{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 200,
  "global_step": 668,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0029940119760479044,
      "grad_norm": 1.1737305073559647,
      "learning_rate": 9.999944704978835e-06,
      "loss": 0.1694,
      "step": 1
    },
    {
      "epoch": 0.005988023952095809,
      "grad_norm": 0.9070630834813923,
      "learning_rate": 9.999778821138357e-06,
      "loss": 0.1245,
      "step": 2
    },
    {
      "epoch": 0.008982035928143712,
      "grad_norm": 0.9772178731809653,
      "learning_rate": 9.999502352147583e-06,
      "loss": 0.1385,
      "step": 3
    },
    {
      "epoch": 0.011976047904191617,
      "grad_norm": 0.9038156971178649,
      "learning_rate": 9.999115304121459e-06,
      "loss": 0.125,
      "step": 4
    },
    {
      "epoch": 0.014970059880239521,
      "grad_norm": 0.7942048261791306,
      "learning_rate": 9.998617685620715e-06,
      "loss": 0.1044,
      "step": 5
    },
    {
      "epoch": 0.017964071856287425,
      "grad_norm": 0.7546598478303629,
      "learning_rate": 9.998009507651683e-06,
      "loss": 0.1013,
      "step": 6
    },
    {
      "epoch": 0.020958083832335328,
      "grad_norm": 0.8272935786162782,
      "learning_rate": 9.997290783666048e-06,
      "loss": 0.104,
      "step": 7
    },
    {
      "epoch": 0.023952095808383235,
      "grad_norm": 0.8600942614404201,
      "learning_rate": 9.996461529560553e-06,
      "loss": 0.1069,
      "step": 8
    },
    {
      "epoch": 0.02694610778443114,
      "grad_norm": 0.8753367007402617,
      "learning_rate": 9.995521763676645e-06,
      "loss": 0.1163,
      "step": 9
    },
    {
      "epoch": 0.029940119760479042,
      "grad_norm": 0.8808689624208605,
      "learning_rate": 9.994471506800078e-06,
      "loss": 0.1126,
      "step": 10
    },
    {
      "epoch": 0.03293413173652695,
      "grad_norm": 0.7794977897566551,
      "learning_rate": 9.993310782160439e-06,
      "loss": 0.1116,
      "step": 11
    },
    {
      "epoch": 0.03592814371257485,
      "grad_norm": 0.8104276990375525,
      "learning_rate": 9.992039615430648e-06,
      "loss": 0.0996,
      "step": 12
    },
    {
      "epoch": 0.038922155688622756,
      "grad_norm": 0.8852372440170901,
      "learning_rate": 9.99065803472638e-06,
      "loss": 0.1146,
      "step": 13
    },
    {
      "epoch": 0.041916167664670656,
      "grad_norm": 0.9103241611178924,
      "learning_rate": 9.989166070605447e-06,
      "loss": 0.1143,
      "step": 14
    },
    {
      "epoch": 0.04491017964071856,
      "grad_norm": 0.8983354352264371,
      "learning_rate": 9.98756375606713e-06,
      "loss": 0.1017,
      "step": 15
    },
    {
      "epoch": 0.04790419161676647,
      "grad_norm": 0.9271898582667116,
      "learning_rate": 9.985851126551428e-06,
      "loss": 0.1132,
      "step": 16
    },
    {
      "epoch": 0.05089820359281437,
      "grad_norm": 0.6709198199840029,
      "learning_rate": 9.9840282199383e-06,
      "loss": 0.0819,
      "step": 17
    },
    {
      "epoch": 0.05389221556886228,
      "grad_norm": 0.8704990912944199,
      "learning_rate": 9.982095076546806e-06,
      "loss": 0.1194,
      "step": 18
    },
    {
      "epoch": 0.05688622754491018,
      "grad_norm": 1.0799111154015777,
      "learning_rate": 9.980051739134235e-06,
      "loss": 0.1406,
      "step": 19
    },
    {
      "epoch": 0.059880239520958084,
      "grad_norm": 0.810563721477535,
      "learning_rate": 9.977898252895133e-06,
      "loss": 0.1168,
      "step": 20
    },
    {
      "epoch": 0.06287425149700598,
      "grad_norm": 0.7882155401025411,
      "learning_rate": 9.975634665460333e-06,
      "loss": 0.1117,
      "step": 21
    },
    {
      "epoch": 0.0658682634730539,
      "grad_norm": 0.8376166560940022,
      "learning_rate": 9.973261026895878e-06,
      "loss": 0.1215,
      "step": 22
    },
    {
      "epoch": 0.0688622754491018,
      "grad_norm": 0.7490393059908862,
      "learning_rate": 9.970777389701927e-06,
      "loss": 0.1099,
      "step": 23
    },
    {
      "epoch": 0.0718562874251497,
      "grad_norm": 0.6905895938913018,
      "learning_rate": 9.968183808811586e-06,
      "loss": 0.0938,
      "step": 24
    },
    {
      "epoch": 0.0748502994011976,
      "grad_norm": 0.8460971903640745,
      "learning_rate": 9.965480341589702e-06,
      "loss": 0.1127,
      "step": 25
    },
    {
      "epoch": 0.07784431137724551,
      "grad_norm": 0.8200670758510404,
      "learning_rate": 9.962667047831585e-06,
      "loss": 0.1119,
      "step": 26
    },
    {
      "epoch": 0.08083832335329341,
      "grad_norm": 0.7802991509316722,
      "learning_rate": 9.95974398976169e-06,
      "loss": 0.1162,
      "step": 27
    },
    {
      "epoch": 0.08383233532934131,
      "grad_norm": 0.7720026369905891,
      "learning_rate": 9.95671123203224e-06,
      "loss": 0.1095,
      "step": 28
    },
    {
      "epoch": 0.08682634730538923,
      "grad_norm": 0.8487594806503318,
      "learning_rate": 9.953568841721796e-06,
      "loss": 0.1149,
      "step": 29
    },
    {
      "epoch": 0.08982035928143713,
      "grad_norm": 0.741296248758953,
      "learning_rate": 9.950316888333775e-06,
      "loss": 0.1116,
      "step": 30
    },
    {
      "epoch": 0.09281437125748503,
      "grad_norm": 0.6527134258072538,
      "learning_rate": 9.946955443794908e-06,
      "loss": 0.0986,
      "step": 31
    },
    {
      "epoch": 0.09580838323353294,
      "grad_norm": 0.859342277292467,
      "learning_rate": 9.943484582453653e-06,
      "loss": 0.1071,
      "step": 32
    },
    {
      "epoch": 0.09880239520958084,
      "grad_norm": 0.837311693290727,
      "learning_rate": 9.939904381078553e-06,
      "loss": 0.0967,
      "step": 33
    },
    {
      "epoch": 0.10179640718562874,
      "grad_norm": 0.8692257518788851,
      "learning_rate": 9.93621491885653e-06,
      "loss": 0.1111,
      "step": 34
    },
    {
      "epoch": 0.10479041916167664,
      "grad_norm": 0.7908613920309305,
      "learning_rate": 9.932416277391144e-06,
      "loss": 0.0996,
      "step": 35
    },
    {
      "epoch": 0.10778443113772455,
      "grad_norm": 0.769749698046481,
      "learning_rate": 9.928508540700775e-06,
      "loss": 0.1127,
      "step": 36
    },
    {
      "epoch": 0.11077844311377245,
      "grad_norm": 0.9941289625268535,
      "learning_rate": 9.924491795216777e-06,
      "loss": 0.137,
      "step": 37
    },
    {
      "epoch": 0.11377245508982035,
      "grad_norm": 0.8186071562643454,
      "learning_rate": 9.920366129781564e-06,
      "loss": 0.1038,
      "step": 38
    },
    {
      "epoch": 0.11676646706586827,
      "grad_norm": 0.6497158875788898,
      "learning_rate": 9.916131635646635e-06,
      "loss": 0.0841,
      "step": 39
    },
    {
      "epoch": 0.11976047904191617,
      "grad_norm": 0.8630253759556468,
      "learning_rate": 9.91178840647057e-06,
      "loss": 0.1084,
      "step": 40
    },
    {
      "epoch": 0.12275449101796407,
      "grad_norm": 0.7483230419615318,
      "learning_rate": 9.907336538316946e-06,
      "loss": 0.1055,
      "step": 41
    },
    {
      "epoch": 0.12574850299401197,
      "grad_norm": 0.5991149716895152,
      "learning_rate": 9.902776129652223e-06,
      "loss": 0.0763,
      "step": 42
    },
    {
      "epoch": 0.12874251497005987,
      "grad_norm": 0.7467680937848625,
      "learning_rate": 9.898107281343557e-06,
      "loss": 0.0933,
      "step": 43
    },
    {
      "epoch": 0.1317365269461078,
      "grad_norm": 0.6750329333250812,
      "learning_rate": 9.893330096656576e-06,
      "loss": 0.08,
      "step": 44
    },
    {
      "epoch": 0.1347305389221557,
      "grad_norm": 0.801668909875063,
      "learning_rate": 9.888444681253087e-06,
      "loss": 0.1019,
      "step": 45
    },
    {
      "epoch": 0.1377245508982036,
      "grad_norm": 0.8391004526525874,
      "learning_rate": 9.883451143188753e-06,
      "loss": 0.1078,
      "step": 46
    },
    {
      "epoch": 0.1407185628742515,
      "grad_norm": 0.7808417335170956,
      "learning_rate": 9.878349592910694e-06,
      "loss": 0.0988,
      "step": 47
    },
    {
      "epoch": 0.1437125748502994,
      "grad_norm": 0.8475076101575896,
      "learning_rate": 9.873140143255035e-06,
      "loss": 0.1131,
      "step": 48
    },
    {
      "epoch": 0.1467065868263473,
      "grad_norm": 0.8402794693155737,
      "learning_rate": 9.867822909444435e-06,
      "loss": 0.1109,
      "step": 49
    },
    {
      "epoch": 0.1497005988023952,
      "grad_norm": 0.8099360357526537,
      "learning_rate": 9.862398009085511e-06,
      "loss": 0.102,
      "step": 50
    },
    {
      "epoch": 0.15269461077844312,
      "grad_norm": 0.8246121328160928,
      "learning_rate": 9.856865562166256e-06,
      "loss": 0.1115,
      "step": 51
    },
    {
      "epoch": 0.15568862275449102,
      "grad_norm": 0.7171176918289516,
      "learning_rate": 9.851225691053382e-06,
      "loss": 0.0946,
      "step": 52
    },
    {
      "epoch": 0.15868263473053892,
      "grad_norm": 0.8416063513163756,
      "learning_rate": 9.8454785204896e-06,
      "loss": 0.1174,
      "step": 53
    },
    {
      "epoch": 0.16167664670658682,
      "grad_norm": 0.7215731678860191,
      "learning_rate": 9.83962417759088e-06,
      "loss": 0.0957,
      "step": 54
    },
    {
      "epoch": 0.16467065868263472,
      "grad_norm": 0.8063708031346318,
      "learning_rate": 9.833662791843628e-06,
      "loss": 0.1113,
      "step": 55
    },
    {
      "epoch": 0.16766467065868262,
      "grad_norm": 0.7602890495702399,
      "learning_rate": 9.827594495101824e-06,
      "loss": 0.1072,
      "step": 56
    },
    {
      "epoch": 0.17065868263473055,
      "grad_norm": 0.7952079097020819,
      "learning_rate": 9.821419421584108e-06,
      "loss": 0.108,
      "step": 57
    },
    {
      "epoch": 0.17365269461077845,
      "grad_norm": 1.0057421871754313,
      "learning_rate": 9.815137707870806e-06,
      "loss": 0.1492,
      "step": 58
    },
    {
      "epoch": 0.17664670658682635,
      "grad_norm": 0.8414859942979821,
      "learning_rate": 9.808749492900917e-06,
      "loss": 0.1119,
      "step": 59
    },
    {
      "epoch": 0.17964071856287425,
      "grad_norm": 0.8063161427732357,
      "learning_rate": 9.802254917969033e-06,
      "loss": 0.1,
      "step": 60
    },
    {
      "epoch": 0.18263473053892215,
      "grad_norm": 0.9419742146461794,
      "learning_rate": 9.795654126722218e-06,
      "loss": 0.1222,
      "step": 61
    },
    {
      "epoch": 0.18562874251497005,
      "grad_norm": 0.7241660570982316,
      "learning_rate": 9.788947265156828e-06,
      "loss": 0.0972,
      "step": 62
    },
    {
      "epoch": 0.18862275449101795,
      "grad_norm": 0.7068246024430127,
      "learning_rate": 9.782134481615282e-06,
      "loss": 0.0885,
      "step": 63
    },
    {
      "epoch": 0.19161676646706588,
      "grad_norm": 0.7694582437308338,
      "learning_rate": 9.775215926782788e-06,
      "loss": 0.1104,
      "step": 64
    },
    {
      "epoch": 0.19461077844311378,
      "grad_norm": 0.7791001332339761,
      "learning_rate": 9.768191753683997e-06,
      "loss": 0.101,
      "step": 65
    },
    {
      "epoch": 0.19760479041916168,
      "grad_norm": 0.7284404135209178,
      "learning_rate": 9.761062117679632e-06,
      "loss": 0.1034,
      "step": 66
    },
    {
      "epoch": 0.20059880239520958,
      "grad_norm": 0.6713727893279953,
      "learning_rate": 9.75382717646304e-06,
      "loss": 0.0941,
      "step": 67
    },
    {
      "epoch": 0.20359281437125748,
      "grad_norm": 0.7316045956646867,
      "learning_rate": 9.746487090056712e-06,
      "loss": 0.0978,
      "step": 68
    },
    {
      "epoch": 0.20658682634730538,
      "grad_norm": 0.7208626813817461,
      "learning_rate": 9.739042020808746e-06,
      "loss": 0.1001,
      "step": 69
    },
    {
      "epoch": 0.20958083832335328,
      "grad_norm": 0.7693532148854401,
      "learning_rate": 9.73149213338924e-06,
      "loss": 0.1163,
      "step": 70
    },
    {
      "epoch": 0.2125748502994012,
      "grad_norm": 0.8126412735258668,
      "learning_rate": 9.72383759478667e-06,
      "loss": 0.1156,
      "step": 71
    },
    {
      "epoch": 0.2155688622754491,
      "grad_norm": 0.7638359957229056,
      "learning_rate": 9.71607857430419e-06,
      "loss": 0.0866,
      "step": 72
    },
    {
      "epoch": 0.218562874251497,
      "grad_norm": 0.8733658571090064,
      "learning_rate": 9.708215243555875e-06,
      "loss": 0.1183,
      "step": 73
    },
    {
      "epoch": 0.2215568862275449,
      "grad_norm": 0.7917814269103062,
      "learning_rate": 9.700247776462944e-06,
      "loss": 0.1019,
      "step": 74
    },
    {
      "epoch": 0.2245508982035928,
      "grad_norm": 0.786192546818783,
      "learning_rate": 9.6921763492499e-06,
      "loss": 0.1034,
      "step": 75
    },
    {
      "epoch": 0.2275449101796407,
      "grad_norm": 0.6041063918492866,
      "learning_rate": 9.68400114044064e-06,
      "loss": 0.0719,
      "step": 76
    },
    {
      "epoch": 0.23053892215568864,
      "grad_norm": 0.7391035633177765,
      "learning_rate": 9.6757223308545e-06,
      "loss": 0.0908,
      "step": 77
    },
    {
      "epoch": 0.23353293413173654,
      "grad_norm": 0.8254079313682828,
      "learning_rate": 9.667340103602263e-06,
      "loss": 0.1159,
      "step": 78
    },
    {
      "epoch": 0.23652694610778444,
      "grad_norm": 0.7368291427945058,
      "learning_rate": 9.658854644082099e-06,
      "loss": 0.0998,
      "step": 79
    },
    {
      "epoch": 0.23952095808383234,
      "grad_norm": 1.032688174427545,
      "learning_rate": 9.650266139975474e-06,
      "loss": 0.1354,
      "step": 80
    },
    {
      "epoch": 0.24251497005988024,
      "grad_norm": 0.7274017500137487,
      "learning_rate": 9.641574781242999e-06,
      "loss": 0.1019,
      "step": 81
    },
    {
      "epoch": 0.24550898203592814,
      "grad_norm": 0.7275195138322714,
      "learning_rate": 9.632780760120217e-06,
      "loss": 0.0953,
      "step": 82
    },
    {
      "epoch": 0.24850299401197604,
      "grad_norm": 0.6616474775147956,
      "learning_rate": 9.62388427111336e-06,
      "loss": 0.0892,
      "step": 83
    },
    {
      "epoch": 0.25149700598802394,
      "grad_norm": 0.654373699832686,
      "learning_rate": 9.614885510995047e-06,
      "loss": 0.0898,
