918424697,
      "grad_norm": 10.016950607299805,
      "learning_rate": 8.633665385485094e-05,
      "loss": 0.2157,
      "step": 6037
    },
    {
      "epoch": 0.244139618102237,
      "grad_norm": 4.006833076477051,
      "learning_rate": 8.633227285468481e-05,
      "loss": 0.232,
      "step": 6038
    },
    {
      "epoch": 0.24418005195750403,
      "grad_norm": 3.8136184215545654,
      "learning_rate": 8.632789126346322e-05,
      "loss": 0.0961,
      "step": 6039
    },
    {
      "epoch": 0.24422048581277103,
      "grad_norm": 7.684732913970947,
      "learning_rate": 8.632350908125748e-05,
      "loss": 0.1847,
      "step": 6040
    },
    {
      "epoch": 0.24426091966803803,
      "grad_norm": 3.978747606277466,
      "learning_rate": 8.631912630813885e-05,
      "loss": 0.3372,
      "step": 6041
    },
    {
      "epoch": 0.24430135352330506,
      "grad_norm": 5.78717565536499,
      "learning_rate": 8.631474294417864e-05,
      "loss": 0.1573,
      "step": 6042
    },
    {
      "epoch": 0.24434178737857207,
      "grad_norm": 3.336500406265259,
      "learning_rate": 8.631035898944817e-05,
      "loss": 0.1248,
      "step": 6043
    },
    {
      "epoch": 0.2443822212338391,
      "grad_norm": 3.878305673599243,
      "learning_rate": 8.630597444401873e-05,
      "loss": 0.1764,
      "step": 6044
    },
    {
      "epoch": 0.2444226550891061,
      "grad_norm": 3.5070252418518066,
      "learning_rate": 8.63015893079617e-05,
      "loss": 0.1417,
      "step": 6045
    },
    {
      "epoch": 0.24446308894437313,
      "grad_norm": 9.929906845092773,
      "learning_rate": 8.629720358134836e-05,
      "loss": 0.2483,
      "step": 6046
    },
    {
      "epoch": 0.24450352279964013,
      "grad_norm": 2.988154888153076,
      "learning_rate": 8.62928172642501e-05,
      "loss": 0.2212,
      "step": 6047
    },
    {
      "epoch": 0.24454395665490716,
      "grad_norm": 2.689389944076538,
      "learning_rate": 8.628843035673824e-05,
      "loss": 0.0792,
      "step": 6048
    },
    {
      "epoch": 0.24458439051017417,
      "grad_norm": 12.93316650390625,
      "learning_rate": 8.628404285888418e-05,
      "loss": 0.2807,
      "step": 6049
    },
    {
      "epoch": 0.2446248243654412,
      "grad_norm": 9.077486038208008,
      "learning_rate": 8.627965477075928e-05,
      "loss": 0.3347,
      "step": 6050
    },
    {
      "epoch": 0.2446652582207082,
      "grad_norm": 15.951316833496094,
      "learning_rate": 8.627526609243492e-05,
      "loss": 0.3689,
      "step": 6051
    },
    {
      "epoch": 0.2447056920759752,
      "grad_norm": 2.9297943115234375,
      "learning_rate": 8.627087682398252e-05,
      "loss": 0.1827,
      "step": 6052
    },
    {
      "epoch": 0.24474612593124223,
      "grad_norm": 3.517852544784546,
      "learning_rate": 8.626648696547346e-05,
      "loss": 0.087,
      "step": 6053
    },
    {
      "epoch": 0.24478655978650923,
      "grad_norm": 11.33461856842041,
      "learning_rate": 8.626209651697916e-05,
      "loss": 0.337,
      "step": 6054
    },
    {
      "epoch": 0.24482699364177626,
      "grad_norm": 3.5571320056915283,
      "learning_rate": 8.625770547857104e-05,
      "loss": 0.1222,
      "step": 6055
    },
    {
      "epoch": 0.24486742749704327,
      "grad_norm": 3.389450788497925,
      "learning_rate": 8.625331385032056e-05,
      "loss": 0.1753,
      "step": 6056
    },
    {
      "epoch": 0.2449078613523103,
      "grad_norm": 4.702533721923828,
      "learning_rate": 8.624892163229912e-05,
      "loss": 0.1709,
      "step": 6057
    },
    {
      "epoch": 0.2449482952075773,
      "grad_norm": 6.528069972991943,
      "learning_rate": 8.624452882457821e-05,
      "loss": 0.1442,
      "step": 6058
    },
    {
      "epoch": 0.24498872906284433,
      "grad_norm": 2.3681833744049072,
      "learning_rate": 8.624013542722927e-05,
      "loss": 0.1071,
      "step": 6059
    },
    {
      "epoch": 0.24502916291811133,
      "grad_norm": 6.04899263381958,
      "learning_rate": 8.623574144032379e-05,
      "loss": 0.2262,
      "step": 6060
    },
    {
      "epoch": 0.24506959677337836,
      "grad_norm": 3.0587050914764404,
      "learning_rate": 8.623134686393324e-05,
      "loss": 0.1647,
      "step": 6061
    },
    {
      "epoch": 0.24511003062864536,
      "grad_norm": 3.833815813064575,
      "learning_rate": 8.622695169812911e-05,
      "loss": 0.1766,
      "step": 6062
    },
    {
      "epoch": 0.24515046448391237,
      "grad_norm": 5.266330718994141,
      "learning_rate": 8.622255594298291e-05,
      "loss": 0.2156,
      "step": 6063
    },
    {
      "epoch": 0.2451908983391794,
      "grad_norm": 3.1624538898468018,
      "learning_rate": 8.621815959856614e-05,
      "loss": 0.1147,
      "step": 6064
    },
    {
      "epoch": 0.2452313321944464,
      "grad_norm": 7.124099254608154,
      "learning_rate": 8.621376266495031e-05,
      "loss": 0.274,
      "step": 6065
    },
    {
      "epoch": 0.24527176604971343,
      "grad_norm": 2.109743118286133,
      "learning_rate": 8.620936514220699e-05,
      "loss": 0.0597,
      "step": 6066
    },
    {
      "epoch": 0.24531219990498043,
      "grad_norm": 5.317684650421143,
      "learning_rate": 8.620496703040769e-05,
      "loss": 0.2001,
      "step": 6067
    },
    {
      "epoch": 0.24535263376024746,
      "grad_norm": 4.574336528778076,
      "learning_rate": 8.620056832962394e-05,
      "loss": 0.1516,
      "step": 6068
    },
    {
      "epoch": 0.24539306761551447,
      "grad_norm": 4.148901462554932,
      "learning_rate": 8.619616903992733e-05,
      "loss": 0.0779,
      "step": 6069
    },
    {
      "epoch": 0.2454335014707815,
      "grad_norm": 5.344276428222656,
      "learning_rate": 8.619176916138944e-05,
      "loss": 0.1638,
      "step": 6070
    },
    {
      "epoch": 0.2454739353260485,
      "grad_norm": 4.519772529602051,
      "learning_rate": 8.61873686940818e-05,
      "loss": 0.2012,
      "step": 6071
    },
    {
      "epoch": 0.24551436918131553,
      "grad_norm": 4.1637115478515625,
      "learning_rate": 8.618296763807603e-05,
      "loss": 0.1288,
      "step": 6072
    },
    {
      "epoch": 0.24555480303658253,
      "grad_norm": 8.29826545715332,
      "learning_rate": 8.617856599344372e-05,
      "loss": 0.2354,
      "step": 6073
    },
    {
      "epoch": 0.24559523689184953,
      "grad_norm": 6.626035213470459,
      "learning_rate": 8.617416376025649e-05,
      "loss": 0.1596,
      "step": 6074
    },
    {
      "epoch": 0.24563567074711656,
      "grad_norm": 5.0871477127075195,
      "learning_rate": 8.616976093858591e-05,
      "loss": 0.1753,
      "step": 6075
    },
    {
      "epoch": 0.24567610460238357,
      "grad_norm": 4.304733753204346,
      "learning_rate": 8.616535752850366e-05,
      "loss": 0.2885,
      "step": 6076
    },
    {
      "epoch": 0.2457165384576506,
      "grad_norm": 3.2352781295776367,
      "learning_rate": 8.616095353008135e-05,
      "loss": 0.1896,
      "step": 6077
    },
    {
      "epoch": 0.2457569723129176,
      "grad_norm": 2.7523484230041504,
      "learning_rate": 8.615654894339063e-05,
      "loss": 0.1781,
      "step": 6078
    },
    {
      "epoch": 0.24579740616818463,
      "grad_norm": 3.727778673171997,
      "learning_rate": 8.615214376850314e-05,
      "loss": 0.1103,
      "step": 6079
    },
    {
      "epoch": 0.24583784002345163,
      "grad_norm": 3.9669647216796875,
      "learning_rate": 8.614773800549055e-05,
      "loss": 0.0879,
      "step": 6080
    },
    {
      "epoch": 0.24587827387871866,
      "grad_norm": 6.558189392089844,
      "learning_rate": 8.614333165442454e-05,
      "loss": 0.1847,
      "step": 6081
    },
    {
      "epoch": 0.24591870773398566,
      "grad_norm": 6.306224346160889,
      "learning_rate": 8.61389247153768e-05,
      "loss": 0.1495,
      "step": 6082
    },
    {
      "epoch": 0.2459591415892527,
      "grad_norm": 3.4565958976745605,
      "learning_rate": 8.6134517188419e-05,
      "loss": 0.0916,
      "step": 6083
    },
    {
      "epoch": 0.2459995754445197,
      "grad_norm": 4.450873851776123,
      "learning_rate": 8.613010907362287e-05,
      "loss": 0.1449,
      "step": 6084
    },
    {
      "epoch": 0.2460400092997867,
      "grad_norm": 6.050655364990234,
      "learning_rate": 8.612570037106009e-05,
      "loss": 0.2411,
      "step": 6085
    },
    {
      "epoch": 0.24608044315505373,
      "grad_norm": 4.706817150115967,
      "learning_rate": 8.61212910808024e-05,
      "loss": 0.0666,
      "step": 6086
    },
    {
      "epoch": 0.24612087701032073,
      "grad_norm": 2.352297067642212,
      "learning_rate": 8.611688120292155e-05,
      "loss": 0.1034,
      "step": 6087
    },
    {
      "epoch": 0.24616131086558776,
      "grad_norm": 4.81610107421875,
      "learning_rate": 8.611247073748922e-05,
      "loss": 0.0981,
      "step": 6088
    },
    {
      "epoch": 0.24620174472085476,
      "grad_norm": 8.78715991973877,
      "learning_rate": 8.610805968457721e-05,
      "loss": 0.211,
      "step": 6089
    },
    {
      "epoch": 0.2462421785761218,
      "grad_norm": 6.008256912231445,
      "learning_rate": 8.610364804425728e-05,
      "loss": 0.0955,
      "step": 6090
    },
    {
      "epoch": 0.2462826124313888,
      "grad_norm": 4.646786212921143,
      "learning_rate": 8.609923581660118e-05,
      "loss": 0.2153,
      "step": 6091
    },
    {
      "epoch": 0.24632304628665583,
      "grad_norm": 1.5845496654510498,
      "learning_rate": 8.609482300168067e-05,
      "loss": 0.0434,
      "step": 6092
    },
    {
      "epoch": 0.24636348014192283,
      "grad_norm": 2.4028897285461426,
      "learning_rate": 8.609040959956759e-05,
      "loss": 0.0609,
      "step": 6093
    },
    {
      "epoch": 0.24640391399718986,
      "grad_norm": 36.975887298583984,
      "learning_rate": 8.608599561033369e-05,
      "loss": 0.1087,
      "step": 6094
    },
    {
      "epoch": 0.24644434785245686,
      "grad_norm": 7.599268913269043,
      "learning_rate": 8.60815810340508e-05,
      "loss": 0.2089,
      "step": 6095
    },
    {
      "epoch": 0.24648478170772387,
      "grad_norm": 2.8600575923919678,
      "learning_rate": 8.607716587079074e-05,
      "loss": 0.0742,
      "step": 6096
    },
    {
      "epoch": 0.2465252155629909,
      "grad_norm": 2.9021806716918945,
      "learning_rate": 8.607275012062532e-05,
      "loss": 0.0799,
      "step": 6097
    },
    {
      "epoch": 0.2465656494182579,
      "grad_norm": 4.6386332511901855,
      "learning_rate": 8.606833378362639e-05,
      "loss": 0.1038,
      "step": 6098
    },
    {
      "epoch": 0.24660608327352493,
      "grad_norm": 5.46524715423584,
      "learning_rate": 8.606391685986579e-05,
      "loss": 0.1079,
      "step": 6099
    },
    {
      "epoch": 0.24664651712879193,
      "grad_norm": 4.315202236175537,
      "learning_rate": 8.605949934941537e-05,
      "loss": 0.1017,
      "step": 6100
    },
    {
      "epoch": 0.24668695098405896,
      "grad_norm": 4.34230375289917,
      "learning_rate": 8.6055081252347e-05,
      "loss": 0.135,
      "step": 6101
    },
    {
      "epoch": 0.24672738483932596,
      "grad_norm": 10.488526344299316,
      "learning_rate": 8.605066256873256e-05,
      "loss": 0.2443,
      "step": 6102
    },
    {
      "epoch": 0.246767818694593,
      "grad_norm": 4.642398357391357,
      "learning_rate": 8.604624329864392e-05,
      "loss": 0.1518,
      "step": 6103
    },
    {
      "epoch": 0.24680825254986,
      "grad_norm": 3.78501296043396,
      "learning_rate": 8.604182344215298e-05,
      "loss": 0.0736,
      "step": 6104
    },
    {
      "epoch": 0.246848686405127,
      "grad_norm": 9.167596817016602,
      "learning_rate": 8.603740299933164e-05,
      "loss": 0.3418,
      "step": 6105
    },
    {
      "epoch": 0.24688912026039403,
      "grad_norm": 7.805789947509766,
      "learning_rate": 8.603298197025181e-05,
      "loss": 0.4001,
      "step": 6106
    },
    {
      "epoch": 0.24692955411566103,
      "grad_norm": 6.4627275466918945,
      "learning_rate": 8.60285603549854e-05,
      "loss": 0.2139,
      "step": 6107
    },
    {
      "epoch": 0.24696998797092806,
      "grad_norm": 2.7034740447998047,
      "learning_rate": 8.602413815360439e-05,
      "loss": 0.0862,
      "step": 6108
    },
    {
      "epoch": 0.24701042182619506,
      "grad_norm": 3.240323066711426,
      "learning_rate": 8.601971536618067e-05,
      "loss": 0.1273,
      "step": 6109
    },
    {
      "epoch": 0.2470508556814621,
      "grad_norm": 4.7289886474609375,
      "learning_rate": 8.60152919927862e-05,
      "loss": 0.2225,
      "step": 6110
    },
    {
      "epoch": 0.2470912895367291,
      "grad_norm": 3.309140920639038,
      "learning_rate": 8.601086803349296e-05,
      "loss": 0.0905,
      "step": 6111
    },
    {
      "epoch": 0.24713172339199613,
      "grad_norm": 6.798061847686768,
      "learning_rate": 8.60064434883729e-05,
      "loss": 0.2208,
      "step": 6112
    },
    {
      "epoch": 0.24717215724726313,
      "grad_norm": 3.6816961765289307,
      "learning_rate": 8.6002018357498e-05,
      "loss": 0.1211,
      "step": 6113
    },
    {
      "epoch": 0.24721259110253016,
      "grad_norm": 3.832526683807373,
      "learning_rate": 8.599759264094025e-05,
      "loss": 0.0823,
      "step": 6114
    },
    {
      "epoch": 0.24725302495779716,
      "grad_norm": 3.5318474769592285,
      "learning_rate": 8.599316633877164e-05,
      "loss": 0.1528,
      "step": 6115
    },
    {
      "epoch": 0.24729345881306417,
      "grad_norm": 3.617408514022827,
      "learning_rate": 8.59887394510642e-05,
      "loss": 0.0787,
      "step": 6116
    },
    {
      "epoch": 0.2473338926683312,
      "grad_norm": 5.636387348175049,
      "learning_rate": 8.598431197788994e-05,
      "loss": 0.1539,
      "step": 6117
    },
    {
      "epoch": 0.2473743265235982,
      "grad_norm": 6.630018711090088,
      "learning_rate": 8.597988391932089e-05,
      "loss": 0.4171,
      "step": 6118
    },
    {
      "epoch": 0.24741476037886523,
      "grad_norm": 7.672569751739502,
      "learning_rate": 8.597545527542905e-05,
      "loss": 0.3314,
      "step": 6119
    },
    {
      "epoch": 0.24745519423413223,
      "grad_norm": 7.134288787841797,
      "learning_rate": 8.597102604628651e-05,
      "loss": 0.2681,
      "step": 6120
    },
    {
      "epoch": 0.24749562808939926,
      "grad_norm": 7.464917182922363,
      "learning_rate": 8.59665962319653e-05,
      "loss": 0.1661,
      "step": 6121
    },
    {
      "epoch": 0.24753606194466626,
      "grad_norm": 4.142001152038574,
      "learning_rate": 8.59621658325375e-05,
      "loss": 0.2082,
      "step": 6122
    },
    {
      "epoch": 0.2475764957999333,
      "grad_norm": 4.803020000457764,
      "learning_rate": 8.595773484807516e-05,
      "loss": 0.2261,
      "step": 6123
    },
    {
      "epoch": 0.2476169296552003,
      "grad_norm": 3.1889495849609375,
      "learning_rate": 8.595330327865038e-05,
      "loss": 0.141,
      "step": 6124
    },
    {
      "epoch": 0.24765736351046733,
      "grad_norm": 3.5087060928344727,
      "learning_rate": 8.594887112433528e-05,
      "loss": 0.1012,
      "step": 6125
    },
    {
      "epoch": 0.24769779736573433,
      "grad_norm": 3.0082900524139404,
      "learning_rate": 8.594443838520191e-05,
      "loss": 0.1087,
      "step": 6126
    },
    {
      "epoch": 0.24773823122100133,
      "grad_norm": 5.638223648071289,
      "learning_rate": 8.594000506132241e-05,
      "loss": 0.1215,
      "step": 6127
    },
    {
      "epoch": 0.24777866507626836,
      "grad_norm": 2.4479122161865234,
      "learning_rate": 8.59355711527689e-05,
      "loss": 0.1683,
      "step": 6128
    },
    {
      "epoch": 0.24781909893153536,
      "grad_norm": 4.197991847991943,
      "learning_rate": 8.59311366596135e-05,
      "loss": 0.093,
      "step": 6129
    },
    {
      "epoch": 0.2478595327868024,
      "grad_norm": 6.671213626861572,
      "learning_rate": 8.592670158192838e-05,
      "loss": 0.2066,
      "step": 6130
    },
    {
      "epoch": 0.2478999666420694,
      "grad_norm": 5.721414566040039,
      "learning_rate": 8.592226591978566e-05,
      "loss": 0.2359,
      "step": 6131
    },
    {
      "epoch": 0.24794040049733643,
      "grad_norm": 7.26845121383667,
      "learning_rate": 8.59178296732575e-05,
      "loss": 0.1464,
      "step": 6132
    },
    {
      "epoch": 0.24798083435260343,
      "grad_norm": 3.415755271911621,
      "learning_rate": 8.59133928424161e-05,
      "loss": 0.1107,
      "step": 6133
    },
    {
      "epoch": 0.24802126820787046,
      "grad_norm": 2.2114908695220947,
      "learning_rate": 8.59089554273336e-05,
      "loss": 0.0952,
      "step": 6134
    },
    {
      "epoch": 0.24806170206313746,
      "grad_norm": 4.224102973937988,
      "learning_rate": 8.590451742808222e-05,
      "loss": 0.1822,
      "step": 6135
    },
    {
      "epoch": 0.2481021359184045,
      "grad_norm": 5.626123428344727,
      "learning_rate": 8.590007884473413e-05,
      "loss": 0.1277,
      "step": 6136
    },
    {
      "epoch": 0.2481425697736715,
      "grad_norm": 2.7514543533325195,
      "learning_rate": 8.589563967736154e-05,
      "loss": 0.2924,
      "step": 6137
    },
    {
      "epoch": 0.2481830036289385,
      "grad_norm": 4.606257438659668,
      "learning_rate": 8.589119992603669e-05,
      "loss": 0.2058,
      "step": 6138
    },
    {
      "epoch": 0.24822343748420553,
      "grad_norm": 6.313202381134033,
      "learning_rate": 8.58867595908318e-05,
      "loss": 0.1745,
      "step": 6139
    },
    {
      "epoch": 0.24826387133947253,
      "grad_norm": 4.425397872924805,
      "learning_rate": 8.588231867181908e-05,
      "loss": 0.1273,
      "step": 6140
    },
    {
      "epoch": 0.24830430519473956,
      "grad_norm": 4.079669952392578,
      "learning_rate": 8.587787716907081e-05,
      "loss": 0.1124,
      "step": 6141
    },
    {
      "epoch": 0.24834473905000656,
      "grad_norm": 2.386849880218506,
      "learning_rate": 8.587343508265923e-05,
      "loss": 0.1633,
      "step": 6142
    },
    {
      "epoch": 0.2483851729052736,
      "grad_norm": 7.280966758728027,
      "learning_rate": 8.58689924126566e-05,
      "loss": 0.2683,
      "step": 6143
    },
    {
      "epoch": 0.2484256067605406,
      "grad_norm": 2.142650604248047,
      "learning_rate": 8.586454915913517e-05,
      "loss": 0.1698,
      "step": 6144
    },
    {
      "epoch": 0.24846604061580763,
      "grad_norm": 7.144465923309326,
      "learning_rate": 8.586010532216728e-05,
      "loss": 0.1161,
      "step": 6145
    },
    {
      "epoch": 0.24850647447107463,
      "grad_norm": 4.681589126586914,
      "learning_rate": 8.585566090182518e-05,
      "loss": 0.1186,
      "step": 6146
    },
    {
      "epoch": 0.24854690832634166,
      "grad_norm": 3.8650546073913574,
      "learning_rate": 8.58512158981812e-05,
      "loss": 0.1997,
      "step": 6147
    },
    {
      "epoch": 0.24858734218160866,
      "grad_norm": 9.318419456481934,
      "learning_rate": 8.584677031130761e-05,
      "loss": 0.2923,
      "step": 6148
    },
    {
      "epoch": 0.24862777603687566,
      "grad_norm": 4.759102821350098,
      "learning_rate": 8.584232414127678e-05,
      "loss": 0.1804,
      "step": 6149
    },
    {
      "epoch": 0.2486682098921427,
      "grad_norm": 3.2844924926757812,
      "learning_rate": 8.5837877388161e-05,
      "loss": 0.1689,
      "step": 6150
    },
    {
      "epoch": 0.2487086437474097,
      "grad_norm": 9.399534225463867,
      "learning_rate": 8.583343005203263e-05,
      "loss": 0.3574,
      "step": 6151
    },
    {
      "epoch": 0.24874907760267673,
      "grad_norm": 4.286575794219971,
      "learning_rate": 8.582898213296402e-05,
      "loss": 0.2597,
      "step": 6152
    },
    {
      "epoch": 0.24878951145794373,
      "grad_norm": 5.7185139656066895,
      "learning_rate": 8.582453363102752e-05,
      "loss": 0.2317,
      "step": 6153
    },
    {
      "epoch": 0.24882994531321076,
      "grad_norm": 5.736219882965088,
      "learning_rate": 8.58200845462955e-05,
      "loss": 0.1812,
      "step": 6154
    },
    {
      "epoch": 0.24887037916847776,
      "grad_norm": 5.837985515594482,
      "learning_rate": 8.581563487884035e-05,
      "loss": 0.1813,
      "step": 6155
    },
    {
      "epoch": 0.2489108130237448,
      "grad_norm": 5.658602237701416,
      "learning_rate": 8.581118462873446e-05,
      "loss": 0.1545,
      "step": 6156
    },
    {
      "epoch": 0.2489512468790118,
      "grad_norm": 4.969576358795166,
      "learning_rate": 8.58067337960502e-05,
      "loss": 0.1486,
      "step": 6157
    },
    {
      "epoch": 0.24899168073427883,
      "grad_norm": 2.319491386413574,
      "learning_rate": 8.580228238085999e-05,
      "loss": 0.1649,
      "step": 6158
    },
    {
      "epoch": 0.24903211458954583,
      "grad_norm": 3.767211437225342,
      "learning_rate": 8.579783038323626e-05,
      "loss": 0.1828,
      "step": 6159
    },
    {
      "epoch": 0.24907254844481283,
      "grad_norm": 3.399191379547119,
      "learning_rate": 8.579337780325142e-05,
      "loss": 0.1291,
      "step": 6160
    },
    {
      "epoch": 0.24911298230007986,
      "grad_norm": 3.8261640071868896,
      "learning_rate": 8.57889246409779e-05,
      "loss": 0.1082,
      "step": 6161
    },
    {
      "epoch": 0.24915341615534686,
      "grad_norm": 2.4140303134918213,
      "learning_rate": 8.578447089648816e-05,
      "loss": 0.1318,
      "step": 6162
    },
    {
      "epoch": 0.2491938500106139,
      "grad_norm": 7.104363918304443,
      "learning_rate": 8.578001656985466e-05,
      "loss": 0.1731,
      "step": 6163
    },
    {
      "epoch": 0.2492342838658809,
      "grad_norm": 7.832826137542725,
      "learning_rate": 8.577556166114984e-05,
      "loss": 0.299,
      "step": 6164
    },
    {
      "epoch": 0.24927471772114793,
      "grad_norm": 8.464363098144531,
      "learning_rate": 8.577110617044618e-05,
      "loss": 0.3275,
      "step": 6165
    },
    {
      "epoch": 0.24931515157641493,
      "grad_norm": 4.684128761291504,
      "learning_rate": 8.576665009781616e-05,
      "loss": 0.2326,
      "step": 6166
    },
    {
      "epoch": 0.24935558543168196,
      "grad_norm": 7.360625743865967,
      "learning_rate": 8.576219344333227e-05,
      "loss": 0.2637,
      "step": 6167
    },
    {
      "epoch": 0.24939601928694896,
      "grad_norm": 5.580816268920898,
      "learning_rate": 8.575773620706704e-05,
      "loss": 0.2078,
      "step": 6168
    },
    {
      "epoch": 0.24943645314221596,
      "grad_norm": 4.870668411254883,
      "learning_rate": 8.575327838909295e-05,
      "loss": 0.1566,
      "step": 6169
    },
    {
      "epoch": 0.249476886997483,
      "grad_norm": 3.1624975204467773,
      "learning_rate": 8.574881998948252e-05,
      "loss": 0.2021,
      "step": 6170
    },
    {
      "epoch": 0.24951732085275,
      "grad_norm": 2.021010398864746,
      "learning_rate": 8.574436100830829e-05,
      "loss": 0.0836,
      "step": 6171
    },
    {
      "epoch": 0.24955775470801703,
      "grad_norm": 2.359396457672119,
      "learning_rate": 8.573990144564281e-05,
      "loss": 0.1261,
      "step": 6172
    },
    {
      "epoch": 0.24959818856328403,
      "grad_norm": 14.166296005249023,
      "learning_rate": 8.57354413015586e-05,
      "loss": 0.3957,
      "step": 6173
    },
    {
      "epoch": 0.24963862241855106,
      "grad_norm": 4.486057281494141,
      "learning_rate": 8.573098057612822e-05,
      "loss": 0.1729,
      "step": 6174
    },
    {
      "epoch": 0.24967905627381806,
      "grad_norm": 4.341570854187012,
      "learning_rate": 8.572651926942428e-05,
      "loss": 0.0956,
      "step": 6175
    },
    {
      "epoch": 0.2497194901290851,
      "grad_norm": 6.579067707061768,
      "learning_rate": 8.572205738151931e-05,
      "loss": 0.1692,
      "step": 6176
    },
    {
      "epoch": 0.2497599239843521,
      "grad_norm": 4.9769206047058105,
      "learning_rate": 8.571759491248593e-05,
      "loss": 0.1608,
      "step": 6177
    },
    {
      "epoch": 0.24980035783961912,
      "grad_norm": 11.039288520812988,
      "learning_rate": 8.57131318623967e-05,
      "loss": 0.3244,
      "step": 6178
    },
    {
      "epoch": 0.24984079169488613,
      "grad_norm": 2.834900140762329,
      "learning_rate": 8.570866823132425e-05,
      "loss": 0.1042,
      "step": 6179
    },
    {
      "epoch": 0.24988122555015313,
      "grad_norm": 5.615602016448975,
      "learning_rate": 8.57042040193412e-05,
      "loss": 0.1991,
      "step": 6180
    },
    {
      "epoch": 0.24992165940542016,
      "grad_norm": 5.0720086097717285,
      "learning_rate": 8.569973922652015e-05,
      "loss": 0.1036,
      "step": 6181
    },
    {
      "epoch": 0.24996209326068716,
      "grad_norm": 4.303607940673828,
      "learning_rate": 8.569527385293375e-05,
      "loss": 0.2971,
      "step": 6182
    },
    {
      "epoch": 0.25000252711595417,
      "grad_norm": 2.68996262550354,
      "learning_rate": 8.569080789865463e-05,
      "loss": 0.0876,
      "step": 6183
    },
    {
      "epoch": 0.2500429609712212,
      "grad_norm": 3.8559789657592773,
      "learning_rate": 8.568634136375547e-05,
      "loss": 0.1207,
      "step": 6184
    },
    {
      "epoch": 0.2500833948264882,
      "grad_norm": 6.243417739868164,
      "learning_rate": 8.568187424830889e-05,
      "loss": 0.2087,
      "step": 6185
    },
    {
      "epoch": 0.25012382868175526,
      "grad_norm": 5.405614376068115,
      "learning_rate": 8.567740655238761e-05,
      "loss": 0.1594,
      "step": 6186
    },
    {
      "epoch": 0.25016426253702223,
      "grad_norm": 3.042715311050415,
      "learning_rate": 8.567293827606427e-05,
      "loss": 0.0963,
      "step": 6187
    },
    {
      "epoch": 0.25020469639228926,
      "grad_norm": 8.87225341796875,
      "learning_rate": 8.566846941941158e-05,
      "loss": 0.243,
      "step": 6188
    },
    {
      "epoch": 0.2502451302475563,
      "grad_norm": 3.334763288497925,
      "learning_rate": 8.566399998250224e-05,
      "loss": 0.0835,
      "step": 6189
    },
    {
      "epoch": 0.2502855641028233,
      "grad_norm": 5.4543538093566895,
      "learning_rate": 8.565952996540893e-05,
      "loss": 0.1584,
      "step": 6190
    },
    {
      "epoch": 0.2503259979580903,
      "grad_norm": 6.160861968994141,
      "learning_rate": 8.565505936820441e-05,
      "loss": 0.1298,
      "step": 6191
    },
    {
      "epoch": 0.2503664318133573,
      "grad_norm": 5.069806098937988,
      "learning_rate": 8.565058819096139e-05,
      "loss": 0.2268,
      "step": 6192
    },
    {
      "epoch": 0.25040686566862436,
      "grad_norm": 5.266380786895752,
      "learning_rate": 8.564611643375263e-05,
      "loss": 0.1274,
      "step": 6193
    },
    {
      "epoch": 0.25044729952389133,
      "grad_norm": 5.647245407104492,
      "learning_rate": 8.564164409665083e-05,
      "loss": 0.1708,
      "step": 6194
    },
    {
      "epoch": 0.25048773337915836,
      "grad_norm": 8.582027435302734,
      "learning_rate": 8.563717117972876e-05,
      "loss": 0.193,
      "step": 6195
    },
    {
      "epoch": 0.2505281672344254,
      "grad_norm": 4.458442687988281,
      "learning_rate": 8.563269768305922e-05,
      "loss": 0.3056,
      "step": 6196
    },
    {
      "epoch": 0.2505686010896924,
      "grad_norm": 3.2652697563171387,
      "learning_rate": 8.562822360671495e-05,
      "loss": 0.1163,
      "step": 6197
    },
    {
      "epoch": 0.2506090349449594,
      "grad_norm": 3.184579849243164,
      "learning_rate": 8.562374895076874e-05,
      "loss": 0.1153,
      "step": 6198
    },
    {
      "epoch": 0.2506494688002264,
      "grad_norm": 4.188009262084961,
      "learning_rate": 8.561927371529341e-05,
      "loss": 0.1579,
      "step": 6199
    },
    {
      "epoch": 0.25068990265549346,
      "grad_norm": 1.9505306482315063,
      "learning_rate": 8.561479790036175e-05,
      "loss": 0.0645,
      "step": 6200
    },
    {
      "epoch": 0.25073033651076043,