      "step": 84
    },
    {
      "epoch": 0.25449101796407186,
      "grad_norm": 0.6907727824064511,
      "learning_rate": 9.605784678799934e-06,
      "loss": 0.0928,
      "step": 85
    },
    {
      "epoch": 0.25748502994011974,
      "grad_norm": 0.7923438288032203,
      "learning_rate": 9.596581975820304e-06,
      "loss": 0.0932,
      "step": 86
    },
    {
      "epoch": 0.26047904191616766,
      "grad_norm": 0.7438727422523921,
      "learning_rate": 9.587277605601617e-06,
      "loss": 0.0994,
      "step": 87
    },
    {
      "epoch": 0.2634730538922156,
      "grad_norm": 0.8776780083489998,
      "learning_rate": 9.577871773938013e-06,
      "loss": 0.1121,
      "step": 88
    },
    {
      "epoch": 0.26646706586826346,
      "grad_norm": 0.8263224583454534,
      "learning_rate": 9.568364688867757e-06,
      "loss": 0.1216,
      "step": 89
    },
    {
      "epoch": 0.2694610778443114,
      "grad_norm": 0.6408025460626702,
      "learning_rate": 9.558756560668637e-06,
      "loss": 0.0835,
      "step": 90
    },
    {
      "epoch": 0.27245508982035926,
      "grad_norm": 0.7911053510264364,
      "learning_rate": 9.549047601853313e-06,
      "loss": 0.1121,
      "step": 91
    },
    {
      "epoch": 0.2754491017964072,
      "grad_norm": 0.7581113749641647,
      "learning_rate": 9.539238027164618e-06,
      "loss": 0.1071,
      "step": 92
    },
    {
      "epoch": 0.27844311377245506,
      "grad_norm": 0.7025350694562589,
      "learning_rate": 9.52932805357081e-06,
      "loss": 0.0963,
      "step": 93
    },
    {
      "epoch": 0.281437125748503,
      "grad_norm": 0.7130944866262018,
      "learning_rate": 9.519317900260769e-06,
      "loss": 0.1027,
      "step": 94
    },
    {
      "epoch": 0.2844311377245509,
      "grad_norm": 0.7238604132991617,
      "learning_rate": 9.509207788639148e-06,
      "loss": 0.0991,
      "step": 95
    },
    {
      "epoch": 0.2874251497005988,
      "grad_norm": 0.7328039174882758,
      "learning_rate": 9.498997942321484e-06,
      "loss": 0.1074,
      "step": 96
    },
    {
      "epoch": 0.2904191616766467,
      "grad_norm": 0.6996428555055889,
      "learning_rate": 9.488688587129243e-06,
      "loss": 0.098,
      "step": 97
    },
    {
      "epoch": 0.2934131736526946,
      "grad_norm": 0.7452905014949691,
      "learning_rate": 9.47827995108483e-06,
      "loss": 0.0845,
      "step": 98
    },
    {
      "epoch": 0.2964071856287425,
      "grad_norm": 0.8296019734045073,
      "learning_rate": 9.467772264406545e-06,
      "loss": 0.1057,
      "step": 99
    },
    {
      "epoch": 0.2994011976047904,
      "grad_norm": 0.6663332474591662,
      "learning_rate": 9.457165759503492e-06,
      "loss": 0.085,
      "step": 100
    },
    {
      "epoch": 0.3023952095808383,
      "grad_norm": 0.707983412401489,
      "learning_rate": 9.446460670970436e-06,
      "loss": 0.0781,
      "step": 101
    },
    {
      "epoch": 0.30538922155688625,
      "grad_norm": 0.6468823971100458,
      "learning_rate": 9.435657235582616e-06,
      "loss": 0.0791,
      "step": 102
    },
    {
      "epoch": 0.3083832335329341,
      "grad_norm": 0.8817479806504642,
      "learning_rate": 9.424755692290507e-06,
      "loss": 0.11,
      "step": 103
    },
    {
      "epoch": 0.31137724550898205,
      "grad_norm": 0.7925188633737331,
      "learning_rate": 9.413756282214538e-06,
      "loss": 0.1072,
      "step": 104
    },
    {
      "epoch": 0.3143712574850299,
      "grad_norm": 0.6903771802627102,
      "learning_rate": 9.402659248639749e-06,
      "loss": 0.0964,
      "step": 105
    },
    {
      "epoch": 0.31736526946107785,
      "grad_norm": 0.7803447350939394,
      "learning_rate": 9.391464837010428e-06,
      "loss": 0.1031,
      "step": 106
    },
    {
      "epoch": 0.3203592814371258,
      "grad_norm": 0.8805435327575107,
      "learning_rate": 9.380173294924661e-06,
      "loss": 0.1131,
      "step": 107
    },
    {
      "epoch": 0.32335329341317365,
      "grad_norm": 0.7787638031143439,
      "learning_rate": 9.368784872128877e-06,
      "loss": 0.1102,
      "step": 108
    },
    {
      "epoch": 0.3263473053892216,
      "grad_norm": 0.8543226039637506,
      "learning_rate": 9.357299820512305e-06,
      "loss": 0.1029,
      "step": 109
    },
    {
      "epoch": 0.32934131736526945,
      "grad_norm": 0.9123789774717929,
      "learning_rate": 9.345718394101412e-06,
      "loss": 0.1183,
      "step": 110
    },
    {
      "epoch": 0.3323353293413174,
      "grad_norm": 0.7962160941480667,
      "learning_rate": 9.334040849054288e-06,
      "loss": 0.118,
      "step": 111
    },
    {
      "epoch": 0.33532934131736525,
      "grad_norm": 0.806096839366154,
      "learning_rate": 9.322267443654974e-06,
      "loss": 0.1058,
      "step": 112
    },
    {
      "epoch": 0.3383233532934132,
      "grad_norm": 0.8277136365187742,
      "learning_rate": 9.310398438307747e-06,
      "loss": 0.128,
      "step": 113
    },
    {
      "epoch": 0.3413173652694611,
      "grad_norm": 0.7508620776978971,
      "learning_rate": 9.29843409553137e-06,
      "loss": 0.1,
      "step": 114
    },
    {
      "epoch": 0.344311377245509,
      "grad_norm": 0.7379992842393038,
      "learning_rate": 9.286374679953278e-06,
      "loss": 0.0994,
      "step": 115
    },
    {
      "epoch": 0.3473053892215569,
      "grad_norm": 0.6384888477835629,
      "learning_rate": 9.274220458303727e-06,
      "loss": 0.0899,
      "step": 116
    },
    {
      "epoch": 0.3502994011976048,
      "grad_norm": 0.7840297852456504,
      "learning_rate": 9.261971699409893e-06,
      "loss": 0.1155,
      "step": 117
    },
    {
      "epoch": 0.3532934131736527,
      "grad_norm": 0.7872703913524547,
      "learning_rate": 9.249628674189928e-06,
      "loss": 0.1033,
      "step": 118
    },
    {
      "epoch": 0.3562874251497006,
      "grad_norm": 0.7355174642641895,
      "learning_rate": 9.237191655646972e-06,
      "loss": 0.1068,
      "step": 119
    },
    {
      "epoch": 0.3592814371257485,
      "grad_norm": 0.781796026193897,
      "learning_rate": 9.224660918863104e-06,
      "loss": 0.1052,
      "step": 120
    },
    {
      "epoch": 0.36227544910179643,
      "grad_norm": 0.7536970037817883,
      "learning_rate": 9.212036740993265e-06,
      "loss": 0.1118,
      "step": 121
    },
    {
      "epoch": 0.3652694610778443,
      "grad_norm": 0.7356018566168909,
      "learning_rate": 9.199319401259132e-06,
      "loss": 0.0919,
      "step": 122
    },
    {
      "epoch": 0.36826347305389223,
      "grad_norm": 0.7184252762290236,
      "learning_rate": 9.186509180942928e-06,
      "loss": 0.0872,
      "step": 123
    },
    {
      "epoch": 0.3712574850299401,
      "grad_norm": 0.7238069108930196,
      "learning_rate": 9.173606363381218e-06,
      "loss": 0.1071,
      "step": 124
    },
    {
      "epoch": 0.37425149700598803,
      "grad_norm": 0.7383242350634773,
      "learning_rate": 9.16061123395863e-06,
      "loss": 0.0945,
      "step": 125
    },
    {
      "epoch": 0.3772455089820359,
      "grad_norm": 0.721334844167743,
      "learning_rate": 9.147524080101543e-06,
      "loss": 0.0937,
      "step": 126
    },
    {
      "epoch": 0.38023952095808383,
      "grad_norm": 0.8035538049784778,
      "learning_rate": 9.134345191271742e-06,
      "loss": 0.1089,
      "step": 127
    },
    {
      "epoch": 0.38323353293413176,
      "grad_norm": 0.7954136412560865,
      "learning_rate": 9.121074858959997e-06,
      "loss": 0.1149,
      "step": 128
    },
    {
      "epoch": 0.38622754491017963,
      "grad_norm": 0.7640052759190399,
      "learning_rate": 9.107713376679634e-06,
      "loss": 0.1065,
      "step": 129
    },
    {
      "epoch": 0.38922155688622756,
      "grad_norm": 0.8073547509330787,
      "learning_rate": 9.094261039960028e-06,
      "loss": 0.1038,
      "step": 130
    },
    {
      "epoch": 0.39221556886227543,
      "grad_norm": 0.8401650327789763,
      "learning_rate": 9.08071814634008e-06,
      "loss": 0.1116,
      "step": 131
    },
    {
      "epoch": 0.39520958083832336,
      "grad_norm": 0.7593900610877398,
      "learning_rate": 9.067084995361623e-06,
      "loss": 0.1039,
      "step": 132
    },
    {
      "epoch": 0.39820359281437123,
      "grad_norm": 0.8027801149072901,
      "learning_rate": 9.053361888562807e-06,
      "loss": 0.115,
      "step": 133
    },
    {
      "epoch": 0.40119760479041916,
      "grad_norm": 0.8190861042528541,
      "learning_rate": 9.039549129471423e-06,
      "loss": 0.1075,
      "step": 134
    },
    {
      "epoch": 0.4041916167664671,
      "grad_norm": 0.6465670646292982,
      "learning_rate": 9.025647023598196e-06,
      "loss": 0.092,
      "step": 135
    },
    {
      "epoch": 0.40718562874251496,
      "grad_norm": 0.6888140864521026,
      "learning_rate": 9.011655878430018e-06,
      "loss": 0.0894,
      "step": 136
    },
    {
      "epoch": 0.4101796407185629,
      "grad_norm": 0.9139222753582523,
      "learning_rate": 8.99757600342316e-06,
      "loss": 0.1304,
      "step": 137
    },
    {
      "epoch": 0.41317365269461076,
      "grad_norm": 0.803957797130848,
      "learning_rate": 8.983407709996415e-06,
      "loss": 0.0993,
      "step": 138
    },
    {
      "epoch": 0.4161676646706587,
      "grad_norm": 0.712965361431581,
      "learning_rate": 8.969151311524215e-06,
      "loss": 0.0931,
      "step": 139
    },
    {
      "epoch": 0.41916167664670656,
      "grad_norm": 0.823005134688049,
      "learning_rate": 8.954807123329703e-06,
      "loss": 0.1145,
      "step": 140
    },
    {
      "epoch": 0.4221556886227545,
      "grad_norm": 0.696900815606486,
      "learning_rate": 8.940375462677758e-06,
      "loss": 0.0964,
      "step": 141
    },
    {
      "epoch": 0.4251497005988024,
      "grad_norm": 0.7390570197099633,
      "learning_rate": 8.92585664876797e-06,
      "loss": 0.1143,
      "step": 142
    },
    {
      "epoch": 0.4281437125748503,
      "grad_norm": 0.6736053956490804,
      "learning_rate": 8.911251002727588e-06,
      "loss": 0.0887,
      "step": 143
    },
    {
      "epoch": 0.4311377245508982,
      "grad_norm": 0.6974067960802459,
      "learning_rate": 8.896558847604414e-06,
      "loss": 0.0972,
      "step": 144
    },
    {
      "epoch": 0.4341317365269461,
      "grad_norm": 0.7670811816656069,
      "learning_rate": 8.881780508359661e-06,
      "loss": 0.1133,
      "step": 145
    },
    {
      "epoch": 0.437125748502994,
      "grad_norm": 0.7988929981711455,
      "learning_rate": 8.86691631186076e-06,
      "loss": 0.1083,
      "step": 146
    },
    {
      "epoch": 0.44011976047904194,
      "grad_norm": 0.6574760720139775,
      "learning_rate": 8.851966586874138e-06,
      "loss": 0.0876,
      "step": 147
    },
    {
      "epoch": 0.4431137724550898,
      "grad_norm": 0.8118859468746412,
      "learning_rate": 8.836931664057935e-06,
      "loss": 0.1163,
      "step": 148
    },
    {
      "epoch": 0.44610778443113774,
      "grad_norm": 0.7312128128161812,
      "learning_rate": 8.821811875954705e-06,
      "loss": 0.094,
      "step": 149
    },
    {
      "epoch": 0.4491017964071856,
      "grad_norm": 0.7326533571151154,
      "learning_rate": 8.806607556984045e-06,
      "loss": 0.1049,
      "step": 150
    },
    {
      "epoch": 0.45209580838323354,
      "grad_norm": 0.7859032654861962,
      "learning_rate": 8.791319043435213e-06,
      "loss": 0.1037,
      "step": 151
    },
    {
      "epoch": 0.4550898203592814,
      "grad_norm": 0.7317068975252268,
      "learning_rate": 8.775946673459682e-06,
      "loss": 0.0976,
      "step": 152
    },
    {
      "epoch": 0.45808383233532934,
      "grad_norm": 0.7172928742274616,
      "learning_rate": 8.76049078706366e-06,
      "loss": 0.0922,
      "step": 153
    },
    {
      "epoch": 0.46107784431137727,
      "grad_norm": 0.7577953423668164,
      "learning_rate": 8.744951726100572e-06,
      "loss": 0.1003,
      "step": 154
    },
    {
      "epoch": 0.46407185628742514,
      "grad_norm": 0.7052509772420711,
      "learning_rate": 8.729329834263503e-06,
      "loss": 0.0896,
      "step": 155
    },
    {
      "epoch": 0.46706586826347307,
      "grad_norm": 0.871925431823152,
      "learning_rate": 8.713625457077585e-06,
      "loss": 0.1213,
      "step": 156
    },
    {
      "epoch": 0.47005988023952094,
      "grad_norm": 0.8856902379350235,
      "learning_rate": 8.697838941892371e-06,
      "loss": 0.123,
      "step": 157
    },
    {
      "epoch": 0.47305389221556887,
      "grad_norm": 0.7167767009449888,
      "learning_rate": 8.681970637874131e-06,
      "loss": 0.104,
      "step": 158
    },
    {
      "epoch": 0.47604790419161674,
      "grad_norm": 0.7816806991057982,
      "learning_rate": 8.666020895998154e-06,
      "loss": 0.1045,
      "step": 159
    },
    {
      "epoch": 0.47904191616766467,
      "grad_norm": 0.7254011773158139,
      "learning_rate": 8.64999006904096e-06,
      "loss": 0.0915,
      "step": 160
    },
    {
      "epoch": 0.4820359281437126,
      "grad_norm": 0.8449274494253806,
      "learning_rate": 8.63387851157252e-06,
      "loss": 0.1184,
      "step": 161
    },
    {
      "epoch": 0.48502994011976047,
      "grad_norm": 0.7353619224734645,
      "learning_rate": 8.617686579948396e-06,
      "loss": 0.0938,
      "step": 162
    },
    {
      "epoch": 0.4880239520958084,
      "grad_norm": 0.8329532959727286,
      "learning_rate": 8.60141463230187e-06,
      "loss": 0.1072,
      "step": 163
    },
    {
      "epoch": 0.49101796407185627,
      "grad_norm": 0.812467464342694,
      "learning_rate": 8.585063028536015e-06,
      "loss": 0.1084,
      "step": 164
    },
    {
      "epoch": 0.4940119760479042,
      "grad_norm": 0.8752927050779713,
      "learning_rate": 8.568632130315747e-06,
      "loss": 0.1211,
      "step": 165
    },
    {
      "epoch": 0.49700598802395207,
      "grad_norm": 0.6910169882233176,
      "learning_rate": 8.552122301059807e-06,
      "loss": 0.0923,
      "step": 166
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7025122056161462,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.0905,
      "step": 167
    },
    {
      "epoch": 0.5029940119760479,
      "grad_norm": 0.7101174625169111,
      "learning_rate": 8.518867311836808e-06,
      "loss": 0.1079,
      "step": 168
    },
    {
      "epoch": 0.5059880239520959,
      "grad_norm": 0.742272798517134,
      "learning_rate": 8.502122887403882e-06,
      "loss": 0.0874,
      "step": 169
    },
    {
      "epoch": 0.5089820359281437,
      "grad_norm": 0.7325181901034072,
      "learning_rate": 8.485301002987285e-06,
      "loss": 0.0893,
      "step": 170
    },
    {
      "epoch": 0.5119760479041916,
      "grad_norm": 0.7553946371945585,
      "learning_rate": 8.468402030653598e-06,
      "loss": 0.1096,
      "step": 171
    },
    {
      "epoch": 0.5149700598802395,
      "grad_norm": 0.6673958593938553,
      "learning_rate": 8.451426344174433e-06,
      "loss": 0.0981,
      "step": 172
    },
    {
      "epoch": 0.5179640718562875,
      "grad_norm": 0.7973294028265521,
      "learning_rate": 8.434374319018165e-06,
      "loss": 0.1009,
      "step": 173
    },
    {
      "epoch": 0.5209580838323353,
      "grad_norm": 0.7153505444376848,
      "learning_rate": 8.417246332341638e-06,
      "loss": 0.096,
      "step": 174
    },
    {
      "epoch": 0.5239520958083832,
      "grad_norm": 0.7977115431795168,
      "learning_rate": 8.4000427629818e-06,
      "loss": 0.109,
      "step": 175
    },
    {
      "epoch": 0.5269461077844312,
      "grad_norm": 0.6624462621489516,
      "learning_rate": 8.382763991447344e-06,
      "loss": 0.0914,
      "step": 176
    },
    {
      "epoch": 0.5299401197604791,
      "grad_norm": 0.8301699904009856,
      "learning_rate": 8.365410399910287e-06,
      "loss": 0.1072,
      "step": 177
    },
    {
      "epoch": 0.5329341317365269,
      "grad_norm": 0.8318638961993523,
      "learning_rate": 8.347982372197515e-06,
      "loss": 0.1071,
      "step": 178
    },
    {
      "epoch": 0.5359281437125748,
      "grad_norm": 0.7446175404267884,
      "learning_rate": 8.33048029378229e-06,
      "loss": 0.0976,
      "step": 179
    },
    {
      "epoch": 0.5389221556886228,
      "grad_norm": 0.7951570931006732,
      "learning_rate": 8.312904551775731e-06,
      "loss": 0.1204,
      "step": 180
    },
    {
      "epoch": 0.5419161676646707,
      "grad_norm": 0.8286340305563193,
      "learning_rate": 8.295255534918249e-06,
      "loss": 0.1271,
      "step": 181
    },
    {
      "epoch": 0.5449101796407185,
      "grad_norm": 0.9610052678401081,
      "learning_rate": 8.277533633570948e-06,
      "loss": 0.1277,
      "step": 182
    },
    {
      "epoch": 0.5479041916167665,
      "grad_norm": 0.830893503714413,
      "learning_rate": 8.25973923970699e-06,
      "loss": 0.1116,
      "step": 183
    },
    {
      "epoch": 0.5508982035928144,
      "grad_norm": 0.7893323184402998,
      "learning_rate": 8.241872746902934e-06,
      "loss": 0.1153,
      "step": 184
    },
    {
      "epoch": 0.5538922155688623,
      "grad_norm": 0.7662292902434731,
      "learning_rate": 8.223934550330015e-06,
      "loss": 0.1091,
      "step": 185
    },
    {
      "epoch": 0.5568862275449101,
      "grad_norm": 0.7270395908721708,
      "learning_rate": 8.20592504674542e-06,
      "loss": 0.1036,
      "step": 186
    },
    {
      "epoch": 0.5598802395209581,
      "grad_norm": 0.9242656917651041,
      "learning_rate": 8.187844634483495e-06,
      "loss": 0.127,
      "step": 187
    },
    {
      "epoch": 0.562874251497006,
      "grad_norm": 0.772698781266155,
      "learning_rate": 8.16969371344696e-06,
      "loss": 0.1097,
      "step": 188
    },
    {
      "epoch": 0.5658682634730539,
      "grad_norm": 0.7318941409205366,
      "learning_rate": 8.151472685098037e-06,
      "loss": 0.1092,
      "step": 189
    },
    {
      "epoch": 0.5688622754491018,
      "grad_norm": 0.6844197260486172,
      "learning_rate": 8.13318195244958e-06,
      "loss": 0.0976,
      "step": 190
    },
    {
      "epoch": 0.5718562874251497,
      "grad_norm": 0.7422308498693158,
      "learning_rate": 8.114821920056177e-06,
      "loss": 0.0996,
      "step": 191
    },
    {
      "epoch": 0.5748502994011976,
      "grad_norm": 0.8731970535519051,
      "learning_rate": 8.096392994005177e-06,
      "loss": 0.1207,
      "step": 192
    },
    {
      "epoch": 0.5778443113772455,
      "grad_norm": 0.6902635056316513,
      "learning_rate": 8.077895581907719e-06,
      "loss": 0.0986,
      "step": 193
    },
    {
      "epoch": 0.5808383233532934,
      "grad_norm": 0.7134497940610738,
      "learning_rate": 8.059330092889724e-06,
      "loss": 0.1089,
      "step": 194
    },
    {
      "epoch": 0.5838323353293413,
      "grad_norm": 0.7647715239989019,
      "learning_rate": 8.040696937582833e-06,
      "loss": 0.1098,
      "step": 195
    },
    {
      "epoch": 0.5868263473053892,
      "grad_norm": 0.6326447048131935,
      "learning_rate": 8.021996528115335e-06,
      "loss": 0.0802,
      "step": 196
    },
    {
      "epoch": 0.5898203592814372,
      "grad_norm": 0.8264396697338311,
      "learning_rate": 8.003229278103044e-06,
      "loss": 0.1062,
      "step": 197
    },
    {
      "epoch": 0.592814371257485,
      "grad_norm": 0.768695291249376,
      "learning_rate": 7.984395602640153e-06,
      "loss": 0.1102,
      "step": 198
    },
    {
      "epoch": 0.5958083832335329,
      "grad_norm": 0.7193212924041615,
      "learning_rate": 7.96549591829006e-06,
      "loss": 0.0993,
      "step": 199
    },
    {
      "epoch": 0.5988023952095808,
      "grad_norm": 0.6558070472143126,
      "learning_rate": 7.946530643076138e-06,
      "loss": 0.089,
      "step": 200
    },
    {
      "epoch": 0.5988023952095808,
      "eval_loss": 0.11316297948360443,
      "eval_runtime": 8.1239,
      "eval_samples_per_second": 6.647,
      "eval_steps_per_second": 1.723,
      "step": 200
    },
    {
      "epoch": 0.6017964071856288,
      "grad_norm": 0.6304545559864521,
      "learning_rate": 7.927500196472506e-06,
      "loss": 0.091,
      "step": 201
    },
    {
      "epoch": 0.6047904191616766,
      "grad_norm": 0.7436848465834952,
      "learning_rate": 7.908404999394747e-06,
      "loss": 0.0921,
      "step": 202
    },
    {
      "epoch": 0.6077844311377245,
      "grad_norm": 0.8032829831660236,
      "learning_rate": 7.889245474190588e-06,
      "loss": 0.1128,
      "step": 203
    },
    {
      "epoch": 0.6107784431137725,
      "grad_norm": 0.9677011367648443,
      "learning_rate": 7.870022044630569e-06,
      "loss": 0.1301,
      "step": 204
    },
    {
      "epoch": 0.6137724550898204,
      "grad_norm": 0.7812827550278139,
      "learning_rate": 7.85073513589867e-06,
      "loss": 0.1079,
      "step": 205
    },
    {
      "epoch": 0.6167664670658682,
      "grad_norm": 0.7170922381465475,
      "learning_rate": 7.831385174582901e-06,
      "loss": 0.0926,
      "step": 206
    },
    {
      "epoch": 0.6197604790419161,
      "grad_norm": 0.7290994599537557,
      "learning_rate": 7.81197258866587e-06,
      "loss": 0.089,
      "step": 207
    },
    {
      "epoch": 0.6227544910179641,
      "grad_norm": 0.735132543308245,
      "learning_rate": 7.792497807515317e-06,
      "loss": 0.0983,
      "step": 208
    },
    {
      "epoch": 0.625748502994012,
      "grad_norm": 0.6610151565435571,
      "learning_rate": 7.772961261874615e-06,
      "loss": 0.0884,
      "step": 209
    },
    {
      "epoch": 0.6287425149700598,
      "grad_norm": 0.8268533248477717,
      "learning_rate": 7.75336338385325e-06,
      "loss": 0.1038,
      "step": 210
    },
    {
      "epoch": 0.6317365269461078,
      "grad_norm": 0.8347853516989824,
      "learning_rate": 7.733704606917248e-06,
      "loss": 0.1279,
      "step": 211
    },
    {
      "epoch": 0.6347305389221557,
      "grad_norm": 0.8027764162345326,
      "learning_rate": 7.713985365879607e-06,
      "loss": 0.0944,
      "step": 212
    },
    {
      "epoch": 0.6377245508982036,
      "grad_norm": 0.8724003224372571,
      "learning_rate": 7.694206096890667e-06,
      "loss": 0.1275,
      "step": 213
    },
    {
      "epoch": 0.6407185628742516,
      "grad_norm": 0.6958571641590033,
      "learning_rate": 7.674367237428467e-06,
      "loss": 0.0936,
      "step": 214
    },
    {
      "epoch": 0.6437125748502994,
      "grad_norm": 0.7284953995152934,
      "learning_rate": 7.654469226289068e-06,
      "loss": 0.0971,
      "step": 215
    },
    {
      "epoch": 0.6467065868263473,
      "grad_norm": 0.8125565813281055,
      "learning_rate": 7.63451250357685e-06,
      "loss": 0.1073,
      "step": 216
    },
    {
      "epoch": 0.6497005988023952,
      "grad_norm": 0.7574445626756828,
      "learning_rate": 7.614497510694774e-06,
      "loss": 0.0951,
      "step": 217
    },
    {
      "epoch": 0.6526946107784432,
      "grad_norm": 0.7958664766821525,
      "learning_rate": 7.5944246903346204e-06,
      "loss": 0.1073,
      "step": 218
    },
    {
      "epoch": 0.655688622754491,
      "grad_norm": 0.638066139127121,
      "learning_rate": 7.574294486467204e-06,
      "loss": 0.0774,
      "step": 219
    },
    {
      "epoch": 0.6586826347305389,
      "grad_norm": 0.6490093504003906,
      "learning_rate": 7.55410734433254e-06,
      "loss": 0.0896,
      "step": 220
    },
    {
      "epoch": 0.6616766467065869,
      "grad_norm": 0.8000395363872616,
      "learning_rate": 7.533863710430011e-06,
      "loss": 0.1048,
      "step": 221
    },
    {
      "epoch": 0.6646706586826348,
      "grad_norm": 0.609150379175774,
      "learning_rate": 7.513564032508484e-06,
      "loss": 0.0803,
      "step": 222
    },
    {
      "epoch": 0.6676646706586826,
      "grad_norm": 0.8314278005098262,
      "learning_rate": 7.493208759556406e-06,
      "loss": 0.1095,
      "step": 223
    },
    {
      "epoch": 0.6706586826347305,
      "grad_norm": 0.752207398717586,
      "learning_rate": 7.472798341791877e-06,
      "loss": 0.1082,
      "step": 224
    },
    {
      "epoch": 0.6736526946107785,
      "grad_norm": 0.6500202377718381,
      "learning_rate": 7.452333230652688e-06,
      "loss": 0.0753,
      "step": 225
    },
    {
      "epoch": 0.6766467065868264,
      "grad_norm": 0.7464396220628815,
      "learning_rate": 7.431813878786343e-06,
      "loss": 0.1083,
      "step": 226
    },
    {
      "epoch": 0.6796407185628742,
      "grad_norm": 0.671897723239332,
      "learning_rate": 7.4112407400400395e-06,
      "loss": 0.0779,
      "step": 227
    },
    {
      "epoch": 0.6826347305389222,
      "grad_norm": 0.6240112302738774,
      "learning_rate": 7.390614269450633e-06,
      "loss": 0.0785,
      "step": 228
    },
    {
      "epoch": 0.6856287425149701,
      "grad_norm": 0.7162947575530185,
      "learning_rate": 7.369934923234577e-06,
      "loss": 0.0903,
      "step": 229
    },
    {
      "epoch": 0.688622754491018,
      "grad_norm": 0.7710167332241861,
      "learning_rate": 7.349203158777826e-06,
      "loss": 0.0971,
      "step": 230
    },
    {
      "epoch": 0.6916167664670658,
      "grad_norm": 0.7422256126617821,
      "learning_rate": 7.32841943462572e-06,
      "loss": 0.0988,
      "step": 231
    },
    {
      "epoch": 0.6946107784431138,
      "grad_norm": 0.82801734766444,
      "learning_rate": 7.3075842104728445e-06,
      "loss": 0.108,
      "step": 232
    },
    {
      "epoch": 0.6976047904191617,
      "grad_norm": 0.7689698855196472,
      "learning_rate": 7.286697947152868e-06,
      "loss": 0.1094,
      "step": 233
    },
    {
      "epoch": 0.7005988023952096,
      "grad_norm": 0.5763214856261514,
      "learning_rate": 7.265761106628338e-06,
      "loss": 0.0802,
      "step": 234
    },
    {
      "epoch": 0.7035928143712575,
      "grad_norm": 0.6782615406210066,
      "learning_rate": 7.244774151980466e-06,
      "loss": 0.0878,
      "step": 235
    },
    {
      "epoch": 0.7065868263473054,
      "grad_norm": 0.790457574949974,
      "learning_rate": 7.223737547398898e-06,
      "loss": 0.0985,
      "step": 236
    },
    {
      "epoch": 0.7095808383233533,
      "grad_norm": 0.7281274864418545,
      "learning_rate": 7.20265175817143e-06,
      "loss": 0.0993,
      "step": 237
    },
    {
      "epoch": 0.7125748502994012,
      "grad_norm": 0.8838121223242332,
      "learning_rate": 7.181517250673729e-06,
      "loss": 0.1304,
      "step": 238
    },
    {
      "epoch": 0.7155688622754491,
      "grad_norm": 0.7459034331423962,
      "learning_rate": 7.1603344923590065e-06,
      "loss": 0.1008,
      "step": 239
    },
    {
      "epoch": 0.718562874251497,
      "grad_norm": 0.8301863104536503,
      "learning_rate": 7.139103951747694e-06,
      "loss": 0.1149,
      "step": 240
    },
    {
      "epoch": 0.7215568862275449,
      "grad_norm": 0.7794850708214441,
      "learning_rate": 7.1178260984170675e-06,
      "loss": 0.1002,
      "step": 241
    },
    {
      "epoch": 0.7245508982035929,
      "grad_norm": 0.699406535250054,
      "learning_rate": 7.0965014029908654e-06,
      "loss": 0.0874,
      "step": 242
    },
    {
      "epoch": 0.7275449101796407,
      "grad_norm": 0.8078329206980916,
      "learning_rate": 7.075130337128883e-06,
      "loss": 0.1284,
      "step": 243
    },
    {
      "epoch": 0.7305389221556886,
      "grad_norm": 0.8900126533526467,
      "learning_rate": 7.053713373516538e-06,
      "loss": 0.1293,
      "step": 244
    },
    {
      "epoch": 0.7335329341317365,
      "grad_norm": 0.7391492289273155,
      "learning_rate": 7.03225098585441e-06,
      "loss": 0.1044,
      "step": 245
    },
    {
      "epoch": 0.7365269461077845,
      "grad_norm": 0.8347608596349876,
      "learning_rate": 7.0107436488477694e-06,
      "loss": 0.1279,
      "step": 246
    },
    {
      "epoch": 0.7395209580838323,
      "grad_norm": 0.8782486227395034,
      "learning_rate": 6.989191838196083e-06,
      "loss": 0.1187,
      "step": 247
    },
    {
      "epoch": 0.7425149700598802,
      "grad_norm": 0.6327806565426204,
      "learning_rate": 6.9675960305824785e-06,
      "loss": 0.088,
      "step": 248
    },
    {
      "epoch": 0.7455089820359282,
      "grad_norm": 0.8527249309848887,
      "learning_rate": 6.945956703663212e-06,
      "loss": 0.114,
      "step": 249
    },
    {
      "epoch": 0.7485029940119761,
      "grad_norm": 0.7468400992242747,
      "learning_rate": 6.9242743360570985e-06,
      "loss": 0.0988,
      "step": 250
    },
    {
      "epoch": 0.7514970059880239,
      "grad_norm": 0.7848352072453517,
      "learning_rate": 6.9025494073349284e-06,
      "loss": 0.1082,
      "step": 251
    },
    {
      "epoch": 0.7544910179640718,