      "grad_norm": 4.53508996963501,
      "learning_rate": 8.561032150604655e-05,
      "loss": 0.1473,
      "step": 6201
    },
    {
      "epoch": 0.25077077036602746,
      "grad_norm": 3.168933391571045,
      "learning_rate": 8.560584453242064e-05,
      "loss": 0.1491,
      "step": 6202
    },
    {
      "epoch": 0.2508112042212945,
      "grad_norm": 3.8186163902282715,
      "learning_rate": 8.560136697955687e-05,
      "loss": 0.1645,
      "step": 6203
    },
    {
      "epoch": 0.2508516380765615,
      "grad_norm": 4.747062683105469,
      "learning_rate": 8.559688884752808e-05,
      "loss": 0.1574,
      "step": 6204
    },
    {
      "epoch": 0.2508920719318285,
      "grad_norm": 4.399488925933838,
      "learning_rate": 8.55924101364071e-05,
      "loss": 0.2242,
      "step": 6205
    },
    {
      "epoch": 0.25093250578709553,
      "grad_norm": 6.263681411743164,
      "learning_rate": 8.55879308462668e-05,
      "loss": 0.2975,
      "step": 6206
    },
    {
      "epoch": 0.25097293964236256,
      "grad_norm": 5.284090518951416,
      "learning_rate": 8.558345097718006e-05,
      "loss": 0.1785,
      "step": 6207
    },
    {
      "epoch": 0.2510133734976296,
      "grad_norm": 5.420502185821533,
      "learning_rate": 8.557897052921975e-05,
      "loss": 0.1035,
      "step": 6208
    },
    {
      "epoch": 0.25105380735289656,
      "grad_norm": 4.130248069763184,
      "learning_rate": 8.557448950245874e-05,
      "loss": 0.2202,
      "step": 6209
    },
    {
      "epoch": 0.2510942412081636,
      "grad_norm": 3.97829008102417,
      "learning_rate": 8.557000789696997e-05,
      "loss": 0.0936,
      "step": 6210
    },
    {
      "epoch": 0.2511346750634306,
      "grad_norm": 5.584280967712402,
      "learning_rate": 8.556552571282631e-05,
      "loss": 0.2014,
      "step": 6211
    },
    {
      "epoch": 0.2511751089186976,
      "grad_norm": 4.516365051269531,
      "learning_rate": 8.55610429501007e-05,
      "loss": 0.2308,
      "step": 6212
    },
    {
      "epoch": 0.25121554277396463,
      "grad_norm": 4.201866626739502,
      "learning_rate": 8.555655960886603e-05,
      "loss": 0.1656,
      "step": 6213
    },
    {
      "epoch": 0.25125597662923166,
      "grad_norm": 5.899411201477051,
      "learning_rate": 8.555207568919529e-05,
      "loss": 0.2022,
      "step": 6214
    },
    {
      "epoch": 0.2512964104844987,
      "grad_norm": 5.131168842315674,
      "learning_rate": 8.554759119116139e-05,
      "loss": 0.1892,
      "step": 6215
    },
    {
      "epoch": 0.25133684433976566,
      "grad_norm": 3.4465293884277344,
      "learning_rate": 8.554310611483727e-05,
      "loss": 0.193,
      "step": 6216
    },
    {
      "epoch": 0.2513772781950327,
      "grad_norm": 1.9942100048065186,
      "learning_rate": 8.553862046029593e-05,
      "loss": 0.0839,
      "step": 6217
    },
    {
      "epoch": 0.2514177120502997,
      "grad_norm": 3.042116403579712,
      "learning_rate": 8.553413422761032e-05,
      "loss": 0.1122,
      "step": 6218
    },
    {
      "epoch": 0.25145814590556675,
      "grad_norm": 4.3754472732543945,
      "learning_rate": 8.552964741685343e-05,
      "loss": 0.1772,
      "step": 6219
    },
    {
      "epoch": 0.25149857976083373,
      "grad_norm": 2.8776440620422363,
      "learning_rate": 8.552516002809824e-05,
      "loss": 0.0951,
      "step": 6220
    },
    {
      "epoch": 0.25153901361610076,
      "grad_norm": 3.3530313968658447,
      "learning_rate": 8.552067206141776e-05,
      "loss": 0.1026,
      "step": 6221
    },
    {
      "epoch": 0.2515794474713678,
      "grad_norm": 7.831663131713867,
      "learning_rate": 8.551618351688502e-05,
      "loss": 0.2196,
      "step": 6222
    },
    {
      "epoch": 0.25161988132663476,
      "grad_norm": 4.250004291534424,
      "learning_rate": 8.5511694394573e-05,
      "loss": 0.2105,
      "step": 6223
    },
    {
      "epoch": 0.2516603151819018,
      "grad_norm": 5.238514423370361,
      "learning_rate": 8.550720469455476e-05,
      "loss": 0.1809,
      "step": 6224
    },
    {
      "epoch": 0.2517007490371688,
      "grad_norm": 4.478693008422852,
      "learning_rate": 8.550271441690333e-05,
      "loss": 0.1392,
      "step": 6225
    },
    {
      "epoch": 0.25174118289243586,
      "grad_norm": 2.730726480484009,
      "learning_rate": 8.549822356169174e-05,
      "loss": 0.1215,
      "step": 6226
    },
    {
      "epoch": 0.25178161674770283,
      "grad_norm": 4.786006450653076,
      "learning_rate": 8.54937321289931e-05,
      "loss": 0.1379,
      "step": 6227
    },
    {
      "epoch": 0.25182205060296986,
      "grad_norm": 3.0412774085998535,
      "learning_rate": 8.548924011888041e-05,
      "loss": 0.0681,
      "step": 6228
    },
    {
      "epoch": 0.2518624844582369,
      "grad_norm": 7.8732500076293945,
      "learning_rate": 8.548474753142678e-05,
      "loss": 0.2025,
      "step": 6229
    },
    {
      "epoch": 0.2519029183135039,
      "grad_norm": 5.494780540466309,
      "learning_rate": 8.54802543667053e-05,
      "loss": 0.1258,
      "step": 6230
    },
    {
      "epoch": 0.2519433521687709,
      "grad_norm": 4.78334379196167,
      "learning_rate": 8.547576062478907e-05,
      "loss": 0.306,
      "step": 6231
    },
    {
      "epoch": 0.2519837860240379,
      "grad_norm": 4.669282913208008,
      "learning_rate": 8.547126630575117e-05,
      "loss": 0.267,
      "step": 6232
    },
    {
      "epoch": 0.25202421987930496,
      "grad_norm": 4.810275077819824,
      "learning_rate": 8.546677140966473e-05,
      "loss": 0.0795,
      "step": 6233
    },
    {
      "epoch": 0.25206465373457193,
      "grad_norm": 6.018242359161377,
      "learning_rate": 8.546227593660286e-05,
      "loss": 0.2052,
      "step": 6234
    },
    {
      "epoch": 0.25210508758983896,
      "grad_norm": 4.984010696411133,
      "learning_rate": 8.545777988663873e-05,
      "loss": 0.1985,
      "step": 6235
    },
    {
      "epoch": 0.252145521445106,
      "grad_norm": 6.683145046234131,
      "learning_rate": 8.545328325984541e-05,
      "loss": 0.0936,
      "step": 6236
    },
    {
      "epoch": 0.252185955300373,
      "grad_norm": 3.997206449508667,
      "learning_rate": 8.544878605629614e-05,
      "loss": 0.1981,
      "step": 6237
    },
    {
      "epoch": 0.25222638915564,
      "grad_norm": 4.37819766998291,
      "learning_rate": 8.544428827606401e-05,
      "loss": 0.1916,
      "step": 6238
    },
    {
      "epoch": 0.252266823010907,
      "grad_norm": 6.649611473083496,
      "learning_rate": 8.543978991922223e-05,
      "loss": 0.1783,
      "step": 6239
    },
    {
      "epoch": 0.25230725686617406,
      "grad_norm": 2.1568360328674316,
      "learning_rate": 8.543529098584395e-05,
      "loss": 0.1872,
      "step": 6240
    },
    {
      "epoch": 0.2523476907214411,
      "grad_norm": 2.9699277877807617,
      "learning_rate": 8.543079147600239e-05,
      "loss": 0.1311,
      "step": 6241
    },
    {
      "epoch": 0.25238812457670806,
      "grad_norm": 1.729875087738037,
      "learning_rate": 8.542629138977072e-05,
      "loss": 0.1097,
      "step": 6242
    },
    {
      "epoch": 0.2524285584319751,
      "grad_norm": 10.33169174194336,
      "learning_rate": 8.542179072722216e-05,
      "loss": 0.2242,
      "step": 6243
    },
    {
      "epoch": 0.2524689922872421,
      "grad_norm": 6.025792121887207,
      "learning_rate": 8.541728948842994e-05,
      "loss": 0.1638,
      "step": 6244
    },
    {
      "epoch": 0.2525094261425091,
      "grad_norm": 4.260978698730469,
      "learning_rate": 8.541278767346728e-05,
      "loss": 0.1998,
      "step": 6245
    },
    {
      "epoch": 0.2525498599977761,
      "grad_norm": 1.7004098892211914,
      "learning_rate": 8.540828528240739e-05,
      "loss": 0.0731,
      "step": 6246
    },
    {
      "epoch": 0.25259029385304316,
      "grad_norm": 4.788060188293457,
      "learning_rate": 8.540378231532356e-05,
      "loss": 0.0807,
      "step": 6247
    },
    {
      "epoch": 0.2526307277083102,
      "grad_norm": 5.246776580810547,
      "learning_rate": 8.5399278772289e-05,
      "loss": 0.1346,
      "step": 6248
    },
    {
      "epoch": 0.25267116156357716,
      "grad_norm": 8.667241096496582,
      "learning_rate": 8.5394774653377e-05,
      "loss": 0.2778,
      "step": 6249
    },
    {
      "epoch": 0.2527115954188442,
      "grad_norm": 2.9778223037719727,
      "learning_rate": 8.539026995866082e-05,
      "loss": 0.1028,
      "step": 6250
    },
    {
      "epoch": 0.2527520292741112,
      "grad_norm": 4.117793560028076,
      "learning_rate": 8.538576468821375e-05,
      "loss": 0.0893,
      "step": 6251
    },
    {
      "epoch": 0.25279246312937825,
      "grad_norm": 2.23382830619812,
      "learning_rate": 8.53812588421091e-05,
      "loss": 0.0806,
      "step": 6252
    },
    {
      "epoch": 0.25283289698464523,
      "grad_norm": 3.395962953567505,
      "learning_rate": 8.537675242042014e-05,
      "loss": 0.1227,
      "step": 6253
    },
    {
      "epoch": 0.25287333083991226,
      "grad_norm": 2.672717571258545,
      "learning_rate": 8.53722454232202e-05,
      "loss": 0.1515,
      "step": 6254
    },
    {
      "epoch": 0.2529137646951793,
      "grad_norm": 3.2836570739746094,
      "learning_rate": 8.536773785058259e-05,
      "loss": 0.0952,
      "step": 6255
    },
    {
      "epoch": 0.25295419855044626,
      "grad_norm": 7.9911980628967285,
      "learning_rate": 8.536322970258064e-05,
      "loss": 0.2271,
      "step": 6256
    },
    {
      "epoch": 0.2529946324057133,
      "grad_norm": 3.975504159927368,
      "learning_rate": 8.535872097928768e-05,
      "loss": 0.1517,
      "step": 6257
    },
    {
      "epoch": 0.2530350662609803,
      "grad_norm": 8.91515064239502,
      "learning_rate": 8.535421168077708e-05,
      "loss": 0.2666,
      "step": 6258
    },
    {
      "epoch": 0.25307550011624735,
      "grad_norm": 4.1537909507751465,
      "learning_rate": 8.534970180712221e-05,
      "loss": 0.1067,
      "step": 6259
    },
    {
      "epoch": 0.25311593397151433,
      "grad_norm": 3.3577077388763428,
      "learning_rate": 8.534519135839638e-05,
      "loss": 0.1172,
      "step": 6260
    },
    {
      "epoch": 0.25315636782678136,
      "grad_norm": 4.607162952423096,
      "learning_rate": 8.534068033467302e-05,
      "loss": 0.2664,
      "step": 6261
    },
    {
      "epoch": 0.2531968016820484,
      "grad_norm": 4.48886251449585,
      "learning_rate": 8.53361687360255e-05,
      "loss": 0.2104,
      "step": 6262
    },
    {
      "epoch": 0.2532372355373154,
      "grad_norm": 3.975416898727417,
      "learning_rate": 8.533165656252719e-05,
      "loss": 0.1517,
      "step": 6263
    },
    {
      "epoch": 0.2532776693925824,
      "grad_norm": 3.51596736907959,
      "learning_rate": 8.532714381425154e-05,
      "loss": 0.208,
      "step": 6264
    },
    {
      "epoch": 0.2533181032478494,
      "grad_norm": 7.504252910614014,
      "learning_rate": 8.532263049127192e-05,
      "loss": 0.4486,
      "step": 6265
    },
    {
      "epoch": 0.25335853710311645,
      "grad_norm": 8.218213081359863,
      "learning_rate": 8.531811659366178e-05,
      "loss": 0.2256,
      "step": 6266
    },
    {
      "epoch": 0.25339897095838343,
      "grad_norm": 3.353452444076538,
      "learning_rate": 8.531360212149455e-05,
      "loss": 0.1014,
      "step": 6267
    },
    {
      "epoch": 0.25343940481365046,
      "grad_norm": 8.242406845092773,
      "learning_rate": 8.530908707484367e-05,
      "loss": 0.3201,
      "step": 6268
    },
    {
      "epoch": 0.2534798386689175,
      "grad_norm": 3.9833481311798096,
      "learning_rate": 8.530457145378258e-05,
      "loss": 0.124,
      "step": 6269
    },
    {
      "epoch": 0.2535202725241845,
      "grad_norm": 3.9097211360931396,
      "learning_rate": 8.530005525838474e-05,
      "loss": 0.1897,
      "step": 6270
    },
    {
      "epoch": 0.2535607063794515,
      "grad_norm": 3.90010666847229,
      "learning_rate": 8.529553848872364e-05,
      "loss": 0.1575,
      "step": 6271
    },
    {
      "epoch": 0.2536011402347185,
      "grad_norm": 2.2660040855407715,
      "learning_rate": 8.529102114487273e-05,
      "loss": 0.0755,
      "step": 6272
    },
    {
      "epoch": 0.25364157408998556,
      "grad_norm": 5.888993263244629,
      "learning_rate": 8.528650322690555e-05,
      "loss": 0.189,
      "step": 6273
    },
    {
      "epoch": 0.2536820079452526,
      "grad_norm": 5.217118263244629,
      "learning_rate": 8.528198473489553e-05,
      "loss": 0.2706,
      "step": 6274
    },
    {
      "epoch": 0.25372244180051956,
      "grad_norm": 5.54807186126709,
      "learning_rate": 8.527746566891622e-05,
      "loss": 0.2454,
      "step": 6275
    },
    {
      "epoch": 0.2537628756557866,
      "grad_norm": 2.200016498565674,
      "learning_rate": 8.527294602904112e-05,
      "loss": 0.064,
      "step": 6276
    },
    {
      "epoch": 0.2538033095110536,
      "grad_norm": 3.341926336288452,
      "learning_rate": 8.526842581534376e-05,
      "loss": 0.086,
      "step": 6277
    },
    {
      "epoch": 0.2538437433663206,
      "grad_norm": 5.0617356300354,
      "learning_rate": 8.526390502789769e-05,
      "loss": 0.1915,
      "step": 6278
    },
    {
      "epoch": 0.2538841772215876,
      "grad_norm": 4.246654510498047,
      "learning_rate": 8.525938366677644e-05,
      "loss": 0.2351,
      "step": 6279
    },
    {
      "epoch": 0.25392461107685466,
      "grad_norm": 2.459105968475342,
      "learning_rate": 8.525486173205355e-05,
      "loss": 0.135,
      "step": 6280
    },
    {
      "epoch": 0.2539650449321217,
      "grad_norm": 6.659311771392822,
      "learning_rate": 8.525033922380261e-05,
      "loss": 0.2173,
      "step": 6281
    },
    {
      "epoch": 0.25400547878738866,
      "grad_norm": 4.312849521636963,
      "learning_rate": 8.524581614209717e-05,
      "loss": 0.1036,
      "step": 6282
    },
    {
      "epoch": 0.2540459126426557,
      "grad_norm": 4.339302062988281,