      "grad_norm": 0.6872779249530724,
      "learning_rate": 6.880782398008862e-06,
      "loss": 0.1001,
      "step": 252
    },
    {
      "epoch": 0.7574850299401198,
      "grad_norm": 0.7076782243110112,
      "learning_rate": 6.858973789521792e-06,
      "loss": 0.091,
      "step": 253
    },
    {
      "epoch": 0.7604790419161677,
      "grad_norm": 0.7192504296528952,
      "learning_rate": 6.837124064236709e-06,
      "loss": 0.0953,
      "step": 254
    },
    {
      "epoch": 0.7634730538922155,
      "grad_norm": 0.6198570724195798,
      "learning_rate": 6.815233705426019e-06,
      "loss": 0.0851,
      "step": 255
    },
    {
      "epoch": 0.7664670658682635,
      "grad_norm": 0.7091267402741389,
      "learning_rate": 6.7933031972608644e-06,
      "loss": 0.0945,
      "step": 256
    },
    {
      "epoch": 0.7694610778443114,
      "grad_norm": 0.8048678733868162,
      "learning_rate": 6.771333024800411e-06,
      "loss": 0.1067,
      "step": 257
    },
    {
      "epoch": 0.7724550898203593,
      "grad_norm": 0.829535205897187,
      "learning_rate": 6.74932367398112e-06,
      "loss": 0.1109,
      "step": 258
    },
    {
      "epoch": 0.7754491017964071,
      "grad_norm": 0.6289007304192439,
      "learning_rate": 6.727275631605996e-06,
      "loss": 0.0822,
      "step": 259
    },
    {
      "epoch": 0.7784431137724551,
      "grad_norm": 0.7856477507347315,
      "learning_rate": 6.70518938533383e-06,
      "loss": 0.0987,
      "step": 260
    },
    {
      "epoch": 0.781437125748503,
      "grad_norm": 0.6575545537772718,
      "learning_rate": 6.683065423668403e-06,
      "loss": 0.0831,
      "step": 261
    },
    {
      "epoch": 0.7844311377245509,
      "grad_norm": 0.7915352317815166,
      "learning_rate": 6.660904235947687e-06,
      "loss": 0.1051,
      "step": 262
    },
    {
      "epoch": 0.7874251497005988,
      "grad_norm": 0.7400481183429451,
      "learning_rate": 6.638706312333018e-06,
      "loss": 0.0948,
      "step": 263
    },
    {
      "epoch": 0.7904191616766467,
      "grad_norm": 0.8485669565284383,
      "learning_rate": 6.61647214379826e-06,
      "loss": 0.1095,
      "step": 264
    },
    {
      "epoch": 0.7934131736526946,
      "grad_norm": 0.9051666838584678,
      "learning_rate": 6.594202222118941e-06,
      "loss": 0.1204,
      "step": 265
    },
    {
      "epoch": 0.7964071856287425,
      "grad_norm": 0.7121199392646126,
      "learning_rate": 6.571897039861377e-06,
      "loss": 0.0951,
      "step": 266
    },
    {
      "epoch": 0.7994011976047904,
      "grad_norm": 0.7453169109951469,
      "learning_rate": 6.549557090371775e-06,
      "loss": 0.1089,
      "step": 267
    },
    {
      "epoch": 0.8023952095808383,
      "grad_norm": 0.5965983117784317,
      "learning_rate": 6.527182867765333e-06,
      "loss": 0.0745,
      "step": 268
    },
    {
      "epoch": 0.8053892215568862,
      "grad_norm": 0.8710522635670452,
      "learning_rate": 6.504774866915291e-06,
      "loss": 0.1263,
      "step": 269
    },
    {
      "epoch": 0.8083832335329342,
      "grad_norm": 0.7428249361033396,
      "learning_rate": 6.482333583442002e-06,
      "loss": 0.0952,
      "step": 270
    },
    {
      "epoch": 0.811377245508982,
      "grad_norm": 0.7590066563934622,
      "learning_rate": 6.459859513701967e-06,
      "loss": 0.0991,
      "step": 271
    },
    {
      "epoch": 0.8143712574850299,
      "grad_norm": 0.8963092338879843,
      "learning_rate": 6.437353154776848e-06,
      "loss": 0.103,
      "step": 272
    },
    {
      "epoch": 0.8173652694610778,
      "grad_norm": 0.7068452673868478,
      "learning_rate": 6.414815004462483e-06,
      "loss": 0.0801,
      "step": 273
    },
    {
      "epoch": 0.8203592814371258,
      "grad_norm": 0.7609504513707516,
      "learning_rate": 6.3922455612578715e-06,
      "loss": 0.1134,
      "step": 274
    },
    {
      "epoch": 0.8233532934131736,
      "grad_norm": 0.7064896922285335,
      "learning_rate": 6.369645324354149e-06,
      "loss": 0.089,
      "step": 275
    },
    {
      "epoch": 0.8263473053892215,
      "grad_norm": 0.8739759345720145,
      "learning_rate": 6.3470147936235485e-06,
      "loss": 0.1203,
      "step": 276
    },
    {
      "epoch": 0.8293413173652695,
      "grad_norm": 0.5782011838402902,
      "learning_rate": 6.3243544696083355e-06,
      "loss": 0.0846,
      "step": 277
    },
    {
      "epoch": 0.8323353293413174,
      "grad_norm": 0.8163189314840777,
      "learning_rate": 6.301664853509755e-06,
      "loss": 0.1133,
      "step": 278
    },
    {
      "epoch": 0.8353293413173652,
      "grad_norm": 0.6858033204742969,
      "learning_rate": 6.278946447176924e-06,
      "loss": 0.0835,
      "step": 279
    },
    {
      "epoch": 0.8383233532934131,
      "grad_norm": 0.7325665885602335,
      "learning_rate": 6.256199753095745e-06,
      "loss": 0.1029,
      "step": 280
    },
    {
      "epoch": 0.8413173652694611,
      "grad_norm": 0.6012169890415201,
      "learning_rate": 6.233425274377793e-06,
      "loss": 0.0819,
      "step": 281
    },
    {
      "epoch": 0.844311377245509,
      "grad_norm": 0.7438888886495855,
      "learning_rate": 6.21062351474918e-06,
      "loss": 0.0968,
      "step": 282
    },
    {
      "epoch": 0.8473053892215568,
      "grad_norm": 0.7467175843801874,
      "learning_rate": 6.18779497853942e-06,
      "loss": 0.0993,
      "step": 283
    },
    {
      "epoch": 0.8502994011976048,
      "grad_norm": 0.7361903845681506,
      "learning_rate": 6.164940170670266e-06,
      "loss": 0.1056,
      "step": 284
    },
    {
      "epoch": 0.8532934131736527,
      "grad_norm": 0.7060976208071157,
      "learning_rate": 6.142059596644557e-06,
      "loss": 0.0849,
      "step": 285
    },
    {
      "epoch": 0.8562874251497006,
      "grad_norm": 0.6780804390450543,
      "learning_rate": 6.11915376253502e-06,
      "loss": 0.0817,
      "step": 286
    },
    {
      "epoch": 0.8592814371257484,
      "grad_norm": 0.7253925510848019,
      "learning_rate": 6.096223174973091e-06,
      "loss": 0.0952,
      "step": 287
    },
    {
      "epoch": 0.8622754491017964,
      "grad_norm": 0.710231420622638,
      "learning_rate": 6.073268341137694e-06,
      "loss": 0.0886,
      "step": 288
    },
    {
      "epoch": 0.8652694610778443,
      "grad_norm": 0.7392452658230668,
      "learning_rate": 6.050289768744042e-06,
      "loss": 0.0936,
      "step": 289
    },
    {
      "epoch": 0.8682634730538922,
      "grad_norm": 0.756665130571337,
      "learning_rate": 6.0272879660323936e-06,
      "loss": 0.0974,
      "step": 290
    },
    {
      "epoch": 0.8712574850299402,
      "grad_norm": 0.7713478746391634,
      "learning_rate": 6.004263441756815e-06,
      "loss": 0.1002,
      "step": 291
    },
    {
      "epoch": 0.874251497005988,
      "grad_norm": 0.8040258494823045,
      "learning_rate": 5.98121670517393e-06,
      "loss": 0.1017,
      "step": 292
    },
    {
      "epoch": 0.8772455089820359,
      "grad_norm": 0.6408252753809833,
      "learning_rate": 5.958148266031654e-06,
      "loss": 0.0969,
      "step": 293
    },
    {
      "epoch": 0.8802395209580839,
      "grad_norm": 0.6728849817323966,
      "learning_rate": 5.935058634557917e-06,
      "loss": 0.0869,
      "step": 294
    },
    {
      "epoch": 0.8832335329341318,
      "grad_norm": 0.6848133060596014,
      "learning_rate": 5.911948321449384e-06,
      "loss": 0.0885,
      "step": 295
    },
    {
      "epoch": 0.8862275449101796,
      "grad_norm": 0.783519122492361,
      "learning_rate": 5.8888178378601565e-06,
      "loss": 0.101,
      "step": 296
    },
    {
      "epoch": 0.8892215568862275,
      "grad_norm": 0.7046353274842205,
      "learning_rate": 5.865667695390468e-06,
      "loss": 0.0932,
      "step": 297
    },
    {
      "epoch": 0.8922155688622755,
      "grad_norm": 0.6258408674384502,
      "learning_rate": 5.842498406075363e-06,
      "loss": 0.0743,
      "step": 298
    },
    {
      "epoch": 0.8952095808383234,
      "grad_norm": 0.6899501104158414,
      "learning_rate": 5.819310482373381e-06,
      "loss": 0.081,
      "step": 299
    },
    {
      "epoch": 0.8982035928143712,
      "grad_norm": 0.7609883445063802,
      "learning_rate": 5.796104437155213e-06,
      "loss": 0.1017,
      "step": 300
    },
    {
      "epoch": 0.9011976047904192,
      "grad_norm": 0.6873816360092,
      "learning_rate": 5.772880783692363e-06,
      "loss": 0.0837,
      "step": 301
    },
    {
      "epoch": 0.9041916167664671,
      "grad_norm": 0.7668016513242792,
      "learning_rate": 5.749640035645798e-06,
      "loss": 0.0988,
      "step": 302
    },
    {
      "epoch": 0.907185628742515,
      "grad_norm": 0.6802941610188087,
      "learning_rate": 5.726382707054578e-06,
      "loss": 0.0845,
      "step": 303
    },
    {
      "epoch": 0.9101796407185628,
      "grad_norm": 0.7932851937438237,
      "learning_rate": 5.703109312324493e-06,
      "loss": 0.097,
      "step": 304
    },
    {
      "epoch": 0.9131736526946108,
      "grad_norm": 0.8139095664770755,
      "learning_rate": 5.679820366216684e-06,
      "loss": 0.1094,
      "step": 305
    },
    {
      "epoch": 0.9161676646706587,
      "grad_norm": 0.6009794212158722,
      "learning_rate": 5.656516383836263e-06,
      "loss": 0.074,
      "step": 306
    },
    {
      "epoch": 0.9191616766467066,
      "grad_norm": 0.8236868897295665,
      "learning_rate": 5.6331978806209044e-06,
      "loss": 0.1125,
      "step": 307
    },
    {
      "epoch": 0.9221556886227545,
      "grad_norm": 0.7664074465141002,
      "learning_rate": 5.609865372329461e-06,
      "loss": 0.0945,
      "step": 308
    },
    {
      "epoch": 0.9251497005988024,
      "grad_norm": 0.7257251057015707,
      "learning_rate": 5.586519375030549e-06,
      "loss": 0.0989,
      "step": 309
    },
    {
      "epoch": 0.9281437125748503,
      "grad_norm": 0.754338036509913,
      "learning_rate": 5.5631604050911354e-06,
      "loss": 0.0943,
      "step": 310
    },
    {
      "epoch": 0.9311377245508982,
      "grad_norm": 0.7636488457535154,
      "learning_rate": 5.539788979165115e-06,
      "loss": 0.0966,
      "step": 311
    },
    {
      "epoch": 0.9341317365269461,
      "grad_norm": 0.648152170923811,
      "learning_rate": 5.516405614181883e-06,
      "loss": 0.0894,
      "step": 312
    },
    {
      "epoch": 0.937125748502994,
      "grad_norm": 0.642586989434421,
      "learning_rate": 5.4930108273349034e-06,
      "loss": 0.0783,
      "step": 313
    },
    {
      "epoch": 0.9401197604790419,
      "grad_norm": 0.7456220984422796,
      "learning_rate": 5.4696051360702725e-06,
      "loss": 0.0898,
      "step": 314
    },
    {
      "epoch": 0.9431137724550899,
      "grad_norm": 0.9011055617017701,
      "learning_rate": 5.446189058075265e-06,
      "loss": 0.1167,
      "step": 315
    },
    {
      "epoch": 0.9461077844311377,
      "grad_norm": 0.8442826019802778,
      "learning_rate": 5.4227631112668955e-06,
      "loss": 0.1077,
      "step": 316
    },
    {
      "epoch": 0.9491017964071856,
      "grad_norm": 0.7461246030744498,
      "learning_rate": 5.39932781378045e-06,
      "loss": 0.0869,
      "step": 317
    },
    {
      "epoch": 0.9520958083832335,
      "grad_norm": 0.7759524349326654,
      "learning_rate": 5.375883683958041e-06,
      "loss": 0.1058,
      "step": 318
    },
    {
      "epoch": 0.9550898203592815,
      "grad_norm": 0.6908245440341831,
      "learning_rate": 5.3524312403371255e-06,
      "loss": 0.085,
      "step": 319
    },
    {
      "epoch": 0.9580838323353293,
      "grad_norm": 0.702680242538725,
      "learning_rate": 5.328971001639054e-06,
      "loss": 0.0904,
      "step": 320
    },
    {
      "epoch": 0.9610778443113772,
      "grad_norm": 0.8439322326649915,
      "learning_rate": 5.3055034867575825e-06,
      "loss": 0.1033,
      "step": 321
    },
    {
      "epoch": 0.9640718562874252,
      "grad_norm": 0.8380510560185647,
      "learning_rate": 5.282029214747404e-06,
      "loss": 0.1012,
      "step": 322
    },
    {
      "epoch": 0.9670658682634731,
      "grad_norm": 0.7919590386089189,
      "learning_rate": 5.258548704812667e-06,
      "loss": 0.1019,
      "step": 323
    },
    {
      "epoch": 0.9700598802395209,
      "grad_norm": 0.8663983698657292,
      "learning_rate": 5.235062476295488e-06,
      "loss": 0.1106,
      "step": 324
    },
    {
      "epoch": 0.9730538922155688,
      "grad_norm": 0.7141558445041686,
      "learning_rate": 5.211571048664469e-06,
      "loss": 0.085,
      "step": 325
    },
    {
      "epoch": 0.9760479041916168,
      "grad_norm": 0.693845225098938,
      "learning_rate": 5.188074941503203e-06,
      "loss": 0.0931,
      "step": 326
    },
    {
      "epoch": 0.9790419161676647,
      "grad_norm": 0.833596856387305,
      "learning_rate": 5.164574674498788e-06,
      "loss": 0.0984,
      "step": 327
    },
    {
      "epoch": 0.9820359281437125,
      "grad_norm": 0.7276379962293017,
      "learning_rate": 5.141070767430331e-06,
      "loss": 0.093,
      "step": 328
    },
    {
      "epoch": 0.9850299401197605,
      "grad_norm": 0.7687001332573924,
      "learning_rate": 5.117563740157444e-06,
      "loss": 0.0948,
      "step": 329
    },
    {
      "epoch": 0.9880239520958084,
      "grad_norm": 0.8444313121451766,
      "learning_rate": 5.094054112608758e-06,
      "loss": 0.1012,
      "step": 330
    },
    {
      "epoch": 0.9910179640718563,
      "grad_norm": 0.7508007120709564,
      "learning_rate": 5.070542404770413e-06,
      "loss": 0.0981,
      "step": 331
    },
    {
      "epoch": 0.9940119760479041,
      "grad_norm": 0.641568995522917,
      "learning_rate": 5.047029136674563e-06,
      "loss": 0.0848,
      "step": 332
    },
    {
      "epoch": 0.9970059880239521,
      "grad_norm": 0.9728819815537252,
      "learning_rate": 5.023514828387868e-06,
      "loss": 0.1241,
      "step": 333
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6151130125840837,
      "learning_rate": 5e-06,
      "loss": 0.0608,
      "step": 334
    },
    {
      "epoch": 1.0029940119760479,
      "grad_norm": 0.520855518282977,
      "learning_rate": 4.976485171612134e-06,
      "loss": 0.0438,
      "step": 335
    },
    {
      "epoch": 1.0059880239520957,
      "grad_norm": 0.5295178507477781,
      "learning_rate": 4.95297086332544e-06,
      "loss": 0.0432,
      "step": 336
    },
    {
      "epoch": 1.0089820359281436,
      "grad_norm": 0.48953467566376024,
      "learning_rate": 4.9294575952295896e-06,
      "loss": 0.0371,
      "step": 337
    },
    {
      "epoch": 1.0119760479041917,
      "grad_norm": 0.47681774766524027,
      "learning_rate": 4.905945887391242e-06,
      "loss": 0.0393,
      "step": 338
    },
    {
      "epoch": 1.0149700598802396,
      "grad_norm": 0.3907312094734483,
      "learning_rate": 4.882436259842556e-06,
      "loss": 0.029,
      "step": 339
    },
    {
      "epoch": 1.0179640718562875,
      "grad_norm": 0.47929472123423666,
      "learning_rate": 4.858929232569671e-06,
      "loss": 0.0382,
      "step": 340
    },
    {
      "epoch": 1.0209580838323353,
      "grad_norm": 0.6032620172287063,
      "learning_rate": 4.835425325501214e-06,
      "loss": 0.052,
      "step": 341
    },
    {
      "epoch": 1.0239520958083832,
      "grad_norm": 0.4658769961010715,
      "learning_rate": 4.811925058496799e-06,
      "loss": 0.0378,
      "step": 342
    },
    {
      "epoch": 1.026946107784431,
      "grad_norm": 0.46225597137071484,
      "learning_rate": 4.788428951335534e-06,
      "loss": 0.0344,
      "step": 343
    },
    {
      "epoch": 1.029940119760479,
      "grad_norm": 0.543946303586491,
      "learning_rate": 4.7649375237045135e-06,
      "loss": 0.0433,
      "step": 344
    },
    {
      "epoch": 1.032934131736527,
      "grad_norm": 0.3706676296201268,
      "learning_rate": 4.741451295187333e-06,
      "loss": 0.0269,
      "step": 345
    },
    {
      "epoch": 1.035928143712575,
      "grad_norm": 0.5087278080472419,
      "learning_rate": 4.717970785252596e-06,
      "loss": 0.0393,
      "step": 346
    },
    {
      "epoch": 1.0389221556886228,
      "grad_norm": 0.5501334737094976,
      "learning_rate": 4.694496513242418e-06,
      "loss": 0.0368,
      "step": 347
    },
    {
      "epoch": 1.0419161676646707,
      "grad_norm": 0.4832730585372898,
      "learning_rate": 4.671028998360947e-06,
      "loss": 0.0344,
      "step": 348
    },
    {
      "epoch": 1.0449101796407185,
      "grad_norm": 0.4822162478683523,
      "learning_rate": 4.647568759662876e-06,
      "loss": 0.0323,
      "step": 349
    },
    {
      "epoch": 1.0479041916167664,
      "grad_norm": 0.514896112806131,
      "learning_rate": 4.624116316041962e-06,
      "loss": 0.0361,
      "step": 350
    },
    {
      "epoch": 1.0508982035928143,
      "grad_norm": 0.4489448042786132,
      "learning_rate": 4.600672186219551e-06,
      "loss": 0.0277,
      "step": 351
    },
    {
      "epoch": 1.0538922155688624,
      "grad_norm": 0.6260360261901665,
      "learning_rate": 4.5772368887331044e-06,
      "loss": 0.0316,
      "step": 352
    },
    {
      "epoch": 1.0568862275449102,
      "grad_norm": 0.6484973364998764,
      "learning_rate": 4.553810941924735e-06,
      "loss": 0.0477,
      "step": 353
    },
    {
      "epoch": 1.0598802395209581,
      "grad_norm": 0.4851749222501382,
      "learning_rate": 4.530394863929728e-06,
      "loss": 0.0319,
      "step": 354
    },
    {
      "epoch": 1.062874251497006,
      "grad_norm": 0.5811881093853859,
      "learning_rate": 4.506989172665097e-06,
      "loss": 0.0348,
      "step": 355
    },
    {
      "epoch": 1.0658682634730539,
      "grad_norm": 0.5818694955561754,
      "learning_rate": 4.483594385818119e-06,
      "loss": 0.0282,
      "step": 356
    },
    {
      "epoch": 1.0688622754491017,
      "grad_norm": 0.764470367501723,
      "learning_rate": 4.460211020834887e-06,
      "loss": 0.0376,
      "step": 357
    },
    {
      "epoch": 1.0718562874251496,
      "grad_norm": 0.600500506572781,
      "learning_rate": 4.436839594908866e-06,
      "loss": 0.0318,
      "step": 358
    },
    {
      "epoch": 1.0748502994011977,
      "grad_norm": 0.4816071273947944,
      "learning_rate": 4.4134806249694514e-06,
      "loss": 0.0215,
      "step": 359
    },
    {
      "epoch": 1.0778443113772456,
      "grad_norm": 0.6914671756249607,
      "learning_rate": 4.39013462767054e-06,
      "loss": 0.0356,
      "step": 360
    },
    {
      "epoch": 1.0808383233532934,
      "grad_norm": 0.5445466145729043,
      "learning_rate": 4.366802119379098e-06,
      "loss": 0.0314,
      "step": 361
    },
    {
      "epoch": 1.0838323353293413,
      "grad_norm": 0.6159950401229202,
      "learning_rate": 4.34348361616374e-06,
      "loss": 0.0292,
      "step": 362
    },
    {
      "epoch": 1.0868263473053892,
      "grad_norm": 0.7519825023230998,
      "learning_rate": 4.3201796337833165e-06,
      "loss": 0.0392,
      "step": 363
    },
    {
      "epoch": 1.089820359281437,
      "grad_norm": 0.7421940535777576,
      "learning_rate": 4.29689068767551e-06,
      "loss": 0.0258,
      "step": 364
    },
    {
      "epoch": 1.092814371257485,
      "grad_norm": 0.6460367808278974,
      "learning_rate": 4.273617292945425e-06,
      "loss": 0.0364,
      "step": 365
    },
    {
      "epoch": 1.095808383233533,
      "grad_norm": 0.5235683650971498,
      "learning_rate": 4.250359964354203e-06,
      "loss": 0.0258,
      "step": 366
    },
    {
      "epoch": 1.098802395209581,
      "grad_norm": 0.675594350362758,
      "learning_rate": 4.227119216307637e-06,
      "loss": 0.0317,
      "step": 367
    },
    {
      "epoch": 1.1017964071856288,
      "grad_norm": 0.6482328208729161,
      "learning_rate": 4.203895562844789e-06,
      "loss": 0.0299,
      "step": 368
    },
    {
      "epoch": 1.1047904191616766,
      "grad_norm": 0.6101102136402125,
      "learning_rate": 4.18068951762662e-06,
      "loss": 0.0238,
      "step": 369
    },
    {
      "epoch": 1.1077844311377245,
      "grad_norm": 0.545985002863651,
      "learning_rate": 4.157501593924638e-06,
      "loss": 0.0278,
      "step": 370
    },
    {
      "epoch": 1.1107784431137724,
      "grad_norm": 0.7190846077424872,
      "learning_rate": 4.134332304609533e-06,
      "loss": 0.0374,
      "step": 371
    },
    {
      "epoch": 1.1137724550898203,
      "grad_norm": 0.5766053933811589,
      "learning_rate": 4.111182162139844e-06,
      "loss": 0.0271,
      "step": 372
    },
    {
      "epoch": 1.1167664670658684,
      "grad_norm": 0.6259397511624576,
      "learning_rate": 4.088051678550617e-06,
      "loss": 0.028,
      "step": 373
    },
    {
      "epoch": 1.1197604790419162,
      "grad_norm": 0.6717503311508694,
      "learning_rate": 4.064941365442084e-06,
      "loss": 0.032,
      "step": 374
    },
    {
      "epoch": 1.122754491017964,
      "grad_norm": 0.8476866750736212,
      "learning_rate": 4.041851733968348e-06,
      "loss": 0.0388,
      "step": 375
    },
    {
      "epoch": 1.125748502994012,
      "grad_norm": 0.8231064142335593,
      "learning_rate": 4.018783294826071e-06,
      "loss": 0.0408,
      "step": 376
    },
    {
      "epoch": 1.1287425149700598,
      "grad_norm": 0.8034796738776325,
      "learning_rate": 3.995736558243186e-06,
      "loss": 0.0381,
      "step": 377
    },
    {
      "epoch": 1.1317365269461077,
      "grad_norm": 0.563563482124499,
      "learning_rate": 3.972712033967608e-06,
      "loss": 0.0269,
      "step": 378
    },
    {
      "epoch": 1.1347305389221556,
      "grad_norm": 0.6403165091281876,
      "learning_rate": 3.949710231255961e-06,
      "loss": 0.0344,
      "step": 379
    },
    {
      "epoch": 1.1377245508982037,
      "grad_norm": 0.6500257224704857,
      "learning_rate": 3.926731658862307e-06,
      "loss": 0.031,
      "step": 380
    },
    {
      "epoch": 1.1407185628742516,
      "grad_norm": 0.5583005387566806,
      "learning_rate": 3.903776825026912e-06,
      "loss": 0.0276,
      "step": 381
    },
    {
      "epoch": 1.1437125748502994,
      "grad_norm": 0.606380535404849,
      "learning_rate": 3.8808462374649805e-06,
      "loss": 0.0335,
      "step": 382
    },
    {
      "epoch": 1.1467065868263473,
      "grad_norm": 0.6299806076306098,
      "learning_rate": 3.857940403355444e-06,
      "loss": 0.0306,
      "step": 383
    },
    {
      "epoch": 1.1497005988023952,
      "grad_norm": 0.5992470579637572,
      "learning_rate": 3.8350598293297345e-06,
      "loss": 0.0347,
      "step": 384
    },
    {
      "epoch": 1.152694610778443,
      "grad_norm": 0.6517254523523051,
      "learning_rate": 3.8122050214605822e-06,
      "loss": 0.0345,
      "step": 385
    },
    {
      "epoch": 1.1556886227544911,
      "grad_norm": 0.7040145436567912,
      "learning_rate": 3.7893764852508207e-06,
      "loss": 0.0357,
      "step": 386
    },
    {
      "epoch": 1.158682634730539,
      "grad_norm": 0.5882493868186172,
      "learning_rate": 3.766574725622208e-06,
      "loss": 0.0274,
      "step": 387
    },
    {
      "epoch": 1.1616766467065869,
      "grad_norm": 0.564087962477084,
      "learning_rate": 3.7438002469042567e-06,
      "loss": 0.027,
      "step": 388
    },
    {
      "epoch": 1.1646706586826348,
      "grad_norm": 0.5859086535052078,
      "learning_rate": 3.721053552823078e-06,
      "loss": 0.03,
      "step": 389
    },
    {
      "epoch": 1.1676646706586826,
      "grad_norm": 0.5639428879642789,
      "learning_rate": 3.698335146490246e-06,
      "loss": 0.0286,
      "step": 390
    },
    {
      "epoch": 1.1706586826347305,
      "grad_norm": 0.6150988045436406,
      "learning_rate": 3.675645530391665e-06,
      "loss": 0.0276,
      "step": 391
    },
    {
      "epoch": 1.1736526946107784,
      "grad_norm": 0.70557716817553,
      "learning_rate": 3.652985206376455e-06,
      "loss": 0.0306,
      "step": 392
    },
    {
      "epoch": 1.1766467065868262,
      "grad_norm": 0.6202065881421004,
      "learning_rate": 3.630354675645853e-06,
      "loss": 0.0338,
      "step": 393
    },
    {
      "epoch": 1.1796407185628743,
      "grad_norm": 0.6882199835077807,
      "learning_rate": 3.6077544387421293e-06,
      "loss": 0.0336,
      "step": 394
    },
    {
      "epoch": 1.1826347305389222,
      "grad_norm": 0.5282836499310867,
      "learning_rate": 3.5851849955375177e-06,
      "loss": 0.0266,
      "step": 395
    },
    {
      "epoch": 1.18562874251497,
      "grad_norm": 0.5465037516279739,
      "learning_rate": 3.5626468452231534e-06,
      "loss": 0.0269,
      "step": 396
    },
    {
      "epoch": 1.188622754491018,
      "grad_norm": 0.5005915068856803,
      "learning_rate": 3.540140486298035e-06,
      "loss": 0.0244,
      "step": 397
    },
    {
      "epoch": 1.1916167664670658,
      "grad_norm": 0.6785371118074063,
      "learning_rate": 3.517666416557999e-06,
      "loss": 0.0403,
      "step": 398
    },
    {
      "epoch": 1.1946107784431137,
      "grad_norm": 0.5878778907323362,
      "learning_rate": 3.495225133084712e-06,
      "loss": 0.0243,
      "step": 399
    },
    {
      "epoch": 1.1976047904191618,
      "grad_norm": 0.48844524063939,
      "learning_rate": 3.472817132234669e-06,
      "loss": 0.023,
      "step": 400
    },
    {
      "epoch": 1.1976047904191618,
      "eval_loss": 0.12322219461202621,
      "eval_runtime": 8.1085,
      "eval_samples_per_second": 6.66,
      "eval_steps_per_second": 1.727,
      "step": 400
    },
    {
      "epoch": 1.2005988023952097,
      "grad_norm": 0.6829333302272778,
      "learning_rate": 3.4504429096282246e-06,
      "loss": 0.0347,
      "step": 401
    },
    {
      "epoch": 1.2035928143712575,
      "grad_norm": 0.5798672272403903,
      "learning_rate": 3.428102960138625e-06,
      "loss": 0.0311,
      "step": 402
    },
    {
      "epoch": 1.2065868263473054,
      "grad_norm": 0.6867005402440246,
      "learning_rate": 3.405797777881059e-06,
      "loss": 0.035,
      "step": 403
    },
    {
      "epoch": 1.2095808383233533,
      "grad_norm": 0.5035948144693635,
      "learning_rate": 3.3835278562017405e-06,
      "loss": 0.0223,
      "step": 404
    },
    {
      "epoch": 1.2125748502994012,
      "grad_norm": 0.5717272965423454,
      "learning_rate": 3.3612936876669834e-06,
      "loss": 0.0282,
      "step": 405
    },
    {
      "epoch": 1.215568862275449,
      "grad_norm": 0.6739676822150292,
      "learning_rate": 3.3390957640523147e-06,
      "loss": 0.0359,
      "step": 406
    },
    {
      "epoch": 1.218562874251497,
      "grad_norm": 0.5939927144786713,
      "learning_rate": 3.3169345763315986e-06,
      "loss": 0.0288,
      "step": 407
    },
    {
      "epoch": 1.221556886227545,
      "grad_norm": 0.6441701309562623,
      "learning_rate": 3.29481061466617e-06,
      "loss": 0.0338,
      "step": 408
    },
    {
      "epoch": 1.2245508982035929,
      "grad_norm": 0.7796008492932952,
      "learning_rate": 3.2727243683940045e-06,
      "loss": 0.0388,
      "step": 409
    },
    {
      "epoch": 1.2275449101796407,
      "grad_norm": 0.538841699281851,
      "learning_rate": 3.2506763260188824e-06,
      "loss": 0.0246,
      "step": 410
    },
    {
      "epoch": 1.2305389221556886,
      "grad_norm": 0.5403177062470527,
      "learning_rate": 3.2286669751995905e-06,
      "loss": 0.0259,
      "step": 411
    },
    {
      "epoch": 1.2335329341317365,
      "grad_norm": 0.5832392664340997,
      "learning_rate": 3.2066968027391377e-06,
      "loss": 0.026,
      "step": 412
    },
    {
      "epoch": 1.2365269461077844,
      "grad_norm": 0.5930307451951693,
      "learning_rate": 3.1847662945739833e-06,
      "loss": 0.0274,
      "step": 413
    },
    {
      "epoch": 1.2395209580838324,
      "grad_norm": 0.5570298595397972,
      "learning_rate": 3.1628759357632943e-06,
      "loss": 0.0294,
      "step": 414
    },
    {
      "epoch": 1.2425149700598803,
      "grad_norm": 0.6326493884444826,
      "learning_rate": 3.1410262104782086e-06,
      "loss": 0.0327,
      "step": 415
    },
    {
      "epoch": 1.2455089820359282,
      "grad_norm": 0.5825709131092239,
      "learning_rate": 3.119217601991139e-06,
      "loss": 0.0319,
      "step": 416
    },
    {
      "epoch": 1.248502994011976,
      "grad_norm": 0.6766963272521479,
      "learning_rate": 3.0974505926650724e-06,
      "loss": 0.0338,
      "step": 417
    },
    {
      "epoch": 1.251497005988024,
      "grad_norm": 0.6473792397816943,
      "learning_rate": 3.0757256639429027e-06,
      "loss": 0.0336,
      "step": 418
    },
    {
      "epoch": 1.2544910179640718,
      "grad_norm": 0.5934284060335054,
      "learning_rate": 3.0540432963367907e-06,
      "loss": 0.0346,