      "learning_rate": 8.524129248701083e-05,
      "loss": 0.1592,
      "step": 6283
    },
    {
      "epoch": 0.2540863464979227,
      "grad_norm": 7.011333465576172,
      "learning_rate": 8.523676825861719e-05,
      "loss": 0.2355,
      "step": 6284
    },
    {
      "epoch": 0.25412678035318975,
      "grad_norm": 3.377826690673828,
      "learning_rate": 8.523224345698981e-05,
      "loss": 0.0733,
      "step": 6285
    },
    {
      "epoch": 0.2541672142084567,
      "grad_norm": 5.842650890350342,
      "learning_rate": 8.522771808220233e-05,
      "loss": 0.2519,
      "step": 6286
    },
    {
      "epoch": 0.25420764806372376,
      "grad_norm": 5.8505659103393555,
      "learning_rate": 8.522319213432837e-05,
      "loss": 0.1435,
      "step": 6287
    },
    {
      "epoch": 0.2542480819189908,
      "grad_norm": 5.959987640380859,
      "learning_rate": 8.521866561344154e-05,
      "loss": 0.2371,
      "step": 6288
    },
    {
      "epoch": 0.25428851577425776,
      "grad_norm": 2.4413418769836426,
      "learning_rate": 8.52141385196155e-05,
      "loss": 0.089,
      "step": 6289
    },
    {
      "epoch": 0.2543289496295248,
      "grad_norm": 4.990113258361816,
      "learning_rate": 8.520961085292387e-05,
      "loss": 0.1568,
      "step": 6290
    },
    {
      "epoch": 0.2543693834847918,
      "grad_norm": 7.304047584533691,
      "learning_rate": 8.520508261344033e-05,
      "loss": 0.2882,
      "step": 6291
    },
    {
      "epoch": 0.25440981734005885,
      "grad_norm": 3.4831466674804688,
      "learning_rate": 8.520055380123855e-05,
      "loss": 0.1084,
      "step": 6292
    },
    {
      "epoch": 0.2544502511953258,
      "grad_norm": 3.8137497901916504,
      "learning_rate": 8.519602441639217e-05,
      "loss": 0.2194,
      "step": 6293
    },
    {
      "epoch": 0.25449068505059286,
      "grad_norm": 5.360727310180664,
      "learning_rate": 8.519149445897491e-05,
      "loss": 0.157,
      "step": 6294
    },
    {
      "epoch": 0.2545311189058599,
      "grad_norm": 6.933289527893066,
      "learning_rate": 8.518696392906045e-05,
      "loss": 0.2431,
      "step": 6295
    },
    {
      "epoch": 0.2545715527611269,
      "grad_norm": 5.3872809410095215,
      "learning_rate": 8.518243282672249e-05,
      "loss": 0.1341,
      "step": 6296
    },
    {
      "epoch": 0.2546119866163939,
      "grad_norm": 2.950385093688965,
      "learning_rate": 8.517790115203475e-05,
      "loss": 0.0747,
      "step": 6297
    },
    {
      "epoch": 0.2546524204716609,
      "grad_norm": 5.414254665374756,
      "learning_rate": 8.517336890507095e-05,
      "loss": 0.2068,
      "step": 6298
    },
    {
      "epoch": 0.25469285432692795,
      "grad_norm": 6.815707206726074,
      "learning_rate": 8.516883608590481e-05,
      "loss": 0.3236,
      "step": 6299
    },
    {
      "epoch": 0.25473328818219493,
      "grad_norm": 4.279081344604492,
      "learning_rate": 8.516430269461009e-05,
      "loss": 0.2014,
      "step": 6300
    },
    {
      "epoch": 0.25477372203746196,
      "grad_norm": 2.455984592437744,
      "learning_rate": 8.515976873126051e-05,
      "loss": 0.1983,
      "step": 6301
    },
    {
      "epoch": 0.254814155892729,
      "grad_norm": 3.6150600910186768,
      "learning_rate": 8.515523419592985e-05,
      "loss": 0.1326,
      "step": 6302
    },
    {
      "epoch": 0.254854589747996,
      "grad_norm": 2.019244432449341,
      "learning_rate": 8.51506990886919e-05,
      "loss": 0.1406,
      "step": 6303
    },
    {
      "epoch": 0.254895023603263,
      "grad_norm": 6.8846635818481445,
      "learning_rate": 8.514616340962038e-05,
      "loss": 0.2659,
      "step": 6304
    },
    {
      "epoch": 0.25493545745853,
      "grad_norm": 3.416940689086914,
      "learning_rate": 8.514162715878912e-05,
      "loss": 0.2755,
      "step": 6305
    },
    {
      "epoch": 0.25497589131379705,
      "grad_norm": 4.088443279266357,
      "learning_rate": 8.51370903362719e-05,
      "loss": 0.0884,
      "step": 6306
    },
    {
      "epoch": 0.2550163251690641,
      "grad_norm": 1.9420490264892578,
      "learning_rate": 8.513255294214255e-05,
      "loss": 0.0845,
      "step": 6307
    },
    {
      "epoch": 0.25505675902433106,
      "grad_norm": 2.7024238109588623,
      "learning_rate": 8.512801497647484e-05,
      "loss": 0.197,
      "step": 6308
    },
    {
      "epoch": 0.2550971928795981,
      "grad_norm": 4.50208044052124,
      "learning_rate": 8.512347643934264e-05,
      "loss": 0.2195,
      "step": 6309
    },
    {
      "epoch": 0.2551376267348651,
      "grad_norm": 5.4285569190979,
      "learning_rate": 8.511893733081975e-05,
      "loss": 0.2219,
      "step": 6310
    },
    {
      "epoch": 0.2551780605901321,
      "grad_norm": 3.334282875061035,
      "learning_rate": 8.511439765098001e-05,
      "loss": 0.2564,
      "step": 6311
    },
    {
      "epoch": 0.2552184944453991,
      "grad_norm": 3.280755043029785,
      "learning_rate": 8.510985739989731e-05,
      "loss": 0.1022,
      "step": 6312
    },
    {
      "epoch": 0.25525892830066615,
      "grad_norm": 3.1413607597351074,
      "learning_rate": 8.510531657764548e-05,
      "loss": 0.1204,
      "step": 6313
    },
    {
      "epoch": 0.2552993621559332,
      "grad_norm": 3.8937203884124756,
      "learning_rate": 8.510077518429839e-05,
      "loss": 0.1427,
      "step": 6314
    },
    {
      "epoch": 0.25533979601120016,
      "grad_norm": 9.105998039245605,
      "learning_rate": 8.509623321992993e-05,
      "loss": 0.2413,
      "step": 6315
    },
    {
      "epoch": 0.2553802298664672,
      "grad_norm": 3.748194694519043,
      "learning_rate": 8.509169068461398e-05,
      "loss": 0.2387,
      "step": 6316
    },
    {
      "epoch": 0.2554206637217342,
      "grad_norm": 2.9352807998657227,
      "learning_rate": 8.508714757842445e-05,
      "loss": 0.1515,
      "step": 6317
    },
    {
      "epoch": 0.25546109757700125,
      "grad_norm": 7.0299482345581055,
      "learning_rate": 8.508260390143524e-05,
      "loss": 0.2198,
      "step": 6318
    },
    {
      "epoch": 0.2555015314322682,
      "grad_norm": 6.920150279998779,
      "learning_rate": 8.507805965372026e-05,
      "loss": 0.2549,
      "step": 6319
    },
    {
      "epoch": 0.25554196528753526,
      "grad_norm": 3.6378939151763916,
      "learning_rate": 8.507351483535346e-05,
      "loss": 0.1282,
      "step": 6320
    },
    {
      "epoch": 0.2555823991428023,
      "grad_norm": 3.546814441680908,
      "learning_rate": 8.506896944640876e-05,
      "loss": 0.2543,
      "step": 6321
    },
    {
      "epoch": 0.25562283299806926,
      "grad_norm": 5.38877534866333,
      "learning_rate": 8.50644234869601e-05,
      "loss": 0.126,
      "step": 6322
    },
    {
      "epoch": 0.2556632668533363,
      "grad_norm": 2.71343994140625,
      "learning_rate": 8.505987695708144e-05,
      "loss": 0.0831,
      "step": 6323
    },
    {
      "epoch": 0.2557037007086033,
      "grad_norm": 4.955891132354736,
      "learning_rate": 8.505532985684673e-05,
      "loss": 0.1422,
      "step": 6324
    },
    {
      "epoch": 0.25574413456387035,
      "grad_norm": 2.52089786529541,
      "learning_rate": 8.505078218632996e-05,
      "loss": 0.1084,
      "step": 6325
    },
    {
      "epoch": 0.2557845684191373,
      "grad_norm": 4.579136848449707,
      "learning_rate": 8.504623394560511e-05,
      "loss": 0.2748,
      "step": 6326
    },
    {
      "epoch": 0.25582500227440436,
      "grad_norm": 3.9160637855529785,
      "learning_rate": 8.504168513474617e-05,
      "loss": 0.1203,
      "step": 6327
    },
    {
      "epoch": 0.2558654361296714,
      "grad_norm": 4.7934160232543945,
      "learning_rate": 8.503713575382712e-05,
      "loss": 0.1336,
      "step": 6328
    },
    {
      "epoch": 0.2559058699849384,
      "grad_norm": 6.95398473739624,
      "learning_rate": 8.5032585802922e-05,
      "loss": 0.1431,
      "step": 6329
    },
    {
      "epoch": 0.2559463038402054,
      "grad_norm": 4.516506671905518,
      "learning_rate": 8.50280352821048e-05,
      "loss": 0.2555,
      "step": 6330
    },
    {
      "epoch": 0.2559867376954724,
      "grad_norm": 5.366239070892334,
      "learning_rate": 8.502348419144959e-05,
      "loss": 0.1554,
      "step": 6331
    },
    {
      "epoch": 0.25602717155073945,
      "grad_norm": 4.950640678405762,
      "learning_rate": 8.501893253103036e-05,
      "loss": 0.4409,
      "step": 6332
    },
    {
      "epoch": 0.2560676054060064,
      "grad_norm": 5.238677024841309,
      "learning_rate": 8.501438030092119e-05,
      "loss": 0.2381,
      "step": 6333
    },
    {
      "epoch": 0.25610803926127346,
      "grad_norm": 3.452730178833008,
      "learning_rate": 8.500982750119613e-05,
      "loss": 0.1077,
      "step": 6334
    },
    {
      "epoch": 0.2561484731165405,
      "grad_norm": 5.175906181335449,
      "learning_rate": 8.500527413192922e-05,
      "loss": 0.2344,
      "step": 6335
    },
    {
      "epoch": 0.2561889069718075,
      "grad_norm": 5.024079322814941,
      "learning_rate": 8.500072019319456e-05,
      "loss": 0.1239,
      "step": 6336
    },
    {
      "epoch": 0.2562293408270745,
      "grad_norm": 3.608715772628784,
      "learning_rate": 8.499616568506624e-05,
      "loss": 0.1508,
      "step": 6337
    },
    {
      "epoch": 0.2562697746823415,
      "grad_norm": 3.933455228805542,
      "learning_rate": 8.499161060761831e-05,
      "loss": 0.1152,
      "step": 6338
    },
    {
      "epoch": 0.25631020853760855,
      "grad_norm": 2.304415702819824,
      "learning_rate": 8.498705496092492e-05,
      "loss": 0.0763,
      "step": 6339
    },
    {
      "epoch": 0.2563506423928755,
      "grad_norm": 2.713329553604126,
      "learning_rate": 8.498249874506016e-05,
      "loss": 0.1807,
      "step": 6340
    },
    {
      "epoch": 0.25639107624814256,
      "grad_norm": 4.124869346618652,
      "learning_rate": 8.497794196009816e-05,
      "loss": 0.1265,
      "step": 6341
    },
    {
      "epoch": 0.2564315101034096,
      "grad_norm": 4.125143051147461,
      "learning_rate": 8.497338460611303e-05,
      "loss": 0.1421,
      "step": 6342
    },
    {
      "epoch": 0.2564719439586766,
      "grad_norm": 5.568498134613037,
      "learning_rate": 8.496882668317896e-05,
      "loss": 0.1275,
      "step": 6343
    },
    {
      "epoch": 0.2565123778139436,
      "grad_norm": 3.043327808380127,
      "learning_rate": 8.496426819137003e-05,
      "loss": 0.089,
      "step": 6344
    },
    {
      "epoch": 0.2565528116692106,
      "grad_norm": 1.9587103128433228,
      "learning_rate": 8.495970913076043e-05,
      "loss": 0.0663,
      "step": 6345
    },
    {
      "epoch": 0.25659324552447765,
      "grad_norm": 5.768272399902344,
      "learning_rate": 8.495514950142433e-05,
      "loss": 0.1119,
      "step": 6346
    },
    {
      "epoch": 0.2566336793797447,
      "grad_norm": 4.563802242279053,
      "learning_rate": 8.495058930343592e-05,
      "loss": 0.1408,
      "step": 6347
    },
    {
      "epoch": 0.25667411323501166,
      "grad_norm": 3.0931496620178223,
      "learning_rate": 8.494602853686936e-05,
      "loss": 0.1668,
      "step": 6348
    },
    {
      "epoch": 0.2567145470902787,
      "grad_norm": 3.356067657470703,
      "learning_rate": 8.494146720179885e-05,
      "loss": 0.1318,
      "step": 6349
    },
    {
      "epoch": 0.2567549809455457,
      "grad_norm": 8.33302116394043,
      "learning_rate": 8.493690529829861e-05,
      "loss": 0.2704,
      "step": 6350
    },
    {
      "epoch": 0.2567954148008127,
      "grad_norm": 7.626204013824463,
      "learning_rate": 8.493234282644282e-05,
      "loss": 0.2377,
      "step": 6351
    },
    {
      "epoch": 0.2568358486560797,
      "grad_norm": 2.6277921199798584,
      "learning_rate": 8.492777978630574e-05,
      "loss": 0.2527,
      "step": 6352
    },
    {
      "epoch": 0.25687628251134675,
      "grad_norm": 1.9599628448486328,
      "learning_rate": 8.492321617796157e-05,
      "loss": 0.0878,
      "step": 6353
    },
    {
      "epoch": 0.2569167163666138,
      "grad_norm": 6.0915703773498535,
      "learning_rate": 8.491865200148458e-05,
      "loss": 0.2377,
      "step": 6354
    },
    {
      "epoch": 0.25695715022188076,
      "grad_norm": 2.4235754013061523,
      "learning_rate": 8.491408725694902e-05,
      "loss": 0.07,
      "step": 6355
    },
    {
      "epoch": 0.2569975840771478,
      "grad_norm": 4.932924747467041,
      "learning_rate": 8.490952194442912e-05,
      "loss": 0.1753,
      "step": 6356
    },
    {
      "epoch": 0.2570380179324148,
      "grad_norm": 2.017972707748413,
      "learning_rate": 8.490495606399915e-05,
      "loss": 0.0895,
      "step": 6357
    },
    {
      "epoch": 0.25707845178768185,
      "grad_norm": 3.0645298957824707,
      "learning_rate": 8.490038961573344e-05,
      "loss": 0.1286,
      "step": 6358
    },
    {
      "epoch": 0.2571188856429488,
      "grad_norm": 6.763504981994629,
      "learning_rate": 8.489582259970621e-05,
      "loss": 0.1615,
      "step": 6359
    },
    {
      "epoch": 0.25715931949821585,
      "grad_norm": 10.51475715637207,
      "learning_rate": 8.48912550159918e-05,
      "loss": 0.2463,
      "step": 6360
    },
    {
      "epoch": 0.2571997533534829,
      "grad_norm": 6.952187538146973,
      "learning_rate": 8.48866868646645e-05,
      "loss": 0.2318,
      "step": 6361
    },
    {
      "epoch": 0.25724018720874986,
      "grad_norm": 7.947668552398682,
      "learning_rate": 8.488211814579862e-05,
      "loss": 0.1819,
      "step": 6362
    },
    {
      "epoch": 0.2572806210640169,
      "grad_norm": 5.73841667175293,
      "learning_rate": 8.48775488594685e-05,
      "loss": 0.1764,
      "step": 6363
    },
    {
      "epoch": 0.2573210549192839,
      "grad_norm": 5.9241533279418945,
      "learning_rate": 8.487297900574847e-05,
      "loss": 0.1824,
      "step": 6364
    },
    {
      "epoch": 0.25736148877455095,
      "grad_norm": 5.1504387855529785,
      "learning_rate": 8.486840858471286e-05,
      "loss": 0.1751,
      "step": 6365
    },
    {
      "epoch": 0.2574019226298179,
      "grad_norm": 5.617988109588623,
      "learning_rate": 8.486383759643602e-05,
      "loss": 0.1754,
      "step": 6366
    },
    {
      "epoch": 0.25744235648508496,
      "grad_norm": 4.458637237548828,
      "learning_rate": 8.485926604099233e-05,
      "loss": 0.1956,
      "step": 6367
    },
    {
      "epoch": 0.257482790340352,
      "grad_norm": 5.067800998687744,
      "learning_rate": 8.485469391845615e-05,
      "loss": 0.1718,
      "step": 6368
    },
    {
      "epoch": 0.257523224195619,
      "grad_norm": 3.406689167022705,
      "learning_rate": 8.485012122890187e-05,
      "loss": 0.1398,
      "step": 6369
    },
    {
      "epoch": 0.257563658050886,
      "grad_norm": 9.782722473144531,
      "learning_rate": 8.484554797240387e-05,
      "loss": 0.2298,
      "step": 6370
    },
    {
      "epoch": 0.257604091906153,
      "grad_norm": 16.46589469909668,
      "learning_rate": 8.484097414903654e-05,
      "loss": 0.3264,
      "step": 6371
    },
    {
      "epoch": 0.25764452576142005,
      "grad_norm": 3.9720211029052734,
      "learning_rate": 8.483639975887429e-05,
      "loss": 0.1507,
      "step": 6372
    },
    {
      "epoch": 0.257684959616687,
      "grad_norm": 2.8719277381896973,
      "learning_rate": 8.483182480199155e-05,
      "loss": 0.0907,
      "step": 6373
    },
    {
      "epoch": 0.25772539347195406,
      "grad_norm": 1.6157563924789429,
      "learning_rate": 8.482724927846273e-05,
      "loss": 0.0721,
      "step": 6374
    },
    {
      "epoch": 0.2577658273272211,
      "grad_norm": 1.839153528213501,
      "learning_rate": 8.482267318836225e-05,
      "loss": 0.1382,
      "step": 6375
    },
    {
      "epoch": 0.2578062611824881,
      "grad_norm": 4.852220058441162,
      "learning_rate": 8.481809653176462e-05,
      "loss": 0.2959,
      "step": 6376
    },
    {
      "epoch": 0.2578466950377551,
      "grad_norm": 2.376772165298462,
      "learning_rate": 8.481351930874423e-05,
      "loss": 0.0954,
      "step": 6377
    },
    {
      "epoch": 0.2578871288930221,
      "grad_norm": 2.801809549331665,
      "learning_rate": 8.480894151937556e-05,
      "loss": 0.1136,
      "step": 6378
    },
    {
      "epoch": 0.25792756274828915,
      "grad_norm": 2.7802624702453613,
      "learning_rate": 8.480436316373308e-05,
      "loss": 0.1464,
      "step": 6379
    },
    {
      "epoch": 0.2579679966035562,
      "grad_norm": 3.514580249786377,
      "learning_rate": 8.479978424189128e-05,
      "loss": 0.0766,
      "step": 6380
    },
    {
      "epoch": 0.25800843045882316,
      "grad_norm": 3.6027603149414062,
      "learning_rate": 8.479520475392463e-05,
      "loss": 0.2039,
      "step": 6381
    },
    {
      "epoch": 0.2580488643140902,
      "grad_norm": 3.665100336074829,
      "learning_rate": 8.479062469990767e-05,
      "loss": 0.1421,
      "step": 6382
    },
    {
      "epoch": 0.2580892981693572,
      "grad_norm": 2.225731611251831,
      "learning_rate": 8.478604407991486e-05,
      "loss": 0.0943,
      "step": 6383
    },
    {
      "epoch": 0.2581297320246242,
      "grad_norm": 5.70697021484375,
      "learning_rate": 8.478146289402074e-05,
      "loss": 0.1902,
      "step": 6384
    },
    {
      "epoch": 0.2581701658798912,
      "grad_norm": 10.20688247680664,
      "learning_rate": 8.477688114229985e-05,
      "loss": 0.3306,
      "step": 6385
    },
    {
      "epoch": 0.25821059973515825,
      "grad_norm": 5.550793647766113,
      "learning_rate": 8.477229882482671e-05,
      "loss": 0.1043,
      "step": 6386
    },
    {
      "epoch": 0.2582510335904253,
      "grad_norm": 5.5182600021362305,
      "learning_rate": 8.476771594167585e-05,
      "loss": 0.1728,
      "step": 6387
    },
    {
      "epoch": 0.25829146744569226,
      "grad_norm": 7.05347204208374,
      "learning_rate": 8.476313249292187e-05,
      "loss": 0.1392,
      "step": 6388
    },
    {
      "epoch": 0.2583319013009593,
      "grad_norm": 4.397397041320801,
      "learning_rate": 8.475854847863929e-05,
      "loss": 0.096,
      "step": 6389
    },
    {
      "epoch": 0.2583723351562263,
      "grad_norm": 5.273245334625244,
      "learning_rate": 8.47539638989027e-05,
      "loss": 0.1418,
      "step": 6390
    },
    {
      "epoch": 0.25841276901149335,
      "grad_norm": 4.012918472290039,
      "learning_rate": 8.474937875378669e-05,
      "loss": 0.1406,
      "step": 6391
    },
    {
      "epoch": 0.2584532028667603,
      "grad_norm": 4.039473056793213,
      "learning_rate": 8.474479304336584e-05,
      "loss": 0.1272,
      "step": 6392
    },
    {
      "epoch": 0.25849363672202735,
      "grad_norm": 2.7828409671783447,
      "learning_rate": 8.474020676771476e-05,
      "loss": 0.0875,
      "step": 6393
    },
    {
      "epoch": 0.2585340705772944,
      "grad_norm": 3.8492186069488525,
      "learning_rate": 8.473561992690805e-05,
      "loss": 0.1361,
      "step": 6394
    },
    {
      "epoch": 0.25857450443256136,
      "grad_norm": 5.878203868865967,
      "learning_rate": 8.473103252102033e-05,
      "loss": 0.2268,
      "step": 6395
    },
    {
      "epoch": 0.2586149382878284,
      "grad_norm": 2.3049988746643066,
      "learning_rate": 8.472644455012623e-05,
      "loss": 0.0793,
      "step": 6396
    },
    {
      "epoch": 0.2586553721430954,
      "grad_norm": 4.549858093261719,
      "learning_rate": 8.47218560143004e-05,
      "loss": 0.0883,
      "step": 6397
    },
    {
      "epoch": 0.25869580599836245,
      "grad_norm": 3.8747036457061768,
      "learning_rate": 8.471726691361747e-05,
      "loss": 0.1501,
      "step": 6398
    },
    {
      "epoch": 0.2587362398536294,
      "grad_norm": 6.015817642211914,
      "learning_rate": 8.471267724815209e-05,
      "loss": 0.2054,
      "step": 6399
    },
    {
      "epoch": 0.25877667370889645,
      "grad_norm": 12.059225082397461,
      "learning_rate": 8.470808701797894e-05,
      "loss": 0.2836,
      "step": 6400
    },
    {
      "epoch": 0.2588171075641635,
      "grad_norm": 7.788859844207764,
      "learning_rate": 8.470349622317269e-05,
      "loss": 0.2607,
      "step": 6401
    },
    {
      "epoch": 0.2588575414194305,
      "grad_norm": 4.372074604034424,
      "learning_rate": 8.469890486380802e-05,
      "loss": 0.2265,
      "step": 6402
    },
    {
      "epoch": 0.2588979752746975,
      "grad_norm": 5.639162540435791,
      "learning_rate": 8.469431293995964e-05,
      "loss": 0.1593,
      "step": 6403
    },
    {
      "epoch": 0.2589384091299645,
      "grad_norm": 5.578627109527588,
      "learning_rate": 8.468972045170223e-05,
      "loss": 0.1215,
      "step": 6404
    },
    {
      "epoch": 0.25897884298523155,
      "grad_norm": 5.033603191375732,
      "learning_rate": 8.468512739911051e-05,
      "loss": 0.2522,
      "step": 6405
    },
    {
      "epoch": 0.2590192768404985,
      "grad_norm": 5.399876594543457,
      "learning_rate": 8.468053378225919e-05,
      "loss": 0.1563,
      "step": 6406
    },
    {
      "epoch": 0.25905971069576555,
      "grad_norm": 5.1479315757751465,
      "learning_rate": 8.467593960122302e-05,
      "loss": 0.0807,
      "step": 6407
    },
    {
      "epoch": 0.2591001445510326,
      "grad_norm": 2.96193265914917,
      "learning_rate": 8.467134485607671e-05,
      "loss": 0.1433,
      "step": 6408
    },
    {
      "epoch": 0.2591405784062996,
      "grad_norm": 8.882711410522461,
      "learning_rate": 8.466674954689504e-05,
      "loss": 0.2025,
      "step": 6409
    },
    {
      "epoch": 0.2591810122615666,
      "grad_norm": 6.4518914222717285,
      "learning_rate": 8.466215367375274e-05,
      "loss": 0.1386,
      "step": 6410
    },
    {
      "epoch": 0.2592214461168336,
      "grad_norm": 2.5367136001586914,
      "learning_rate": 8.465755723672459e-05,
      "loss": 0.1707,
      "step": 6411
    },
    {
      "epoch": 0.25926187997210065,
      "grad_norm": 6.0566182136535645,
      "learning_rate": 8.465296023588536e-05,
      "loss": 0.2651,
      "step": 6412
    },
    {
      "epoch": 0.2593023138273677,
      "grad_norm": 4.471607208251953,
      "learning_rate": 8.464836267130982e-05,
      "loss": 0.2542,
      "step": 6413
    },
    {
      "epoch": 0.25934274768263466,
      "grad_norm": 8.036651611328125,
      "learning_rate": 8.46437645430728e-05,
      "loss": 0.1938,
      "step": 6414
    },
    {
      "epoch": 0.2593831815379017,
      "grad_norm": 6.1860833168029785,
      "learning_rate": 8.463916585124908e-05,
      "loss": 0.3583,
      "step": 6415
    },
    {
      "epoch": 0.2594236153931687,
      "grad_norm": 2.4040839672088623,
      "learning_rate": 8.463456659591347e-05,
      "loss": 0.0842,
      "step": 6416
    },
    {
      "epoch": 0.2594640492484357,
      "grad_norm": 4.007444858551025,
      "learning_rate": 8.462996677714078e-05,
      "loss": 0.2408,
      "step": 6417
    },
    {
      "epoch": 0.2595044831037027,
      "grad_norm": 2.925011157989502,
      "learning_rate": 8.462536639500587e-05,
      "loss": 0.1108,
      "step": 6418
    },
    {
      "epoch": 0.25954491695896975,
      "grad_norm": 8.07170581817627,
      "learning_rate": 8.462076544958357e-05,
      "loss": 0.2668,
      "step": 6419
    },
    {
      "epoch": 0.2595853508142368,
      "grad_norm": 5.603867053985596,
      "learning_rate": 8.461616394094871e-05,
      "loss": 0.189,
      "step": 6420
    },
    {
      "epoch": 0.25962578466950376,
      "grad_norm": 3.7268362045288086,
      "learning_rate": 8.461156186917617e-05,
      "loss": 0.0929,
      "step": 6421
    },
    {
      "epoch": 0.2596662185247708,
      "grad_norm": 4.756371021270752,
      "learning_rate": 8.460695923434079e-05,
      "loss": 0.1305,
      "step": 6422
    },
    {
      "epoch": 0.2597066523800378,
      "grad_norm": 7.001537322998047,
      "learning_rate": 8.460235603651746e-05,
      "loss": 0.2624,
      "step": 6423
    },
    {
      "epoch": 0.25974708623530485,
      "grad_norm": 3.183443784713745,
      "learning_rate": 8.459775227578109e-05,
      "loss": 0.2155,
      "step": 6424
    },
    {
      "epoch": 0.2597875200905718,
      "grad_norm": 3.8235130310058594,
      "learning_rate": 8.459314795220652e-05,
      "loss": 0.2506,
      "step": 6425
    },
    {
      "epoch": 0.25982795394583885,
      "grad_norm": 2.616227149963379,
      "learning_rate": 8.458854306586872e-05,
      "loss": 0.1201,
      "step": 6426
    },
    {
      "epoch": 0.2598683878011059,
      "grad_norm": 1.2675552368164062,
      "learning_rate": 8.458393761684255e-05,
      "loss": 0.0644,
      "step": 6427
    },
    {
      "epoch": 0.25990882165637286,
      "grad_norm": 3.470249652862549,
      "learning_rate": 8.457933160520295e-05,
      "loss": 0.1924,
      "step": 6428
    },
    {
      "epoch": 0.2599492555116399,
      "grad_norm": 3.0673654079437256,
      "learning_rate": 8.457472503102484e-05,
      "loss": 0.1778,
      "step": 6429
    },
    {
      "epoch": 0.2599896893669069,
      "grad_norm": 5.870110511779785,
      "learning_rate": 8.457011789438319e-05,
      "loss": 0.3368,
      "step": 6430
    },
    {
      "epoch": 0.26003012322217395,
      "grad_norm": 2.752155303955078,
      "learning_rate": 8.456551019535293e-05,
      "loss": 0.1311,
      "step": 6431
    },
    {
      "epoch": 0.2600705570774409,
      "grad_norm": 4.4634833335876465,
      "learning_rate": 8.4560901934009e-05,
      "loss": 0.2769,
      "step": 6432
    },
    {
      "epoch": 0.26011099093270795,
      "grad_norm": 5.72117280960083,
      "learning_rate": 8.455629311042639e-05,
      "loss": 0.1678,
      "step": 6433
    },
    {
      "epoch": 0.260151424787975,
      "grad_norm": 3.4487555027008057,
      "learning_rate": 8.455168372468007e-05,
      "loss": 0.1195,
      "step": 6434
    },
    {
      "epoch": 0.260191858643242,
      "grad_norm": 2.938300132751465,
      "learning_rate": 8.454707377684503e-05,
      "loss": 0.1321,
      "step": 6435
    },
    {
      "epoch": 0.260232292498509,
      "grad_norm": 4.485689640045166,
      "learning_rate": 8.454246326699626e-05,
      "loss": 0.1388,
      "step": 6436
    },
    {
      "epoch": 0.260272726353776,
      "grad_norm": 11.9823637008667,
      "learning_rate": 8.453785219520878e-05,
      "loss": 0.2942,
      "step": 6437
    },
    {
      "epoch": 0.26031316020904305,
      "grad_norm": 5.374208927154541,
      "learning_rate": 8.453324056155758e-05,
      "loss": 0.3151,
      "step": 6438
    },
    {
      "epoch": 0.26035359406431,
      "grad_norm": 6.791523456573486,
      "learning_rate": 8.452862836611768e-05,
      "loss": 0.1854,
      "step": 6439
    },
    {
      "epoch": 0.26039402791957705,
      "grad_norm": 3.6393191814422607,
      "learning_rate": 8.452401560896415e-05,
      "loss": 0.1342,
      "step": 6440
    },
    {
      "epoch": 0.2604344617748441,
      "grad_norm": 3.012181282043457,
      "learning_rate": 8.4519402290172e-05,
      "loss": 0.1095,
      "step": 6441
    },
    {
      "epoch": 0.2604748956301111,
      "grad_norm": 5.776113033294678,
      "learning_rate": 8.451478840981626e-05,
      "loss": 0.1903,
      "step": 6442
    },
    {
      "epoch": 0.2605153294853781,
      "grad_norm": 2.5943546295166016,
      "learning_rate": 8.451017396797202e-05,
      "loss": 0.0931,
      "step": 6443
    },
    {
      "epoch": 0.2605557633406451,
      "grad_norm": 2.9449124336242676,
      "learning_rate": 8.450555896471434e-05,
      "loss": 0.1039,
      "step": 6444
    },
    {
      "epoch": 0.26059619719591215,
      "grad_norm": 2.590688943862915,
      "learning_rate": 8.450094340011829e-05,
      "loss": 0.1321,
      "step": 6445
    },
    {
      "epoch": 0.2606366310511792,
      "grad_norm": 5.73232364654541,
      "learning_rate": 8.449632727425899e-05,
      "loss": 0.1826,
      "step": 6446
    },
    {
      "epoch": 0.26067706490644615,
      "grad_norm": 5.102660655975342,
      "learning_rate": 8.449171058721151e-05,