      "step": 419
    },
    {
      "epoch": 1.2574850299401197,
      "grad_norm": 0.6086280459997492,
      "learning_rate": 3.032403969417523e-06,
      "loss": 0.0305,
      "step": 420
    },
    {
      "epoch": 1.2604790419161676,
      "grad_norm": 0.48833023196064684,
      "learning_rate": 3.010808161803917e-06,
      "loss": 0.0252,
      "step": 421
    },
    {
      "epoch": 1.2634730538922156,
      "grad_norm": 0.6824141224504635,
      "learning_rate": 2.9892563511522305e-06,
      "loss": 0.0341,
      "step": 422
    },
    {
      "epoch": 1.2664670658682635,
      "grad_norm": 0.5740342444777348,
      "learning_rate": 2.9677490141455915e-06,
      "loss": 0.0273,
      "step": 423
    },
    {
      "epoch": 1.2694610778443114,
      "grad_norm": 0.5754071380571743,
      "learning_rate": 2.946286626483463e-06,
      "loss": 0.0303,
      "step": 424
    },
    {
      "epoch": 1.2724550898203593,
      "grad_norm": 0.5492557771422122,
      "learning_rate": 2.924869662871117e-06,
      "loss": 0.0264,
      "step": 425
    },
    {
      "epoch": 1.2754491017964071,
      "grad_norm": 0.6340244115849604,
      "learning_rate": 2.903498597009136e-06,
      "loss": 0.0373,
      "step": 426
    },
    {
      "epoch": 1.278443113772455,
      "grad_norm": 0.5584430583170973,
      "learning_rate": 2.8821739015829338e-06,
      "loss": 0.0263,
      "step": 427
    },
    {
      "epoch": 1.281437125748503,
      "grad_norm": 0.45258799384262166,
      "learning_rate": 2.8608960482523058e-06,
      "loss": 0.0222,
      "step": 428
    },
    {
      "epoch": 1.284431137724551,
      "grad_norm": 0.5930063563736776,
      "learning_rate": 2.839665507640992e-06,
      "loss": 0.0315,
      "step": 429
    },
    {
      "epoch": 1.2874251497005988,
      "grad_norm": 0.6046333452773391,
      "learning_rate": 2.818482749326272e-06,
      "loss": 0.0277,
      "step": 430
    },
    {
      "epoch": 1.2904191616766467,
      "grad_norm": 0.6720189361309344,
      "learning_rate": 2.797348241828569e-06,
      "loss": 0.0325,
      "step": 431
    },
    {
      "epoch": 1.2934131736526946,
      "grad_norm": 0.4776689219499248,
      "learning_rate": 2.776262452601104e-06,
      "loss": 0.022,
      "step": 432
    },
    {
      "epoch": 1.2964071856287425,
      "grad_norm": 0.6574757756150779,
      "learning_rate": 2.7552258480195348e-06,
      "loss": 0.0313,
      "step": 433
    },
    {
      "epoch": 1.2994011976047903,
      "grad_norm": 0.5802622813671241,
      "learning_rate": 2.734238893371667e-06,
      "loss": 0.0263,
      "step": 434
    },
    {
      "epoch": 1.3023952095808382,
      "grad_norm": 0.6021251940143294,
      "learning_rate": 2.7133020528471322e-06,
      "loss": 0.0282,
      "step": 435
    },
    {
      "epoch": 1.3053892215568863,
      "grad_norm": 0.5836365068157243,
      "learning_rate": 2.6924157895271563e-06,
      "loss": 0.0283,
      "step": 436
    },
    {
      "epoch": 1.3083832335329342,
      "grad_norm": 0.6268947849189795,
      "learning_rate": 2.671580565374282e-06,
      "loss": 0.0344,
      "step": 437
    },
    {
      "epoch": 1.311377245508982,
      "grad_norm": 0.7170945102104322,
      "learning_rate": 2.6507968412221763e-06,
      "loss": 0.0456,
      "step": 438
    },
    {
      "epoch": 1.31437125748503,
      "grad_norm": 0.5731831617064475,
      "learning_rate": 2.6300650767654234e-06,
      "loss": 0.0272,
      "step": 439
    },
    {
      "epoch": 1.3173652694610778,
      "grad_norm": 0.7785193476081234,
      "learning_rate": 2.6093857305493666e-06,
      "loss": 0.0384,
      "step": 440
    },
    {
      "epoch": 1.3203592814371259,
      "grad_norm": 0.6408769611314832,
      "learning_rate": 2.588759259959962e-06,
      "loss": 0.0359,
      "step": 441
    },
    {
      "epoch": 1.3233532934131738,
      "grad_norm": 0.6253860118122385,
      "learning_rate": 2.568186121213658e-06,
      "loss": 0.0299,
      "step": 442
    },
    {
      "epoch": 1.3263473053892216,
      "grad_norm": 0.7909425862208924,
      "learning_rate": 2.547666769347312e-06,
      "loss": 0.0441,
      "step": 443
    },
    {
      "epoch": 1.3293413173652695,
      "grad_norm": 0.6729481178557017,
      "learning_rate": 2.5272016582081236e-06,
      "loss": 0.0343,
      "step": 444
    },
    {
      "epoch": 1.3323353293413174,
      "grad_norm": 0.6638279710265204,
      "learning_rate": 2.5067912404435952e-06,
      "loss": 0.0298,
      "step": 445
    },
    {
      "epoch": 1.3353293413173652,
      "grad_norm": 0.7761199518210663,
      "learning_rate": 2.486435967491516e-06,
      "loss": 0.0357,
      "step": 446
    },
    {
      "epoch": 1.3383233532934131,
      "grad_norm": 0.619637437401186,
      "learning_rate": 2.4661362895699903e-06,
      "loss": 0.0312,
      "step": 447
    },
    {
      "epoch": 1.341317365269461,
      "grad_norm": 0.6633771642078193,
      "learning_rate": 2.445892655667462e-06,
      "loss": 0.0326,
      "step": 448
    },
    {
      "epoch": 1.3443113772455089,
      "grad_norm": 0.6705491467113616,
      "learning_rate": 2.425705513532798e-06,
      "loss": 0.0406,
      "step": 449
    },
    {
      "epoch": 1.347305389221557,
      "grad_norm": 0.6644248101184801,
      "learning_rate": 2.4055753096653795e-06,
      "loss": 0.0367,
      "step": 450
    },
    {
      "epoch": 1.3502994011976048,
      "grad_norm": 0.6092734200107877,
      "learning_rate": 2.3855024893052286e-06,
      "loss": 0.0291,
      "step": 451
    },
    {
      "epoch": 1.3532934131736527,
      "grad_norm": 0.6297964417683745,
      "learning_rate": 2.365487496423152e-06,
      "loss": 0.0345,
      "step": 452
    },
    {
      "epoch": 1.3562874251497006,
      "grad_norm": 0.5974214044277423,
      "learning_rate": 2.3455307737109338e-06,
      "loss": 0.0251,
      "step": 453
    },
    {
      "epoch": 1.3592814371257484,
      "grad_norm": 0.6947225367283539,
      "learning_rate": 2.3256327625715345e-06,
      "loss": 0.0352,
      "step": 454
    },
    {
      "epoch": 1.3622754491017965,
      "grad_norm": 0.514536816513198,
      "learning_rate": 2.3057939031093346e-06,
      "loss": 0.0255,
      "step": 455
    },
    {
      "epoch": 1.3652694610778444,
      "grad_norm": 0.43738409515356963,
      "learning_rate": 2.2860146341203936e-06,
      "loss": 0.0237,
      "step": 456
    },
    {
      "epoch": 1.3682634730538923,
      "grad_norm": 0.5915553891264541,
      "learning_rate": 2.2662953930827546e-06,
      "loss": 0.031,
      "step": 457
    },
    {
      "epoch": 1.3712574850299402,
      "grad_norm": 0.48654889087022873,
      "learning_rate": 2.2466366161467528e-06,
      "loss": 0.0193,
      "step": 458
    },
    {
      "epoch": 1.374251497005988,
      "grad_norm": 0.5958970247583097,
      "learning_rate": 2.227038738125385e-06,
      "loss": 0.0312,
      "step": 459
    },
    {
      "epoch": 1.377245508982036,
      "grad_norm": 0.8099651868744516,
      "learning_rate": 2.207502192484685e-06,
      "loss": 0.0357,
      "step": 460
    },
    {
      "epoch": 1.3802395209580838,
      "grad_norm": 0.45847633096298135,
      "learning_rate": 2.188027411334131e-06,
      "loss": 0.0199,
      "step": 461
    },
    {
      "epoch": 1.3832335329341316,
      "grad_norm": 0.6537978957951901,
      "learning_rate": 2.1686148254171012e-06,
      "loss": 0.032,
      "step": 462
    },
    {
      "epoch": 1.3862275449101795,
      "grad_norm": 0.5247313174683882,
      "learning_rate": 2.1492648641013305e-06,
      "loss": 0.0252,
      "step": 463
    },
    {
      "epoch": 1.3892215568862276,
      "grad_norm": 0.5037368952463198,
      "learning_rate": 2.1299779553694323e-06,
      "loss": 0.0232,
      "step": 464
    },
    {
      "epoch": 1.3922155688622755,
      "grad_norm": 0.6349329206780726,
      "learning_rate": 2.1107545258094135e-06,
      "loss": 0.0297,
      "step": 465
    },
    {
      "epoch": 1.3952095808383234,
      "grad_norm": 0.623675864984584,
      "learning_rate": 2.0915950006052555e-06,
      "loss": 0.0271,
      "step": 466
    },
    {
      "epoch": 1.3982035928143712,
      "grad_norm": 0.5669933361773911,
      "learning_rate": 2.0724998035274947e-06,
      "loss": 0.0234,
      "step": 467
    },
    {
      "epoch": 1.401197604790419,
      "grad_norm": 0.49921963894248095,
      "learning_rate": 2.053469356923865e-06,
      "loss": 0.0232,
      "step": 468
    },
    {
      "epoch": 1.4041916167664672,
      "grad_norm": 0.5203721475858618,
      "learning_rate": 2.0345040817099433e-06,
      "loss": 0.023,
      "step": 469
    },
    {
      "epoch": 1.407185628742515,
      "grad_norm": 0.5814975744772958,
      "learning_rate": 2.0156043973598475e-06,
      "loss": 0.0296,
      "step": 470
    },
    {
      "epoch": 1.410179640718563,
      "grad_norm": 0.63825481799683,
      "learning_rate": 1.996770721896957e-06,
      "loss": 0.0268,
      "step": 471
    },
    {
      "epoch": 1.4131736526946108,
      "grad_norm": 0.47476230492853,
      "learning_rate": 1.9780034718846653e-06,
      "loss": 0.0192,
      "step": 472
    },
    {
      "epoch": 1.4161676646706587,
      "grad_norm": 0.7188814640896392,
      "learning_rate": 1.9593030624171683e-06,
      "loss": 0.0374,
      "step": 473
    },
    {
      "epoch": 1.4191616766467066,
      "grad_norm": 0.7234308861035899,
      "learning_rate": 1.9406699071102774e-06,
      "loss": 0.0352,
      "step": 474
    },
    {
      "epoch": 1.4221556886227544,
      "grad_norm": 0.4490144220271361,
      "learning_rate": 1.9221044180922833e-06,
      "loss": 0.0202,
      "step": 475
    },
    {
      "epoch": 1.4251497005988023,
      "grad_norm": 0.6699395347643555,
      "learning_rate": 1.9036070059948253e-06,
      "loss": 0.0306,
      "step": 476
    },
    {
      "epoch": 1.4281437125748502,
      "grad_norm": 0.7004047251138517,
      "learning_rate": 1.885178079943823e-06,
      "loss": 0.0343,
      "step": 477
    },
    {
      "epoch": 1.4311377245508983,
      "grad_norm": 0.7093888636289429,
      "learning_rate": 1.866818047550419e-06,
      "loss": 0.0382,
      "step": 478
    },
    {
      "epoch": 1.4341317365269461,
      "grad_norm": 0.46378554847252906,
      "learning_rate": 1.8485273149019655e-06,
      "loss": 0.0212,
      "step": 479
    },
    {
      "epoch": 1.437125748502994,
      "grad_norm": 0.5800001380928546,
      "learning_rate": 1.8303062865530407e-06,
      "loss": 0.0257,
      "step": 480
    },
    {
      "epoch": 1.4401197604790419,
      "grad_norm": 0.551981244757976,
      "learning_rate": 1.8121553655165058e-06,
      "loss": 0.0258,
      "step": 481
    },
    {
      "epoch": 1.4431137724550898,
      "grad_norm": 0.6027320140862785,
      "learning_rate": 1.7940749532545832e-06,
      "loss": 0.0305,
      "step": 482
    },
    {
      "epoch": 1.4461077844311379,
      "grad_norm": 0.7162248473244751,
      "learning_rate": 1.7760654496699876e-06,
      "loss": 0.0295,
      "step": 483
    },
    {
      "epoch": 1.4491017964071857,
      "grad_norm": 0.5155423770257815,
      "learning_rate": 1.7581272530970666e-06,
      "loss": 0.0243,
      "step": 484
    },
    {
      "epoch": 1.4520958083832336,
      "grad_norm": 0.5097548721223473,
      "learning_rate": 1.7402607602930106e-06,
      "loss": 0.027,
      "step": 485
    },
    {
      "epoch": 1.4550898203592815,
      "grad_norm": 0.5467178579245988,
      "learning_rate": 1.7224663664290537e-06,
      "loss": 0.0256,
      "step": 486
    },
    {
      "epoch": 1.4580838323353293,
      "grad_norm": 0.5957752274068268,
      "learning_rate": 1.7047444650817518e-06,
      "loss": 0.0277,
      "step": 487
    },
    {
      "epoch": 1.4610778443113772,
      "grad_norm": 0.5372855604949256,
      "learning_rate": 1.6870954482242707e-06,
      "loss": 0.0243,
      "step": 488
    },
    {
      "epoch": 1.464071856287425,
      "grad_norm": 0.6408063298808008,
      "learning_rate": 1.669519706217711e-06,
      "loss": 0.0229,
      "step": 489
    },
    {
      "epoch": 1.467065868263473,
      "grad_norm": 0.804216073216019,
      "learning_rate": 1.652017627802487e-06,
      "loss": 0.0383,
      "step": 490
    },
    {
      "epoch": 1.4700598802395208,
      "grad_norm": 0.5966498038069312,
      "learning_rate": 1.6345896000897122e-06,
      "loss": 0.0241,
      "step": 491
    },
    {
      "epoch": 1.473053892215569,
      "grad_norm": 0.6557245590889959,
      "learning_rate": 1.6172360085526567e-06,
      "loss": 0.0374,
      "step": 492
    },
    {
      "epoch": 1.4760479041916168,
      "grad_norm": 0.6084084963366193,
      "learning_rate": 1.5999572370182016e-06,
      "loss": 0.0318,
      "step": 493
    },
    {
      "epoch": 1.4790419161676647,
      "grad_norm": 0.5823323137174284,
      "learning_rate": 1.5827536676583643e-06,
      "loss": 0.0274,
      "step": 494
    },
    {
      "epoch": 1.4820359281437125,
      "grad_norm": 0.5721312132472964,
      "learning_rate": 1.5656256809818343e-06,
      "loss": 0.0246,
      "step": 495
    },
    {
      "epoch": 1.4850299401197604,
      "grad_norm": 0.5760581017723354,
      "learning_rate": 1.54857365582557e-06,
      "loss": 0.0267,
      "step": 496
    },
    {
      "epoch": 1.4880239520958085,
      "grad_norm": 0.7843178833407437,
      "learning_rate": 1.5315979693464039e-06,
      "loss": 0.0416,
      "step": 497
    },
    {
      "epoch": 1.4910179640718564,
      "grad_norm": 0.5674711279338015,
      "learning_rate": 1.5146989970127158e-06,
      "loss": 0.0258,
      "step": 498
    },
    {
      "epoch": 1.4940119760479043,
      "grad_norm": 0.6449652171183057,
      "learning_rate": 1.4978771125961177e-06,
      "loss": 0.0288,
      "step": 499
    },
    {
      "epoch": 1.4970059880239521,
      "grad_norm": 0.588099563365675,
      "learning_rate": 1.4811326881631937e-06,
      "loss": 0.0307,
      "step": 500
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.6490452726526647,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.0268,
      "step": 501
    },
    {
      "epoch": 1.5029940119760479,
      "grad_norm": 0.8071473384791795,
      "learning_rate": 1.4478776989401949e-06,
      "loss": 0.0321,
      "step": 502
    },
    {
      "epoch": 1.5059880239520957,
      "grad_norm": 0.6152267550348851,
      "learning_rate": 1.4313678696842559e-06,
      "loss": 0.0281,
      "step": 503
    },