      "loss": 0.2592,
      "step": 6447
    },
    {
      "epoch": 0.2607174987617132,
      "grad_norm": 6.164446830749512,
      "learning_rate": 8.448709333905092e-05,
      "loss": 0.2107,
      "step": 6448
    },
    {
      "epoch": 0.2607579326169802,
      "grad_norm": 5.607487678527832,
      "learning_rate": 8.448247552985237e-05,
      "loss": 0.1682,
      "step": 6449
    },
    {
      "epoch": 0.2607983664722472,
      "grad_norm": 3.105562686920166,
      "learning_rate": 8.447785715969102e-05,
      "loss": 0.1193,
      "step": 6450
    },
    {
      "epoch": 0.2608388003275142,
      "grad_norm": 3.8859353065490723,
      "learning_rate": 8.447323822864192e-05,
      "loss": 0.1914,
      "step": 6451
    },
    {
      "epoch": 0.26087923418278125,
      "grad_norm": 3.000746726989746,
      "learning_rate": 8.44686187367803e-05,
      "loss": 0.15,
      "step": 6452
    },
    {
      "epoch": 0.2609196680380483,
      "grad_norm": 7.3277716636657715,
      "learning_rate": 8.446399868418122e-05,
      "loss": 0.2542,
      "step": 6453
    },
    {
      "epoch": 0.26096010189331525,
      "grad_norm": 4.479053020477295,
      "learning_rate": 8.44593780709199e-05,
      "loss": 0.1764,
      "step": 6454
    },
    {
      "epoch": 0.2610005357485823,
      "grad_norm": 3.381807804107666,
      "learning_rate": 8.445475689707147e-05,
      "loss": 0.1947,
      "step": 6455
    },
    {
      "epoch": 0.2610409696038493,
      "grad_norm": 3.185697555541992,
      "learning_rate": 8.445013516271116e-05,
      "loss": 0.1445,
      "step": 6456
    },
    {
      "epoch": 0.26108140345911635,
      "grad_norm": 3.825044870376587,
      "learning_rate": 8.444551286791411e-05,
      "loss": 0.1587,
      "step": 6457
    },
    {
      "epoch": 0.2611218373143833,
      "grad_norm": 2.2519567012786865,
      "learning_rate": 8.444089001275554e-05,
      "loss": 0.0917,
      "step": 6458
    },
    {
      "epoch": 0.26116227116965035,
      "grad_norm": 2.907572031021118,
      "learning_rate": 8.443626659731064e-05,
      "loss": 0.159,
      "step": 6459
    },
    {
      "epoch": 0.2612027050249174,
      "grad_norm": 3.546018123626709,
      "learning_rate": 8.443164262165461e-05,
      "loss": 0.2036,
      "step": 6460
    },
    {
      "epoch": 0.26124313888018436,
      "grad_norm": 4.949718475341797,
      "learning_rate": 8.44270180858627e-05,
      "loss": 0.1441,
      "step": 6461
    },
    {
      "epoch": 0.2612835727354514,
      "grad_norm": 4.2386674880981445,
      "learning_rate": 8.442239299001016e-05,
      "loss": 0.1225,
      "step": 6462
    },
    {
      "epoch": 0.2613240065907184,
      "grad_norm": 2.96069598197937,
      "learning_rate": 8.44177673341722e-05,
      "loss": 0.0953,
      "step": 6463
    },
    {
      "epoch": 0.26136444044598545,
      "grad_norm": 6.37571382522583,
      "learning_rate": 8.441314111842407e-05,
      "loss": 0.2887,
      "step": 6464
    },
    {
      "epoch": 0.2614048743012524,
      "grad_norm": 6.25275182723999,
      "learning_rate": 8.440851434284103e-05,
      "loss": 0.2222,
      "step": 6465
    },
    {
      "epoch": 0.26144530815651945,
      "grad_norm": 3.5580968856811523,
      "learning_rate": 8.440388700749835e-05,
      "loss": 0.1493,
      "step": 6466
    },
    {
      "epoch": 0.2614857420117865,
      "grad_norm": 8.079176902770996,
      "learning_rate": 8.439925911247132e-05,
      "loss": 0.1547,
      "step": 6467
    },
    {
      "epoch": 0.26152617586705346,
      "grad_norm": 4.430794715881348,
      "learning_rate": 8.439463065783523e-05,
      "loss": 0.077,
      "step": 6468
    },
    {
      "epoch": 0.2615666097223205,
      "grad_norm": 4.564385414123535,
      "learning_rate": 8.439000164366536e-05,
      "loss": 0.1131,
      "step": 6469
    },
    {
      "epoch": 0.2616070435775875,
      "grad_norm": 4.047739505767822,
      "learning_rate": 8.438537207003701e-05,
      "loss": 0.2903,
      "step": 6470
    },
    {
      "epoch": 0.26164747743285455,
      "grad_norm": 3.771772623062134,
      "learning_rate": 8.43807419370255e-05,
      "loss": 0.1291,
      "step": 6471
    },
    {
      "epoch": 0.2616879112881215,
      "grad_norm": 7.2533955574035645,
      "learning_rate": 8.437611124470616e-05,
      "loss": 0.2869,
      "step": 6472
    },
    {
      "epoch": 0.26172834514338855,
      "grad_norm": 7.090485572814941,
      "learning_rate": 8.437147999315436e-05,
      "loss": 0.1909,
      "step": 6473
    },
    {
      "epoch": 0.2617687789986556,
      "grad_norm": 8.594810485839844,
      "learning_rate": 8.436684818244535e-05,
      "loss": 0.2963,
      "step": 6474
    },
    {
      "epoch": 0.2618092128539226,
      "grad_norm": 10.380483627319336,
      "learning_rate": 8.436221581265456e-05,
      "loss": 0.4436,
      "step": 6475
    },
    {
      "epoch": 0.2618496467091896,
      "grad_norm": 5.023477554321289,
      "learning_rate": 8.435758288385732e-05,
      "loss": 0.1768,
      "step": 6476
    },
    {
      "epoch": 0.2618900805644566,
      "grad_norm": 3.6216633319854736,
      "learning_rate": 8.435294939612898e-05,
      "loss": 0.1173,
      "step": 6477
    },
    {
      "epoch": 0.26193051441972365,
      "grad_norm": 4.181687831878662,
      "learning_rate": 8.434831534954497e-05,
      "loss": 0.1436,
      "step": 6478
    },
    {
      "epoch": 0.2619709482749906,
      "grad_norm": 3.5942511558532715,
      "learning_rate": 8.434368074418063e-05,
      "loss": 0.1774,
      "step": 6479
    },
    {
      "epoch": 0.26201138213025765,
      "grad_norm": 4.015344142913818,
      "learning_rate": 8.433904558011136e-05,
      "loss": 0.1306,
      "step": 6480
    },
    {
      "epoch": 0.2620518159855247,
      "grad_norm": 1.9899013042449951,
      "learning_rate": 8.43344098574126e-05,
      "loss": 0.0908,
      "step": 6481
    },
    {
      "epoch": 0.2620922498407917,
      "grad_norm": 7.040347099304199,
      "learning_rate": 8.432977357615973e-05,
      "loss": 0.1741,
      "step": 6482
    },
    {
      "epoch": 0.2621326836960587,
      "grad_norm": 5.5926690101623535,
      "learning_rate": 8.432513673642817e-05,
      "loss": 0.2538,
      "step": 6483
    },
    {
      "epoch": 0.2621731175513257,
      "grad_norm": 5.606401443481445,
      "learning_rate": 8.432049933829339e-05,
      "loss": 0.179,
      "step": 6484
    },
    {
      "epoch": 0.26221355140659275,
      "grad_norm": 2.6667726039886475,
      "learning_rate": 8.43158613818308e-05,
      "loss": 0.1405,
      "step": 6485
    },
    {
      "epoch": 0.2622539852618598,
      "grad_norm": 3.666436195373535,
      "learning_rate": 8.431122286711586e-05,
      "loss": 0.1345,
      "step": 6486
    },
    {
      "epoch": 0.26229441911712675,
      "grad_norm": 1.826880693435669,
      "learning_rate": 8.430658379422401e-05,
      "loss": 0.0768,
      "step": 6487
    },
    {
      "epoch": 0.2623348529723938,
      "grad_norm": 8.996906280517578,
      "learning_rate": 8.430194416323076e-05,
      "loss": 0.3116,
      "step": 6488
    },
    {
      "epoch": 0.2623752868276608,
      "grad_norm": 7.172107696533203,
      "learning_rate": 8.429730397421155e-05,
      "loss": 0.2038,
      "step": 6489
    },
    {
      "epoch": 0.2624157206829278,
      "grad_norm": 4.744250774383545,
      "learning_rate": 8.42926632272419e-05,
      "loss": 0.2677,
      "step": 6490
    },
    {
      "epoch": 0.2624561545381948,
      "grad_norm": 6.7520012855529785,
      "learning_rate": 8.428802192239726e-05,
      "loss": 0.1968,
      "step": 6491
    },
    {
      "epoch": 0.26249658839346185,
      "grad_norm": 3.461371898651123,
      "learning_rate": 8.428338005975318e-05,
      "loss": 0.1477,
      "step": 6492
    },
    {
      "epoch": 0.2625370222487289,
      "grad_norm": 1.8159852027893066,
      "learning_rate": 8.427873763938514e-05,
      "loss": 0.067,
      "step": 6493
    },
    {
      "epoch": 0.26257745610399585,
      "grad_norm": 9.266837120056152,
      "learning_rate": 8.42740946613687e-05,
      "loss": 0.2863,
      "step": 6494
    },
    {
      "epoch": 0.2626178899592629,
      "grad_norm": 3.8387975692749023,
      "learning_rate": 8.426945112577937e-05,
      "loss": 0.1482,
      "step": 6495
    },
    {
      "epoch": 0.2626583238145299,
      "grad_norm": 2.9947237968444824,
      "learning_rate": 8.426480703269269e-05,
      "loss": 0.1598,
      "step": 6496
    },
    {
      "epoch": 0.26269875766979695,
      "grad_norm": 13.141474723815918,
      "learning_rate": 8.426016238218419e-05,
      "loss": 0.2665,
      "step": 6497
    },
    {
      "epoch": 0.2627391915250639,
      "grad_norm": 8.485552787780762,
      "learning_rate": 8.425551717432948e-05,
      "loss": 0.2041,
      "step": 6498
    },
    {
      "epoch": 0.26277962538033095,
      "grad_norm": 6.097568511962891,
      "learning_rate": 8.425087140920409e-05,
      "loss": 0.1209,
      "step": 6499
    },
    {
      "epoch": 0.262820059235598,
      "grad_norm": 6.649581432342529,
      "learning_rate": 8.424622508688363e-05,
      "loss": 0.1078,
      "step": 6500
    },
    {
      "epoch": 0.26286049309086496,
      "grad_norm": 5.81403112411499,
      "learning_rate": 8.424157820744364e-05,
      "loss": 0.2087,
      "step": 6501
    },
    {
      "epoch": 0.262900926946132,
      "grad_norm": 4.567441940307617,
      "learning_rate": 8.423693077095975e-05,
      "loss": 0.1554,
      "step": 6502
    },
    {
      "epoch": 0.262941360801399,
      "grad_norm": 6.034025192260742,
      "learning_rate": 8.423228277750755e-05,
      "loss": 0.197,
      "step": 6503
    },
    {
      "epoch": 0.26298179465666605,
      "grad_norm": 2.8581862449645996,
      "learning_rate": 8.422763422716267e-05,
      "loss": 0.1916,
      "step": 6504
    },
    {
      "epoch": 0.263022228511933,
      "grad_norm": 2.978179931640625,
      "learning_rate": 8.42229851200007e-05,
      "loss": 0.1003,
      "step": 6505
    },
    {
      "epoch": 0.26306266236720005,
      "grad_norm": 3.627148151397705,
      "learning_rate": 8.421833545609733e-05,
      "loss": 0.2098,
      "step": 6506
    },
    {
      "epoch": 0.2631030962224671,
      "grad_norm": 3.9005444049835205,
      "learning_rate": 8.421368523552815e-05,
      "loss": 0.1233,
      "step": 6507
    },
    {
      "epoch": 0.2631435300777341,
      "grad_norm": 6.3803181648254395,
      "learning_rate": 8.420903445836882e-05,
      "loss": 0.3498,
      "step": 6508
    },
    {
      "epoch": 0.2631839639330011,
      "grad_norm": 5.617633819580078,
      "learning_rate": 8.420438312469501e-05,
      "loss": 0.1782,
      "step": 6509
    },
    {
      "epoch": 0.2632243977882681,
      "grad_norm": 2.999321937561035,
      "learning_rate": 8.419973123458238e-05,
      "loss": 0.1713,
      "step": 6510
    },
    {
      "epoch": 0.26326483164353515,
      "grad_norm": 7.433416843414307,
      "learning_rate": 8.419507878810662e-05,
      "loss": 0.2645,
      "step": 6511
    },
    {
      "epoch": 0.2633052654988021,
      "grad_norm": 4.668217182159424,
      "learning_rate": 8.419042578534339e-05,
      "loss": 0.1612,
      "step": 6512
    },
    {
      "epoch": 0.26334569935406915,
      "grad_norm": 3.0045735836029053,
      "learning_rate": 8.41857722263684e-05,
      "loss": 0.2459,
      "step": 6513
    },
    {
      "epoch": 0.2633861332093362,
      "grad_norm": 1.820125937461853,
      "learning_rate": 8.418111811125737e-05,
      "loss": 0.1569,
      "step": 6514
    },
    {
      "epoch": 0.2634265670646032,
      "grad_norm": 2.8025760650634766,
      "learning_rate": 8.4176463440086e-05,
      "loss": 0.1436,
      "step": 6515
    },
    {
      "epoch": 0.2634670009198702,
      "grad_norm": 2.2238306999206543,
      "learning_rate": 8.417180821293e-05,
      "loss": 0.1373,
      "step": 6516
    },
    {
      "epoch": 0.2635074347751372,
      "grad_norm": 5.5867919921875,
      "learning_rate": 8.416715242986514e-05,
      "loss": 0.1568,
      "step": 6517
    },
    {
      "epoch": 0.26354786863040425,
      "grad_norm": 4.6446123123168945,
      "learning_rate": 8.416249609096712e-05,
      "loss": 0.1215,
      "step": 6518
    },
    {
      "epoch": 0.2635883024856713,
      "grad_norm": 1.404708981513977,
      "learning_rate": 8.415783919631169e-05,
      "loss": 0.095,
      "step": 6519
    },
    {
      "epoch": 0.26362873634093825,
      "grad_norm": 6.0271406173706055,
      "learning_rate": 8.415318174597461e-05,
      "loss": 0.1576,
      "step": 6520
    },
    {
      "epoch": 0.2636691701962053,
      "grad_norm": 6.995720386505127,
      "learning_rate": 8.41485237400317e-05,
      "loss": 0.2199,
      "step": 6521
    },
    {
      "epoch": 0.2637096040514723,
      "grad_norm": 6.817748069763184,
      "learning_rate": 8.414386517855868e-05,
      "loss": 0.2366,
      "step": 6522
    },
    {
      "epoch": 0.2637500379067393,
      "grad_norm": 4.0690765380859375,
      "learning_rate": 8.413920606163134e-05,
      "loss": 0.1487,
      "step": 6523
    },
    {
      "epoch": 0.2637904717620063,
      "grad_norm": 3.969592332839966,
      "learning_rate": 8.41345463893255e-05,
      "loss": 0.2808,
      "step": 6524
    },
    {
      "epoch": 0.26383090561727335,
      "grad_norm": 4.069786071777344,
      "learning_rate": 8.412988616171694e-05,
      "loss": 0.1688,
      "step": 6525
    },
    {
      "epoch": 0.2638713394725404,
      "grad_norm": 8.43940258026123,
      "learning_rate": 8.412522537888149e-05,
      "loss": 0.1824,
      "step": 6526
    },
    {
      "epoch": 0.26391177332780735,
      "grad_norm": 2.263195753097534,
      "learning_rate": 8.412056404089496e-05,
      "loss": 0.1638,
      "step": 6527
    },
    {
      "epoch": 0.2639522071830744,
      "grad_norm": 2.8283069133758545,
      "learning_rate": 8.41159021478332e-05,
      "loss": 0.1456,
      "step": 6528
    },
    {
      "epoch": 0.2639926410383414,
      "grad_norm": 5.247934341430664,
      "learning_rate": 8.411123969977204e-05,
      "loss": 0.157,
      "step": 6529
    },