    {
      "epoch": 1.5089820359281436,
      "grad_norm": 0.5722716352657163,
      "learning_rate": 1.4149369714639856e-06,
      "loss": 0.0293,
      "step": 504
    },
    {
      "epoch": 1.5119760479041915,
      "grad_norm": 0.6253653205495314,
      "learning_rate": 1.3985853676981316e-06,
      "loss": 0.026,
      "step": 505
    },
    {
      "epoch": 1.5149700598802394,
      "grad_norm": 0.6617074245156088,
      "learning_rate": 1.3823134200516043e-06,
      "loss": 0.0297,
      "step": 506
    },
    {
      "epoch": 1.5179640718562875,
      "grad_norm": 0.6367009177011983,
      "learning_rate": 1.366121488427481e-06,
      "loss": 0.0288,
      "step": 507
    },
    {
      "epoch": 1.5209580838323353,
      "grad_norm": 0.5536892431568364,
      "learning_rate": 1.3500099309590397e-06,
      "loss": 0.0221,
      "step": 508
    },
    {
      "epoch": 1.5239520958083832,
      "grad_norm": 0.5954412993447297,
      "learning_rate": 1.3339791040018479e-06,
      "loss": 0.0254,
      "step": 509
    },
    {
      "epoch": 1.5269461077844313,
      "grad_norm": 0.6357209161595516,
      "learning_rate": 1.3180293621258694e-06,
      "loss": 0.0266,
      "step": 510
    },
    {
      "epoch": 1.5299401197604792,
      "grad_norm": 0.548277873240108,
      "learning_rate": 1.3021610581076316e-06,
      "loss": 0.0256,
      "step": 511
    },
    {
      "epoch": 1.532934131736527,
      "grad_norm": 0.6406384428381766,
      "learning_rate": 1.2863745429224145e-06,
      "loss": 0.025,
      "step": 512
    },
    {
      "epoch": 1.535928143712575,
      "grad_norm": 0.5572561986021211,
      "learning_rate": 1.270670165736499e-06,
      "loss": 0.0251,
      "step": 513
    },
    {
      "epoch": 1.5389221556886228,
      "grad_norm": 0.6237757345698409,
      "learning_rate": 1.2550482738994284e-06,
      "loss": 0.0274,
      "step": 514
    },
    {
      "epoch": 1.5419161676646707,
      "grad_norm": 0.5649263976594836,
      "learning_rate": 1.239509212936343e-06,
      "loss": 0.0289,
      "step": 515
    },
    {
      "epoch": 1.5449101796407185,
      "grad_norm": 0.603795835515118,
      "learning_rate": 1.22405332654032e-06,
      "loss": 0.0284,
      "step": 516
    },
    {
      "epoch": 1.5479041916167664,
      "grad_norm": 0.6292211509271561,
      "learning_rate": 1.2086809565647877e-06,
      "loss": 0.0325,
      "step": 517
    },
    {
      "epoch": 1.5508982035928143,
      "grad_norm": 0.584940623509583,
      "learning_rate": 1.1933924430159571e-06,
      "loss": 0.028,
      "step": 518
    },
    {
      "epoch": 1.5538922155688621,
      "grad_norm": 0.6052978823795032,
      "learning_rate": 1.1781881240452958e-06,
      "loss": 0.0266,
      "step": 519
    },
    {
      "epoch": 1.55688622754491,
      "grad_norm": 0.6589235197600025,
      "learning_rate": 1.1630683359420653e-06,
      "loss": 0.033,
      "step": 520
    },
    {
      "epoch": 1.5598802395209581,
      "grad_norm": 0.5499617308913013,
      "learning_rate": 1.1480334131258626e-06,
      "loss": 0.0237,
      "step": 521
    },
    {
      "epoch": 1.562874251497006,
      "grad_norm": 0.7354307688645517,
      "learning_rate": 1.1330836881392405e-06,
      "loss": 0.0337,
      "step": 522
    },
    {
      "epoch": 1.5658682634730539,
      "grad_norm": 0.5284463794106306,
      "learning_rate": 1.11821949164034e-06,
      "loss": 0.022,
      "step": 523
    },
    {
      "epoch": 1.568862275449102,
      "grad_norm": 0.4737309804945315,
      "learning_rate": 1.103441152395588e-06,
      "loss": 0.0183,
      "step": 524
    },
    {
      "epoch": 1.5718562874251498,
      "grad_norm": 0.7011432548863663,
      "learning_rate": 1.088748997272414e-06,
      "loss": 0.0312,
      "step": 525
    },
    {
      "epoch": 1.5748502994011977,
      "grad_norm": 0.6207774632639615,
      "learning_rate": 1.0741433512320316e-06,
      "loss": 0.0252,
      "step": 526
    },
    {
      "epoch": 1.5778443113772456,
      "grad_norm": 0.5606854066681984,
      "learning_rate": 1.0596245373222424e-06,
      "loss": 0.0222,
      "step": 527
    },
    {
      "epoch": 1.5808383233532934,
      "grad_norm": 0.6349861069383583,
      "learning_rate": 1.045192876670298e-06,
      "loss": 0.026,
      "step": 528
    },
    {
      "epoch": 1.5838323353293413,
      "grad_norm": 0.5858273797564397,
      "learning_rate": 1.0308486884757868e-06,
      "loss": 0.0235,
      "step": 529
    },
    {
      "epoch": 1.5868263473053892,
      "grad_norm": 0.6939159531447912,
      "learning_rate": 1.0165922900035886e-06,
      "loss": 0.0289,
      "step": 530
    },
    {
      "epoch": 1.589820359281437,
      "grad_norm": 0.43538106428338835,
      "learning_rate": 1.0024239965768417e-06,
      "loss": 0.0207,
      "step": 531
    },
    {
      "epoch": 1.592814371257485,
      "grad_norm": 0.5916399536601251,
      "learning_rate": 9.883441215699824e-07,
      "loss": 0.023,
      "step": 532
    },
    {
      "epoch": 1.5958083832335328,
      "grad_norm": 0.7145050641328856,
      "learning_rate": 9.74352976401805e-07,
      "loss": 0.0306,
      "step": 533
    },
    {
      "epoch": 1.5988023952095807,
      "grad_norm": 0.6724029190514954,
      "learning_rate": 9.604508705285765e-07,
      "loss": 0.0285,
      "step": 534
    },
    {
      "epoch": 1.6017964071856288,
      "grad_norm": 0.5290231611789109,
      "learning_rate": 9.466381114371942e-07,
      "loss": 0.0247,
      "step": 535
    },
    {
      "epoch": 1.6047904191616766,
      "grad_norm": 0.7100067677088676,
      "learning_rate": 9.329150046383773e-07,
      "loss": 0.0367,
      "step": 536
    },
    {
      "epoch": 1.6077844311377245,
      "grad_norm": 0.590326010111365,
      "learning_rate": 9.192818536599213e-07,
      "loss": 0.0282,
      "step": 537
    },
    {
      "epoch": 1.6107784431137726,
      "grad_norm": 0.6249586798112137,
      "learning_rate": 9.057389600399719e-07,
      "loss": 0.0284,
      "step": 538
    },
    {
      "epoch": 1.6137724550898205,
      "grad_norm": 0.6334869431531361,
      "learning_rate": 8.922866233203681e-07,
      "loss": 0.0308,
      "step": 539
    },
    {
      "epoch": 1.6167664670658684,
      "grad_norm": 0.5080078055173646,
      "learning_rate": 8.789251410400024e-07,
      "loss": 0.0242,
      "step": 540
    },
    {
      "epoch": 1.6197604790419162,
      "grad_norm": 0.6019174873987064,
      "learning_rate": 8.65654808728259e-07,
      "loss": 0.0289,
      "step": 541
    },
    {
      "epoch": 1.622754491017964,
      "grad_norm": 0.6130442149565656,
      "learning_rate": 8.524759198984567e-07,
      "loss": 0.0254,
      "step": 542
    },
    {
      "epoch": 1.625748502994012,
      "grad_norm": 0.7206470715904347,
      "learning_rate": 8.393887660413719e-07,
      "loss": 0.0327,
      "step": 543
    },
    {
      "epoch": 1.6287425149700598,
      "grad_norm": 0.9006704618146641,
      "learning_rate": 8.263936366187825e-07,
      "loss": 0.0455,
      "step": 544
    },
    {
      "epoch": 1.6317365269461077,
      "grad_norm": 0.5602105919008891,
      "learning_rate": 8.134908190570723e-07,
      "loss": 0.0259,
      "step": 545
    },
    {
      "epoch": 1.6347305389221556,
      "grad_norm": 0.6163730922289473,
      "learning_rate": 8.006805987408705e-07,
      "loss": 0.0285,
      "step": 546
    },
    {
      "epoch": 1.6377245508982035,
      "grad_norm": 0.523684798831296,
      "learning_rate": 7.879632590067354e-07,
      "loss": 0.0241,
      "step": 547
    },
    {
      "epoch": 1.6407185628742516,
      "grad_norm": 0.5714157826411278,
      "learning_rate": 7.753390811368972e-07,
      "loss": 0.0277,
      "step": 548
    },
    {
      "epoch": 1.6437125748502994,
      "grad_norm": 0.7305449368542581,
      "learning_rate": 7.628083443530287e-07,
      "loss": 0.0343,
      "step": 549
    },
    {
      "epoch": 1.6467065868263473,
      "grad_norm": 0.7245934167877404,
      "learning_rate": 7.503713258100726e-07,
      "loss": 0.0317,
      "step": 550
    },
    {
      "epoch": 1.6497005988023952,
      "grad_norm": 0.7043682910702175,
      "learning_rate": 7.380283005901084e-07,
      "loss": 0.0321,
      "step": 551
    },
    {
      "epoch": 1.6526946107784433,
      "grad_norm": 0.5838719786853032,
      "learning_rate": 7.257795416962754e-07,
      "loss": 0.0283,
      "step": 552
    },
    {
      "epoch": 1.6556886227544911,
      "grad_norm": 0.45452535724741283,
      "learning_rate": 7.136253200467231e-07,
      "loss": 0.0148,
      "step": 553
    },
    {
      "epoch": 1.658682634730539,
      "grad_norm": 0.48939458214463044,
      "learning_rate": 7.015659044686307e-07,
      "loss": 0.0201,
      "step": 554
    },
    {
      "epoch": 1.6616766467065869,
      "grad_norm": 0.6391073075953003,
      "learning_rate": 6.896015616922535e-07,
      "loss": 0.0295,
      "step": 555
    },
    {
      "epoch": 1.6646706586826348,
      "grad_norm": 0.7789097833154494,
      "learning_rate": 6.777325563450282e-07,
      "loss": 0.0326,
      "step": 556
    },
    {
      "epoch": 1.6676646706586826,
      "grad_norm": 0.6761947507213802,
      "learning_rate": 6.659591509457125e-07,
      "loss": 0.0295,
      "step": 557
    },
    {
      "epoch": 1.6706586826347305,
      "grad_norm": 0.49096765064102593,
      "learning_rate": 6.542816058985896e-07,
      "loss": 0.0195,
      "step": 558
    },
    {
      "epoch": 1.6736526946107784,
      "grad_norm": 0.5694843433927046,
      "learning_rate": 6.427001794876974e-07,
      "loss": 0.0233,
      "step": 559
    },
    {
      "epoch": 1.6766467065868262,
      "grad_norm": 0.5605406919245851,
      "learning_rate": 6.312151278711237e-07,
      "loss": 0.024,
      "step": 560
    },
    {
      "epoch": 1.6796407185628741,
      "grad_norm": 0.5424311618706831,
      "learning_rate": 6.198267050753387e-07,
      "loss": 0.0242,
      "step": 561
    },
    {
      "epoch": 1.6826347305389222,
      "grad_norm": 0.6431045821651563,
      "learning_rate": 6.085351629895736e-07,
      "loss": 0.0277,
      "step": 562
    },
    {
      "epoch": 1.68562874251497,
      "grad_norm": 0.6768998068357851,
      "learning_rate": 5.973407513602514e-07,
      "loss": 0.0306,
      "step": 563
    },
    {
      "epoch": 1.688622754491018,
      "grad_norm": 0.6637622249363229,
      "learning_rate": 5.862437177854629e-07,
      "loss": 0.0311,
      "step": 564
    },
    {
      "epoch": 1.6916167664670658,
      "grad_norm": 0.8845927965842152,
      "learning_rate": 5.752443077094927e-07,
      "loss": 0.0311,
      "step": 565
    },
    {
      "epoch": 1.694610778443114,
      "grad_norm": 0.552546382035408,
      "learning_rate": 5.643427644173838e-07,
      "loss": 0.0246,
      "step": 566
    },
    {
      "epoch": 1.6976047904191618,
      "grad_norm": 0.6237733171168331,
      "learning_rate": 5.535393290295643e-07,
      "loss": 0.0265,
      "step": 567
    },
    {
      "epoch": 1.7005988023952097,
      "grad_norm": 0.6472498545711429,
      "learning_rate": 5.428342404965076e-07,
      "loss": 0.0306,
      "step": 568
    },
    {
      "epoch": 1.7035928143712575,
      "grad_norm": 0.5154587665621893,
      "learning_rate": 5.322277355934557e-07,
      "loss": 0.021,
      "step": 569
    },
    {
      "epoch": 1.7065868263473054,
      "grad_norm": 0.4791024175484916,
      "learning_rate": 5.217200489151714e-07,
      "loss": 0.0187,
      "step": 570
    },
    {
      "epoch": 1.7095808383233533,
      "grad_norm": 0.5808107840726128,
      "learning_rate": 5.113114128707592e-07,
      "loss": 0.0262,
      "step": 571
    },
    {
      "epoch": 1.7125748502994012,
      "grad_norm": 0.6326519158180665,
      "learning_rate": 5.010020576785174e-07,
      "loss": 0.0324,
      "step": 572
    },
    {
      "epoch": 1.715568862275449,
      "grad_norm": 0.6185489927879538,
      "learning_rate": 4.907922113608532e-07,
      "loss": 0.0251,
      "step": 573
    },
    {
      "epoch": 1.718562874251497,
      "grad_norm": 0.5974432066065593,
      "learning_rate": 4.806820997392325e-07,
      "loss": 0.0269,
      "step": 574
    },
    {
      "epoch": 1.7215568862275448,
      "grad_norm": 0.449869231445647,
      "learning_rate": 4.7067194642919036e-07,
      "loss": 0.0185,
      "step": 575
    },
    {
      "epoch": 1.7245508982035929,
      "grad_norm": 0.5613487273181738,
      "learning_rate": 4.607619728353818e-07,
      "loss": 0.0213,
      "step": 576
    },
    {
      "epoch": 1.7275449101796407,
      "grad_norm": 0.6267465171721688,
      "learning_rate": 4.50952398146689e-07,
      "loss": 0.0295,
      "step": 577
    },
    {
      "epoch": 1.7305389221556886,
      "grad_norm": 0.5184959766256656,
      "learning_rate": 4.4124343933136525e-07,
      "loss": 0.0227,
      "step": 578
    },
    {
      "epoch": 1.7335329341317365,
      "grad_norm": 0.6014093848698644,
      "learning_rate": 4.3163531113224466e-07,
      "loss": 0.0277,
      "step": 579
    },
    {
      "epoch": 1.7365269461077846,
      "grad_norm": 0.8251975760444311,
      "learning_rate": 4.221282260619891e-07,
      "loss": 0.0369,
      "step": 580
    },
    {
      "epoch": 1.7395209580838324,
      "grad_norm": 0.5025143660776764,
      "learning_rate": 4.127223943983849e-07,
      "loss": 0.0262,
      "step": 581
    },
    {
      "epoch": 1.7425149700598803,
      "grad_norm": 0.5338982021625645,
      "learning_rate": 4.03418024179697e-07,
      "loss": 0.0236,
      "step": 582
    },
    {
      "epoch": 1.7455089820359282,
      "grad_norm": 0.721065200986303,
      "learning_rate": 3.9421532120006544e-07,
      "loss": 0.0381,
      "step": 583
    },
    {
      "epoch": 1.748502994011976,
      "grad_norm": 0.5462352934425309,
      "learning_rate": 3.851144890049535e-07,
      "loss": 0.025,
      "step": 584
    },
    {
      "epoch": 1.751497005988024,
      "grad_norm": 0.5609381380961513,
      "learning_rate": 3.761157288866418e-07,
      "loss": 0.0273,
      "step": 585
    },
    {
      "epoch": 1.7544910179640718,
      "grad_norm": 0.6828597948094135,
      "learning_rate": 3.672192398797858e-07,
      "loss": 0.0289,
      "step": 586
    },
    {
      "epoch": 1.7574850299401197,
      "grad_norm": 0.7071438835182291,
      "learning_rate": 3.58425218757002e-07,
      "loss": 0.027,
      "step": 587
    },
    {
      "epoch": 1.7604790419161676,