    {
      "epoch": 0.26403307489360844,
      "grad_norm": 3.966691493988037,
      "learning_rate": 8.41065766967873e-05,
      "loss": 0.179,
      "step": 6530
    },
    {
      "epoch": 0.2640735087488754,
      "grad_norm": 5.709972381591797,
      "learning_rate": 8.410191313895489e-05,
      "loss": 0.2276,
      "step": 6531
    },
    {
      "epoch": 0.26411394260414245,
      "grad_norm": 3.630922555923462,
      "learning_rate": 8.409724902635064e-05,
      "loss": 0.1604,
      "step": 6532
    },
    {
      "epoch": 0.2641543764594095,
      "grad_norm": 1.5789424180984497,
      "learning_rate": 8.409258435905043e-05,
      "loss": 0.058,
      "step": 6533
    },
    {
      "epoch": 0.26419481031467645,
      "grad_norm": 5.126832008361816,
      "learning_rate": 8.408791913713015e-05,
      "loss": 0.2515,
      "step": 6534
    },
    {
      "epoch": 0.2642352441699435,
      "grad_norm": 5.651684761047363,
      "learning_rate": 8.40832533606657e-05,
      "loss": 0.1203,
      "step": 6535
    },
    {
      "epoch": 0.2642756780252105,
      "grad_norm": 3.721672296524048,
      "learning_rate": 8.407858702973299e-05,
      "loss": 0.1965,
      "step": 6536
    },
    {
      "epoch": 0.26431611188047754,
      "grad_norm": 2.2916038036346436,
      "learning_rate": 8.40739201444079e-05,
      "loss": 0.1112,
      "step": 6537
    },
    {
      "epoch": 0.2643565457357445,
      "grad_norm": 3.3301517963409424,
      "learning_rate": 8.406925270476638e-05,
      "loss": 0.1123,
      "step": 6538
    },
    {
      "epoch": 0.26439697959101155,
      "grad_norm": 4.490497589111328,
      "learning_rate": 8.406458471088437e-05,
      "loss": 0.11,
      "step": 6539
    },
    {
      "epoch": 0.2644374134462786,
      "grad_norm": 4.2007269859313965,
      "learning_rate": 8.405991616283778e-05,
      "loss": 0.2107,
      "step": 6540
    },
    {
      "epoch": 0.2644778473015456,
      "grad_norm": 4.226531028747559,
      "learning_rate": 8.405524706070255e-05,
      "loss": 0.1342,
      "step": 6541
    },
    {
      "epoch": 0.2645182811568126,
      "grad_norm": 4.862766265869141,
      "learning_rate": 8.405057740455468e-05,
      "loss": 0.1249,
      "step": 6542
    },
    {
      "epoch": 0.2645587150120796,
      "grad_norm": 5.847499847412109,
      "learning_rate": 8.40459071944701e-05,
      "loss": 0.1532,
      "step": 6543
    },
    {
      "epoch": 0.26459914886734665,
      "grad_norm": 4.505615711212158,
      "learning_rate": 8.40412364305248e-05,
      "loss": 0.0901,
      "step": 6544
    },
    {
      "epoch": 0.2646395827226136,
      "grad_norm": 3.2695295810699463,
      "learning_rate": 8.403656511279477e-05,
      "loss": 0.1241,
      "step": 6545
    },
    {
      "epoch": 0.26468001657788065,
      "grad_norm": 2.6395678520202637,
      "learning_rate": 8.403189324135599e-05,
      "loss": 0.2737,
      "step": 6546
    },
    {
      "epoch": 0.2647204504331477,
      "grad_norm": 5.6780219078063965,
      "learning_rate": 8.402722081628447e-05,
      "loss": 0.1713,
      "step": 6547
    },
    {
      "epoch": 0.2647608842884147,
      "grad_norm": 7.018209934234619,
      "learning_rate": 8.402254783765622e-05,
      "loss": 0.2443,
      "step": 6548
    },
    {
      "epoch": 0.2648013181436817,
      "grad_norm": 5.417405128479004,
      "learning_rate": 8.401787430554726e-05,
      "loss": 0.1917,
      "step": 6549
    },
    {
      "epoch": 0.2648417519989487,
      "grad_norm": 6.50978946685791,
      "learning_rate": 8.401320022003362e-05,
      "loss": 0.2162,
      "step": 6550
    },
    {
      "epoch": 0.26488218585421575,
      "grad_norm": 6.596667766571045,
      "learning_rate": 8.400852558119133e-05,
      "loss": 0.2816,
      "step": 6551
    },
    {
      "epoch": 0.2649226197094828,
      "grad_norm": 4.556078910827637,
      "learning_rate": 8.400385038909646e-05,
      "loss": 0.1147,
      "step": 6552
    },
    {
      "epoch": 0.26496305356474975,
      "grad_norm": 5.520486831665039,
      "learning_rate": 8.399917464382503e-05,
      "loss": 0.1135,
      "step": 6553
    },
    {
      "epoch": 0.2650034874200168,
      "grad_norm": 3.928405284881592,
      "learning_rate": 8.399449834545313e-05,
      "loss": 0.2043,
      "step": 6554
    },
    {
      "epoch": 0.2650439212752838,
      "grad_norm": 3.1504178047180176,
      "learning_rate": 8.398982149405685e-05,
      "loss": 0.105,
      "step": 6555
    },
    {
      "epoch": 0.2650843551305508,
      "grad_norm": 2.864487409591675,
      "learning_rate": 8.398514408971222e-05,
      "loss": 0.1212,
      "step": 6556
    },
    {
      "epoch": 0.2651247889858178,
      "grad_norm": 5.958441734313965,
      "learning_rate": 8.39804661324954e-05,
      "loss": 0.1383,
      "step": 6557
    },
    {
      "epoch": 0.26516522284108485,
      "grad_norm": 4.321859836578369,
      "learning_rate": 8.397578762248243e-05,
      "loss": 0.1522,
      "step": 6558
    },
    {
      "epoch": 0.2652056566963519,
      "grad_norm": 1.879861831665039,
      "learning_rate": 8.397110855974947e-05,
      "loss": 0.1188,
      "step": 6559
    },
    {
      "epoch": 0.26524609055161885,
      "grad_norm": 3.717170476913452,
      "learning_rate": 8.39664289443726e-05,
      "loss": 0.2723,
      "step": 6560
    },
    {
      "epoch": 0.2652865244068859,
      "grad_norm": 4.7497639656066895,
      "learning_rate": 8.396174877642798e-05,
      "loss": 0.2185,
      "step": 6561
    },
    {
      "epoch": 0.2653269582621529,
      "grad_norm": 10.374393463134766,
      "learning_rate": 8.395706805599173e-05,
      "loss": 0.2596,
      "step": 6562
    },
    {
      "epoch": 0.26536739211741994,
      "grad_norm": 4.920714855194092,
      "learning_rate": 8.395238678313999e-05,
      "loss": 0.1864,
      "step": 6563
    },
    {
      "epoch": 0.2654078259726869,
      "grad_norm": 2.298659086227417,
      "learning_rate": 8.394770495794891e-05,
      "loss": 0.2215,
      "step": 6564
    },
    {
      "epoch": 0.26544825982795395,
      "grad_norm": 2.148195505142212,
      "learning_rate": 8.394302258049469e-05,
      "loss": 0.0792,
      "step": 6565
    },
    {
      "epoch": 0.265488693683221,
      "grad_norm": 2.573965311050415,
      "learning_rate": 8.393833965085347e-05,
      "loss": 0.2546,
      "step": 6566
    },
    {
      "epoch": 0.26552912753848795,
      "grad_norm": 2.622670888900757,
      "learning_rate": 8.393365616910145e-05,
      "loss": 0.1245,
      "step": 6567
    },
    {
      "epoch": 0.265569561393755,
      "grad_norm": 2.1678662300109863,
      "learning_rate": 8.39289721353148e-05,
      "loss": 0.1488,
      "step": 6568
    },
    {
      "epoch": 0.265609995249022,
      "grad_norm": 2.495394468307495,
      "learning_rate": 8.392428754956976e-05,
      "loss": 0.1709,
      "step": 6569
    },
    {
      "epoch": 0.26565042910428904,
      "grad_norm": 3.6703124046325684,
      "learning_rate": 8.39196024119425e-05,
      "loss": 0.1122,
      "step": 6570
    },
    {
      "epoch": 0.265690862959556,
      "grad_norm": 4.449604034423828,
      "learning_rate": 8.391491672250925e-05,
      "loss": 0.2264,
      "step": 6571
    },
    {
      "epoch": 0.26573129681482305,
      "grad_norm": 7.208311557769775,
      "learning_rate": 8.391023048134624e-05,
      "loss": 0.2274,
      "step": 6572
    },
    {
      "epoch": 0.2657717306700901,
      "grad_norm": 3.3646368980407715,
      "learning_rate": 8.39055436885297e-05,
      "loss": 0.1179,
      "step": 6573
    },
    {
      "epoch": 0.2658121645253571,
      "grad_norm": 3.414924383163452,
      "learning_rate": 8.390085634413588e-05,
      "loss": 0.1123,
      "step": 6574
    },
    {
      "epoch": 0.2658525983806241,
      "grad_norm": 4.110054016113281,
      "learning_rate": 8.389616844824103e-05,
      "loss": 0.1368,
      "step": 6575
    },
    {
      "epoch": 0.2658930322358911,
      "grad_norm": 3.7811076641082764,
      "learning_rate": 8.389148000092143e-05,
      "loss": 0.11,
      "step": 6576
    },
    {
      "epoch": 0.26593346609115814,
      "grad_norm": 3.9020886421203613,
      "learning_rate": 8.388679100225334e-05,
      "loss": 0.1784,
      "step": 6577
    },
    {
      "epoch": 0.2659738999464251,
      "grad_norm": 4.886922836303711,
      "learning_rate": 8.388210145231302e-05,
      "loss": 0.1216,
      "step": 6578
    },
    {
      "epoch": 0.26601433380169215,
      "grad_norm": 6.976899147033691,
      "learning_rate": 8.387741135117679e-05,
      "loss": 0.1608,
      "step": 6579
    },
    {
      "epoch": 0.2660547676569592,
      "grad_norm": 2.9632537364959717,
      "learning_rate": 8.387272069892094e-05,
      "loss": 0.1304,
      "step": 6580
    },
    {
      "epoch": 0.2660952015122262,
      "grad_norm": 5.112422943115234,
      "learning_rate": 8.386802949562177e-05,
      "loss": 0.2,
      "step": 6581
    },
    {
      "epoch": 0.2661356353674932,
      "grad_norm": 2.054858684539795,
      "learning_rate": 8.386333774135561e-05,
      "loss": 0.0589,
      "step": 6582
    },
    {
      "epoch": 0.2661760692227602,
      "grad_norm": 5.725131511688232,
      "learning_rate": 8.385864543619877e-05,
      "loss": 0.1647,
      "step": 6583
    },
    {
      "epoch": 0.26621650307802724,
      "grad_norm": 4.986160755157471,
      "learning_rate": 8.385395258022759e-05,
      "loss": 0.3018,
      "step": 6584
    },
    {
      "epoch": 0.2662569369332943,
      "grad_norm": 3.6309750080108643,
      "learning_rate": 8.384925917351842e-05,
      "loss": 0.13,
      "step": 6585
    },
    {
      "epoch": 0.26629737078856125,
      "grad_norm": 2.1312034130096436,
      "learning_rate": 8.384456521614761e-05,
      "loss": 0.1472,
      "step": 6586
    },
    {
      "epoch": 0.2663378046438283,
      "grad_norm": 8.192390441894531,
      "learning_rate": 8.383987070819154e-05,
      "loss": 0.1566,
      "step": 6587
    },
    {
      "epoch": 0.2663782384990953,
      "grad_norm": 3.709893226623535,
      "learning_rate": 8.383517564972653e-05,
      "loss": 0.2068,
      "step": 6588
    },
    {
      "epoch": 0.2664186723543623,
      "grad_norm": 6.357100486755371,
      "learning_rate": 8.383048004082902e-05,
      "loss": 0.1536,
      "step": 6589
    },
    {
      "epoch": 0.2664591062096293,
      "grad_norm": 9.251782417297363,
      "learning_rate": 8.382578388157536e-05,
      "loss": 0.259,
      "step": 6590
    },
    {
      "epoch": 0.26649954006489635,
      "grad_norm": 6.364524841308594,
      "learning_rate": 8.382108717204195e-05,
      "loss": 0.1661,
      "step": 6591
    },
    {
      "epoch": 0.2665399739201634,
      "grad_norm": 4.140508651733398,
      "learning_rate": 8.38163899123052e-05,
      "loss": 0.3332,
      "step": 6592
    },
    {
      "epoch": 0.26658040777543035,
      "grad_norm": 5.130396366119385,
      "learning_rate": 8.381169210244153e-05,
      "loss": 0.1297,
      "step": 6593
    },
    {
      "epoch": 0.2666208416306974,
      "grad_norm": 4.069079399108887,
      "learning_rate": 8.380699374252737e-05,
      "loss": 0.2063,
      "step": 6594
    },
    {
      "epoch": 0.2666612754859644,
      "grad_norm": 3.3463737964630127,
      "learning_rate": 8.380229483263913e-05,
      "loss": 0.1234,
      "step": 6595
    },
    {
      "epoch": 0.2667017093412314,
      "grad_norm": 3.121365785598755,
      "learning_rate": 8.379759537285328e-05,
      "loss": 0.3259,
      "step": 6596
    },
    {
      "epoch": 0.2667421431964984,
      "grad_norm": 3.143514394760132,
      "learning_rate": 8.379289536324626e-05,
      "loss": 0.1241,
      "step": 6597
    },
    {
      "epoch": 0.26678257705176545,
      "grad_norm": 1.6280158758163452,
      "learning_rate": 8.378819480389454e-05,
      "loss": 0.0781,
      "step": 6598
    },
    {
      "epoch": 0.2668230109070325,
      "grad_norm": 3.641113758087158,
      "learning_rate": 8.378349369487455e-05,
      "loss": 0.0979,
      "step": 6599
    },
    {
      "epoch": 0.26686344476229945,
      "grad_norm": 1.5974067449569702,
      "learning_rate": 8.377879203626283e-05,
      "loss": 0.0629,
      "step": 6600
    },
    {
      "epoch": 0.2669038786175665,
      "grad_norm": 2.053572654724121,
      "learning_rate": 8.377408982813579e-05,
      "loss": 0.0955,
      "step": 6601
    },
    {
      "epoch": 0.2669443124728335,
      "grad_norm": 3.7002804279327393,
      "learning_rate": 8.376938707056999e-05,
      "loss": 0.2416,
      "step": 6602
    },
    {
      "epoch": 0.26698474632810054,
      "grad_norm": 8.693068504333496,
      "learning_rate": 8.376468376364193e-05,
      "loss": 0.164,
      "step": 6603
    },
    {
      "epoch": 0.2670251801833675,
      "grad_norm": 3.3940460681915283,
      "learning_rate": 8.375997990742808e-05,
      "loss": 0.1395,
      "step": 6604
    },
    {
      "epoch": 0.26706561403863455,
      "grad_norm": 4.1176581382751465,
      "learning_rate": 8.3755275502005e-05,
      "loss": 0.1798,
      "step": 6605
    },
    {
      "epoch": 0.2671060478939016,
      "grad_norm": 3.1726737022399902,
      "learning_rate": 8.375057054744921e-05,
      "loss": 0.0952,
      "step": 6606
    },
    {
      "epoch": 0.26714648174916855,
      "grad_norm": 4.6025190353393555,
      "learning_rate": 8.374586504383724e-05,
      "loss": 0.1456,
      "step": 6607
    },
    {
      "epoch": 0.2671869156044356,
      "grad_norm": 4.858215808868408,
      "learning_rate": 8.374115899124568e-05,
      "loss": 0.1347,
      "step": 6608
    },
    {
      "epoch": 0.2672273494597026,
      "grad_norm": 6.650022029876709,
      "learning_rate": 8.373645238975104e-05,
      "loss": 0.1621,
      "step": 6609
    },
    {
      "epoch": 0.26726778331496964,
      "grad_norm": 4.232663154602051,
      "learning_rate": 8.37317452394299e-05,
      "loss": 0.1919,
      "step": 6610
    },
    {
      "epoch": 0.2673082171702366,
      "grad_norm": 6.808403968811035,
      "learning_rate": 8.372703754035885e-05,
      "loss": 0.1747,
      "step": 6611
    },
    {
      "epoch": 0.26734865102550365,
      "grad_norm": 9.129340171813965,
      "learning_rate": 8.372232929261448e-05,
      "loss": 0.1383,
      "step": 6612
    },
    {
      "epoch": 0.2673890848807707,
      "grad_norm": 1.4686884880065918,
      "learning_rate": 8.371762049627334e-05,
      "loss": 0.0541,
      "step": 6613
    },
    {
      "epoch": 0.2674295187360377,
      "grad_norm": 5.3172078132629395,
      "learning_rate": 8.371291115141209e-05,
      "loss": 0.1711,
      "step": 6614
    },
    {
      "epoch": 0.2674699525913047,
      "grad_norm": 1.6350047588348389,
      "learning_rate": 8.370820125810733e-05,
      "loss": 0.0823,
      "step": 6615
    },
    {
      "epoch": 0.2675103864465717,
      "grad_norm": 4.700094699859619,
      "learning_rate": 8.370349081643563e-05,
      "loss": 0.1098,
      "step": 6616
    },
    {
      "epoch": 0.26755082030183874,
      "grad_norm": 6.265729904174805,
      "learning_rate": 8.369877982647367e-05,
      "loss": 0.2921,
      "step": 6617
    },
    {
      "epoch": 0.2675912541571057,
      "grad_norm": 4.136383056640625,
      "learning_rate": 8.369406828829809e-05,
      "loss": 0.206,
      "step": 6618
    },
    {
      "epoch": 0.26763168801237275,
      "grad_norm": 4.891862392425537,
      "learning_rate": 8.36893562019855e-05,
      "loss": 0.2128,
      "step": 6619
    },
    {
      "epoch": 0.2676721218676398,
      "grad_norm": 6.812446117401123,
      "learning_rate": 8.36846435676126e-05,
      "loss": 0.2589,
      "step": 6620
    },
    {
      "epoch": 0.2677125557229068,
      "grad_norm": 3.122558832168579,
      "learning_rate": 8.367993038525603e-05,
      "loss": 0.1824,
      "step": 6621
    },
    {
      "epoch": 0.2677529895781738,
      "grad_norm": 4.192493438720703,
      "learning_rate": 8.367521665499248e-05,
      "loss": 0.1088,
      "step": 6622
    },
    {
      "epoch": 0.2677934234334408,
      "grad_norm": 5.159574031829834,
      "learning_rate": 8.36705023768986e-05,
      "loss": 0.2182,
      "step": 6623
    },
    {
      "epoch": 0.26783385728870784,
      "grad_norm": 5.709356784820557,
      "learning_rate": 8.366578755105111e-05,
      "loss": 0.1795,
      "step": 6624
    },
    {
      "epoch": 0.2678742911439749,
      "grad_norm": 5.803701877593994,
      "learning_rate": 8.366107217752672e-05,
      "loss": 0.2336,
      "step": 6625
    },
    {
      "epoch": 0.26791472499924185,
      "grad_norm": 2.842479705810547,
      "learning_rate": 8.365635625640212e-05,
      "loss": 0.1535,
      "step": 6626
    },
    {
      "epoch": 0.2679551588545089,
      "grad_norm": 8.77539348602295,
      "learning_rate": 8.365163978775403e-05,
      "loss": 0.3746,
      "step": 6627
    },
    {
      "epoch": 0.2679955927097759,
      "grad_norm": 6.282550811767578,
      "learning_rate": 8.364692277165918e-05,
      "loss": 0.1306,
      "step": 6628
    },
    {
      "epoch": 0.2680360265650429,
      "grad_norm": 6.026726722717285,
      "learning_rate": 8.36422052081943e-05,
      "loss": 0.1601,
      "step": 6629
    },
    {
      "epoch": 0.2680764604203099,
      "grad_norm": 2.6439599990844727,
      "learning_rate": 8.363748709743617e-05,
      "loss": 0.2097,
      "step": 6630
    },
    {
      "epoch": 0.26811689427557694,
      "grad_norm": 3.718888759613037,
      "learning_rate": 8.36327684394615e-05,
      "loss": 0.1553,
      "step": 6631
    },
    {
      "epoch": 0.268157328130844,
      "grad_norm": 5.948970317840576,
      "learning_rate": 8.362804923434709e-05,
      "loss": 0.1228,
      "step": 6632
    },
    {
      "epoch": 0.26819776198611095,
      "grad_norm": 2.3430135250091553,
      "learning_rate": 8.362332948216968e-05,
      "loss": 0.2042,
      "step": 6633
    },
    {
      "epoch": 0.268238195841378,
      "grad_norm": 2.8977715969085693,
      "learning_rate": 8.361860918300605e-05,
      "loss": 0.0938,
      "step": 6634
    },
    {
      "epoch": 0.268278629696645,
      "grad_norm": 6.015623569488525,
      "learning_rate": 8.361388833693304e-05,
      "loss": 0.2318,
      "step": 6635
    },
    {
      "epoch": 0.26831906355191204,
      "grad_norm": 3.8209362030029297,
      "learning_rate": 8.36091669440274e-05,
      "loss": 0.1284,
      "step": 6636
    },
    {
      "epoch": 0.268359497407179,
      "grad_norm": 5.317723274230957,
      "learning_rate": 8.360444500436597e-05,
      "loss": 0.167,
      "step": 6637
    },
    {
      "epoch": 0.26839993126244605,
      "grad_norm": 4.560892105102539,
      "learning_rate": 8.359972251802552e-05,
      "loss": 0.1182,
      "step": 6638
    },
    {
      "epoch": 0.2684403651177131,
      "grad_norm": 1.9216376543045044,
      "learning_rate": 8.359499948508291e-05,
      "loss": 0.1043,
      "step": 6639
    },
    {
      "epoch": 0.26848079897298005,
      "grad_norm": 3.5545313358306885,
      "learning_rate": 8.3590275905615e-05,
      "loss": 0.1981,
      "step": 6640
    },
    {
      "epoch": 0.2685212328282471,
      "grad_norm": 4.940046787261963,
      "learning_rate": 8.358555177969857e-05,
      "loss": 0.1473,
      "step": 6641
    },
    {
      "epoch": 0.2685616666835141,
      "grad_norm": 3.4543397426605225,
      "learning_rate": 8.358082710741053e-05,
      "loss": 0.1641,
      "step": 6642
    },
    {
      "epoch": 0.26860210053878114,
      "grad_norm": 5.4971418380737305,
      "learning_rate": 8.35761018888277e-05,
      "loss": 0.2024,
      "step": 6643
    },
    {
      "epoch": 0.2686425343940481,
      "grad_norm": 17.138687133789062,
      "learning_rate": 8.357137612402697e-05,
      "loss": 0.5021,
      "step": 6644
    },
    {
      "epoch": 0.26868296824931515,
      "grad_norm": 3.5056703090667725,
      "learning_rate": 8.356664981308522e-05,
      "loss": 0.1467,
      "step": 6645
    },
    {
      "epoch": 0.2687234021045822,
      "grad_norm": 6.707460880279541,
      "learning_rate": 8.356192295607934e-05,
      "loss": 0.2155,
      "step": 6646
    },
    {
      "epoch": 0.2687638359598492,
      "grad_norm": 5.155137062072754,
      "learning_rate": 8.355719555308622e-05,
      "loss": 0.2303,
      "step": 6647
    },
    {
      "epoch": 0.2688042698151162,
      "grad_norm": 2.9357731342315674,
      "learning_rate": 8.355246760418277e-05,
      "loss": 0.0866,
      "step": 6648
    },
    {
      "epoch": 0.2688447036703832,
      "grad_norm": 7.29972505569458,
      "learning_rate": 8.354773910944588e-05,
      "loss": 0.2833,
      "step": 6649
    },
    {
      "epoch": 0.26888513752565024,
      "grad_norm": 4.35799503326416,
      "learning_rate": 8.35430100689525e-05,
      "loss": 0.1768,
      "step": 6650
    },
    {
      "epoch": 0.2689255713809172,
      "grad_norm": 7.227541446685791,
      "learning_rate": 8.353828048277957e-05,
      "loss": 0.2469,
      "step": 6651
    },
    {
      "epoch": 0.26896600523618425,
      "grad_norm": 5.29514741897583,
      "learning_rate": 8.353355035100402e-05,
      "loss": 0.3047,
      "step": 6652
    },
    {
      "epoch": 0.2690064390914513,
      "grad_norm": 5.829686164855957,
      "learning_rate": 8.352881967370279e-05,
      "loss": 0.1513,
      "step": 6653
    },
    {
      "epoch": 0.2690468729467183,
      "grad_norm": 4.818840026855469,
      "learning_rate": 8.352408845095283e-05,
      "loss": 0.2807,
      "step": 6654
    },
    {
      "epoch": 0.2690873068019853,
      "grad_norm": 5.2545013427734375,
      "learning_rate": 8.351935668283115e-05,
      "loss": 0.2154,
      "step": 6655
    },
    {
      "epoch": 0.2691277406572523,
      "grad_norm": 2.9671027660369873,
      "learning_rate": 8.351462436941469e-05,
      "loss": 0.158,
      "step": 6656
    },
    {
      "epoch": 0.26916817451251934,
      "grad_norm": 5.487869739532471,
      "learning_rate": 8.350989151078046e-05,
      "loss": 0.0881,
      "step": 6657
    },
    {
      "epoch": 0.2692086083677864,
      "grad_norm": 4.169815540313721,
      "learning_rate": 8.350515810700541e-05,
      "loss": 0.1348,
      "step": 6658
    },
    {
      "epoch": 0.26924904222305335,
      "grad_norm": 2.261898994445801,
      "learning_rate": 8.350042415816661e-05,
      "loss": 0.1066,
      "step": 6659
    },
    {
      "epoch": 0.2692894760783204,
      "grad_norm": 2.4516499042510986,
      "learning_rate": 8.349568966434102e-05,
      "loss": 0.1044,
      "step": 6660
    },
    {
      "epoch": 0.2693299099335874,
      "grad_norm": 2.239933490753174,
      "learning_rate": 8.349095462560567e-05,
      "loss": 0.1095,
      "step": 6661
    },
    {
      "epoch": 0.2693703437888544,
      "grad_norm": 4.995827674865723,
      "learning_rate": 8.348621904203762e-05,
      "loss": 0.2587,
      "step": 6662
    },
    {
      "epoch": 0.2694107776441214,
      "grad_norm": 3.9526302814483643,
      "learning_rate": 8.348148291371387e-05,
      "loss": 0.1517,
      "step": 6663
    },
    {
      "epoch": 0.26945121149938844,
      "grad_norm": 4.4188761711120605,
      "learning_rate": 8.347674624071149e-05,
      "loss": 0.2291,
      "step": 6664
    },
    {
      "epoch": 0.2694916453546555,
      "grad_norm": 4.185996055603027,
      "learning_rate": 8.347200902310753e-05,
      "loss": 0.24,
      "step": 6665
    },
    {
      "epoch": 0.26953207920992245,
      "grad_norm": 4.599777698516846,
      "learning_rate": 8.346727126097904e-05,
      "loss": 0.1806,
      "step": 6666
    },
    {
      "epoch": 0.2695725130651895,
      "grad_norm": 6.624025821685791,
      "learning_rate": 8.346253295440312e-05,
      "loss": 0.2144,
      "step": 6667
    },
    {
      "epoch": 0.2696129469204565,
      "grad_norm": 3.7423858642578125,
      "learning_rate": 8.345779410345687e-05,
      "loss": 0.2625,
      "step": 6668
    },
    {
      "epoch": 0.26965338077572354,
      "grad_norm": 2.7171175479888916,
      "learning_rate": 8.345305470821732e-05,
      "loss": 0.1175,
      "step": 6669
    },
    {
      "epoch": 0.2696938146309905,
      "grad_norm": 5.327414035797119,
      "learning_rate": 8.344831476876161e-05,
      "loss": 0.1192,
      "step": 6670
    },
    {
      "epoch": 0.26973424848625754,
      "grad_norm": 3.4716527462005615,
      "learning_rate": 8.344357428516685e-05,
      "loss": 0.1002,
      "step": 6671
    },
    {
      "epoch": 0.2697746823415246,
      "grad_norm": 3.382917642593384,
      "learning_rate": 8.343883325751016e-05,
      "loss": 0.2696,
      "step": 6672
    },
    {
      "epoch": 0.26981511619679155,
      "grad_norm": 5.441958427429199,
      "learning_rate": 8.343409168586867e-05,
      "loss": 0.1752,
      "step": 6673
    },
    {
      "epoch": 0.2698555500520586,
      "grad_norm": 3.3768417835235596,
      "learning_rate": 8.34293495703195e-05,
      "loss": 0.092,
      "step": 6674
    },
    {
      "epoch": 0.2698959839073256,
      "grad_norm": 5.813772201538086,
      "learning_rate": 8.342460691093978e-05,
      "loss": 0.2326,
      "step": 6675
    },
    {
      "epoch": 0.26993641776259264,
      "grad_norm": 3.0050082206726074,
      "learning_rate": 8.34198637078067e-05,
      "loss": 0.089,
      "step": 6676
    },
    {
      "epoch": 0.2699768516178596,
      "grad_norm": 4.313435077667236,
      "learning_rate": 8.341511996099741e-05,
      "loss": 0.2665,
      "step": 6677
    },
    {
      "epoch": 0.27001728547312664,
      "grad_norm": 2.853736162185669,
      "learning_rate": 8.34103756705891e-05,
      "loss": 0.066,
      "step": 6678
    },
    {
      "epoch": 0.2700577193283937,
      "grad_norm": 5.7998552322387695,
      "learning_rate": 8.34056308366589e-05,
      "loss": 0.1508,
      "step": 6679
    },
    {
      "epoch": 0.2700981531836607,
      "grad_norm": 2.8775806427001953,
      "learning_rate": 8.340088545928404e-05,
      "loss": 0.1989,
      "step": 6680
    },
    {
      "epoch": 0.2701385870389277,
      "grad_norm": 2.0381646156311035,
      "learning_rate": 8.33961395385417e-05,
      "loss": 0.0942,
      "step": 6681
    },
    {
      "epoch": 0.2701790208941947,
      "grad_norm": 3.8013174533843994,
      "learning_rate": 8.339139307450911e-05,
      "loss": 0.155,
      "step": 6682
    },
    {
      "epoch": 0.27021945474946174,
      "grad_norm": 2.5843374729156494,
      "learning_rate": 8.338664606726347e-05,
      "loss": 0.1153,
      "step": 6683
    },
    {
      "epoch": 0.2702598886047287,
      "grad_norm": 4.373371124267578,
      "learning_rate": 8.338189851688201e-05,
      "loss": 0.2366,
      "step": 6684
    },
    {
      "epoch": 0.27030032245999575,
      "grad_norm": 2.953427791595459,
      "learning_rate": 8.337715042344194e-05,
      "loss": 0.2328,
      "step": 6685
    },
    {
      "epoch": 0.2703407563152628,
      "grad_norm": 4.133306980133057,
      "learning_rate": 8.337240178702052e-05,
      "loss": 0.1676,
      "step": 6686
    },
    {
      "epoch": 0.2703811901705298,
      "grad_norm": 5.411648273468018,
      "learning_rate": 8.3367652607695e-05,
      "loss": 0.1771,
      "step": 6687
    },
    {
      "epoch": 0.2704216240257968,
      "grad_norm": 4.436055660247803,
      "learning_rate": 8.336290288554268e-05,
      "loss": 0.1585,
      "step": 6688
    },
    {
      "epoch": 0.2704620578810638,
      "grad_norm": 4.031379222869873,
      "learning_rate": 8.335815262064075e-05,
      "loss": 0.2252,
      "step": 6689
    },
    {
      "epoch": 0.27050249173633084,
      "grad_norm": 6.782537460327148,
      "learning_rate": 8.335340181306654e-05,
      "loss": 0.1625,
      "step": 6690
    },
    {
      "epoch": 0.27054292559159787,
      "grad_norm": 3.9009714126586914,
      "learning_rate": 8.334865046289733e-05,
      "loss": 0.1869,
      "step": 6691
    },
    {
      "epoch": 0.27058335944686485,
      "grad_norm": 3.912127733230591,
      "learning_rate": 8.33438985702104e-05,
      "loss": 0.1963,
      "step": 6692
    }
  ],
  "logging_steps": 1,
  "max_steps": 24731,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 239,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.1390589456521626e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}