      "grad_norm": 0.5512666482402668,
      "learning_rate": 3.497338600245254e-07,
      "loss": 0.022,
      "step": 588
    },
    {
      "epoch": 1.7634730538922154,
      "grad_norm": 0.5771884614633598,
      "learning_rate": 3.4114535591790233e-07,
      "loss": 0.0281,
      "step": 589
    },
    {
      "epoch": 1.7664670658682635,
      "grad_norm": 0.7905582497482192,
      "learning_rate": 3.326598963977395e-07,
      "loss": 0.0305,
      "step": 590
    },
    {
      "epoch": 1.7694610778443114,
      "grad_norm": 0.5876172627837322,
      "learning_rate": 3.242776691455013e-07,
      "loss": 0.0264,
      "step": 591
    },
    {
      "epoch": 1.7724550898203593,
      "grad_norm": 0.5521657100135328,
      "learning_rate": 3.159988595593616e-07,
      "loss": 0.024,
      "step": 592
    },
    {
      "epoch": 1.7754491017964071,
      "grad_norm": 0.4733053022335487,
      "learning_rate": 3.078236507501015e-07,
      "loss": 0.0244,
      "step": 593
    },
    {
      "epoch": 1.7784431137724552,
      "grad_norm": 0.6973036153863111,
      "learning_rate": 2.9975222353705757e-07,
      "loss": 0.0326,
      "step": 594
    },
    {
      "epoch": 1.781437125748503,
      "grad_norm": 0.5940394347647507,
      "learning_rate": 2.917847564441256e-07,
      "loss": 0.0248,
      "step": 595
    },
    {
      "epoch": 1.784431137724551,
      "grad_norm": 0.603276325967987,
      "learning_rate": 2.839214256958106e-07,
      "loss": 0.0253,
      "step": 596
    },
    {
      "epoch": 1.7874251497005988,
      "grad_norm": 0.5535699613155315,
      "learning_rate": 2.7616240521332884e-07,
      "loss": 0.0238,
      "step": 597
    },
    {
      "epoch": 1.7904191616766467,
      "grad_norm": 0.6113014188786479,
      "learning_rate": 2.6850786661076047e-07,
      "loss": 0.0268,
      "step": 598
    },
    {
      "epoch": 1.7934131736526946,
      "grad_norm": 0.502529774058875,
      "learning_rate": 2.6095797919125533e-07,
      "loss": 0.0215,
      "step": 599
    },
    {
      "epoch": 1.7964071856287425,
      "grad_norm": 0.4669390990436676,
      "learning_rate": 2.5351290994328703e-07,
      "loss": 0.0213,
      "step": 600
    },
    {
      "epoch": 1.7964071856287425,
      "eval_loss": 0.11722549051046371,
      "eval_runtime": 8.1085,
      "eval_samples_per_second": 6.66,
      "eval_steps_per_second": 1.727,
      "step": 600
    },
    {
      "epoch": 1.7994011976047903,
      "grad_norm": 0.5400390995338958,
      "learning_rate": 2.4617282353696093e-07,
      "loss": 0.0258,
      "step": 601
    },
    {
      "epoch": 1.8023952095808382,
      "grad_norm": 0.8005101072010351,
      "learning_rate": 2.3893788232036807e-07,
      "loss": 0.0361,
      "step": 602
    },
    {
      "epoch": 1.805389221556886,
      "grad_norm": 0.6303667413965671,
      "learning_rate": 2.318082463160032e-07,
      "loss": 0.0329,
      "step": 603
    },
    {
      "epoch": 1.8083832335329342,
      "grad_norm": 0.5322324323955108,
      "learning_rate": 2.2478407321721295e-07,
      "loss": 0.0301,
      "step": 604
    },
    {
      "epoch": 1.811377245508982,
      "grad_norm": 0.5737041350884969,
      "learning_rate": 2.1786551838471892e-07,
      "loss": 0.0294,
      "step": 605
    },
    {
      "epoch": 1.81437125748503,
      "grad_norm": 0.500239192417263,
      "learning_rate": 2.1105273484317402e-07,
      "loss": 0.0226,
      "step": 606
    },
    {
      "epoch": 1.8173652694610778,
      "grad_norm": 0.46725859713572976,
      "learning_rate": 2.043458732777831e-07,
      "loss": 0.0198,
      "step": 607
    },
    {
      "epoch": 1.8203592814371259,
      "grad_norm": 0.7965919351101484,
      "learning_rate": 1.9774508203096843e-07,
      "loss": 0.036,
      "step": 608
    },
    {
      "epoch": 1.8233532934131738,
      "grad_norm": 0.41438342762939184,
      "learning_rate": 1.9125050709908388e-07,
      "loss": 0.0151,
      "step": 609
    },
    {
      "epoch": 1.8263473053892216,
      "grad_norm": 0.5841035393467985,
      "learning_rate": 1.8486229212919482e-07,
      "loss": 0.0258,
      "step": 610
    },
    {
      "epoch": 1.8293413173652695,
      "grad_norm": 0.5694000966408276,
      "learning_rate": 1.7858057841589281e-07,
      "loss": 0.0243,
      "step": 611
    },
    {
      "epoch": 1.8323353293413174,
      "grad_norm": 0.5301254029113557,
      "learning_rate": 1.7240550489817652e-07,
      "loss": 0.022,
      "step": 612
    },
    {
      "epoch": 1.8353293413173652,
      "grad_norm": 0.573042269872994,
      "learning_rate": 1.66337208156373e-07,
      "loss": 0.0221,
      "step": 613
    },
    {
      "epoch": 1.8383233532934131,
      "grad_norm": 0.6548288294045421,
      "learning_rate": 1.6037582240912175e-07,
      "loss": 0.0278,
      "step": 614
    },
    {
      "epoch": 1.841317365269461,
      "grad_norm": 0.6629443680209643,
      "learning_rate": 1.5452147951040165e-07,
      "loss": 0.0294,
      "step": 615
    },
    {
      "epoch": 1.8443113772455089,
      "grad_norm": 0.608563047821321,
      "learning_rate": 1.4877430894662037e-07,
      "loss": 0.0266,
      "step": 616
    },
    {
      "epoch": 1.8473053892215567,
      "grad_norm": 0.601956390390165,
      "learning_rate": 1.4313443783374405e-07,
      "loss": 0.0286,
      "step": 617
    },
    {
      "epoch": 1.8502994011976048,
      "grad_norm": 0.5433711867396067,
      "learning_rate": 1.3760199091449045e-07,
      "loss": 0.0214,
      "step": 618
    },
    {
      "epoch": 1.8532934131736527,
      "grad_norm": 0.6234281640927353,
      "learning_rate": 1.3217709055556638e-07,
      "loss": 0.0302,
      "step": 619
    },
    {
      "epoch": 1.8562874251497006,
      "grad_norm": 0.6313565432499837,
      "learning_rate": 1.268598567449647e-07,
      "loss": 0.0266,
      "step": 620
    },
    {
      "epoch": 1.8592814371257484,
      "grad_norm": 0.5722130308833123,
      "learning_rate": 1.2165040708930763e-07,
      "loss": 0.0266,
      "step": 621
    },
    {
      "epoch": 1.8622754491017965,
      "grad_norm": 0.43014735505016405,
      "learning_rate": 1.1654885681124661e-07,
      "loss": 0.0204,
      "step": 622
    },
    {
      "epoch": 1.8652694610778444,
      "grad_norm": 0.5305475132536474,
      "learning_rate": 1.1155531874691372e-07,
      "loss": 0.023,
      "step": 623
    },
    {
      "epoch": 1.8682634730538923,
      "grad_norm": 0.6147246210703419,
      "learning_rate": 1.0666990334342708e-07,
      "loss": 0.0255,
      "step": 624
    },
    {
      "epoch": 1.8712574850299402,
      "grad_norm": 0.4893731601337044,
      "learning_rate": 1.0189271865644445e-07,
      "loss": 0.0188,
      "step": 625
    },
    {
      "epoch": 1.874251497005988,
      "grad_norm": 0.647380007719925,
      "learning_rate": 9.722387034777847e-08,
      "loss": 0.027,
      "step": 626
    },
    {
      "epoch": 1.877245508982036,
      "grad_norm": 0.6069960434886762,
      "learning_rate": 9.266346168305518e-08,
      "loss": 0.0276,
      "step": 627
    },
    {
      "epoch": 1.8802395209580838,
      "grad_norm": 0.508612509960378,
      "learning_rate": 8.821159352943142e-08,
      "loss": 0.0212,
      "step": 628
    },
    {
      "epoch": 1.8832335329341316,
      "grad_norm": 0.525549449768915,
      "learning_rate": 8.38683643533661e-08,
      "loss": 0.0197,
      "step": 629
    },
    {
      "epoch": 1.8862275449101795,
      "grad_norm": 0.595433388646461,
      "learning_rate": 7.963387021843683e-08,
      "loss": 0.0232,
      "step": 630
    },
    {
      "epoch": 1.8892215568862274,
      "grad_norm": 0.633683780909806,
      "learning_rate": 7.550820478322285e-08,
      "loss": 0.0283,
      "step": 631
    },
    {
      "epoch": 1.8922155688622755,
      "grad_norm": 0.5681296045644338,
      "learning_rate": 7.149145929922607e-08,
      "loss": 0.0322,
      "step": 632
    },
    {
      "epoch": 1.8952095808383234,
      "grad_norm": 0.6428469179083691,
      "learning_rate": 6.758372260885714e-08,
      "loss": 0.0274,
      "step": 633
    },
    {
      "epoch": 1.8982035928143712,
      "grad_norm": 0.6709835738162027,
      "learning_rate": 6.378508114346982e-08,
      "loss": 0.0317,
      "step": 634
    },
    {
      "epoch": 1.9011976047904193,
      "grad_norm": 0.5337875321059655,
      "learning_rate": 6.009561892144744e-08,
      "loss": 0.0236,
      "step": 635
    },
    {
      "epoch": 1.9041916167664672,
      "grad_norm": 0.5805824139798283,
      "learning_rate": 5.651541754634726e-08,
      "loss": 0.0248,
      "step": 636
    },
    {
      "epoch": 1.907185628742515,
      "grad_norm": 0.6979227682580144,
      "learning_rate": 5.304455620509297e-08,
      "loss": 0.0327,
      "step": 637
    },
    {
      "epoch": 1.910179640718563,
      "grad_norm": 0.5789599966754212,
      "learning_rate": 4.968311166622553e-08,
      "loss": 0.0264,
      "step": 638
    },
    {
      "epoch": 1.9131736526946108,
      "grad_norm": 0.7031123325231106,
      "learning_rate": 4.643115827820399e-08,
      "loss": 0.0313,
      "step": 639
    },
    {
      "epoch": 1.9161676646706587,
      "grad_norm": 0.5673402405927495,
      "learning_rate": 4.328876796776071e-08,
      "loss": 0.0215,
      "step": 640
    },
    {
      "epoch": 1.9191616766467066,
      "grad_norm": 0.6431003067996562,
      "learning_rate": 4.0256010238310936e-08,
      "loss": 0.0336,
      "step": 641
    },
    {
      "epoch": 1.9221556886227544,
      "grad_norm": 0.5999957257163617,
      "learning_rate": 3.733295216841626e-08,
      "loss": 0.0268,
      "step": 642
    },
    {
      "epoch": 1.9251497005988023,
      "grad_norm": 0.6854285275369748,
      "learning_rate": 3.451965841029914e-08,
      "loss": 0.0287,
      "step": 643
    },
    {
      "epoch": 1.9281437125748502,
      "grad_norm": 0.48113373261459796,
      "learning_rate": 3.181619118841517e-08,
      "loss": 0.0196,
      "step": 644
    },
    {
      "epoch": 1.931137724550898,
      "grad_norm": 0.5180192727431109,
      "learning_rate": 2.9222610298074717e-08,
      "loss": 0.0211,
      "step": 645
    },
    {
      "epoch": 1.9341317365269461,
      "grad_norm": 0.6152776910708702,
      "learning_rate": 2.673897310412288e-08,
      "loss": 0.0207,
      "step": 646
    },
    {
      "epoch": 1.937125748502994,
      "grad_norm": 0.51092306660443,
      "learning_rate": 2.4365334539667717e-08,
      "loss": 0.0197,
      "step": 647
    },
    {
      "epoch": 1.9401197604790419,
      "grad_norm": 0.6004281839070754,
      "learning_rate": 2.210174710486679e-08,
      "loss": 0.0268,
      "step": 648
    },
    {
      "epoch": 1.94311377245509,
      "grad_norm": 0.5080853706084675,
      "learning_rate": 1.99482608657664e-08,
      "loss": 0.0223,
      "step": 649
    },
    {
      "epoch": 1.9461077844311379,
      "grad_norm": 0.5765693450203558,
      "learning_rate": 1.7904923453193056e-08,
      "loss": 0.0257,
      "step": 650
    },
    {
      "epoch": 1.9491017964071857,
      "grad_norm": 0.615183228781466,
      "learning_rate": 1.5971780061701524e-08,
      "loss": 0.0266,
      "step": 651
    },
    {
      "epoch": 1.9520958083832336,
      "grad_norm": 0.5494659296305936,
      "learning_rate": 1.4148873448573408e-08,
      "loss": 0.0255,
      "step": 652
    },
    {
      "epoch": 1.9550898203592815,
      "grad_norm": 0.6023489062647699,
      "learning_rate": 1.2436243932872349e-08,
      "loss": 0.0281,
      "step": 653
    },
    {
      "epoch": 1.9580838323353293,
      "grad_norm": 0.5370693118796576,
      "learning_rate": 1.0833929394552523e-08,
      "loss": 0.0191,
      "step": 654
    },
    {
      "epoch": 1.9610778443113772,
      "grad_norm": 0.4975445952595503,
      "learning_rate": 9.341965273621522e-09,
      "loss": 0.0216,
      "step": 655
    },
    {
      "epoch": 1.964071856287425,
      "grad_norm": 0.6985668051103773,
      "learning_rate": 7.96038456935322e-09,
      "loss": 0.0322,
      "step": 656
    },
    {
      "epoch": 1.967065868263473,
      "grad_norm": 0.7397360703330154,
      "learning_rate": 6.6892178395611125e-09,
      "loss": 0.0337,
      "step": 657
    },
    {
      "epoch": 1.9700598802395208,
      "grad_norm": 0.5924284276764363,
      "learning_rate": 5.528493199922769e-09,
      "loss": 0.0256,
      "step": 658
    },
    {
      "epoch": 1.9730538922155687,
      "grad_norm": 0.5879974051742082,
      "learning_rate": 4.478236323355312e-09,
      "loss": 0.0285,
      "step": 659
    },
    {
      "epoch": 1.9760479041916168,
      "grad_norm": 0.5847065188408496,
      "learning_rate": 3.538470439448105e-09,
      "loss": 0.0246,
      "step": 660
    },
    {
      "epoch": 1.9790419161676647,
      "grad_norm": 0.7291276124667593,
      "learning_rate": 2.709216333952602e-09,
      "loss": 0.0358,
      "step": 661
    },
    {
      "epoch": 1.9820359281437125,
      "grad_norm": 0.5590061616056822,
      "learning_rate": 1.9904923483171632e-09,
      "loss": 0.0272,
      "step": 662
    },
    {
      "epoch": 1.9850299401197606,
      "grad_norm": 0.811564010293061,
      "learning_rate": 1.3823143792851545e-09,
      "loss": 0.0312,
      "step": 663
    },
    {
      "epoch": 1.9880239520958085,
      "grad_norm": 0.7491076241903057,
      "learning_rate": 8.846958785418969e-10,
      "loss": 0.0359,
      "step": 664
    },
    {
      "epoch": 1.9910179640718564,
      "grad_norm": 0.4851659101440372,
      "learning_rate": 4.97647852417682e-10,
      "loss": 0.0179,
      "step": 665
    },
    {
      "epoch": 1.9940119760479043,
      "grad_norm": 0.5342896224203642,
      "learning_rate": 2.2117886164407797e-10,
      "loss": 0.021,
      "step": 666
    },
    {
      "epoch": 1.9970059880239521,
      "grad_norm": 0.5555583617167489,
      "learning_rate": 5.529502116519148e-11,
      "loss": 0.0227,
      "step": 667
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.3006977946789636,
      "learning_rate": 0.0,
      "loss": 0.0089,
      "step": 668
    },
    {
      "epoch": 2.0,
      "step": 668,
      "total_flos": 16797260906496.0,
      "train_loss": 0.06560490493144015,
      "train_runtime": 1761.391,
      "train_samples_per_second": 6.051,
      "train_steps_per_second": 0.379
    }
  ],
  "logging_steps": 1,
  "max_steps": 668,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 16797260906496.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}