{
  "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": 1.4565563281991678,
      "learning_rate": 9.999944704978835e-06,
      "loss": 0.1837,
      "step": 1
    },
    {
      "epoch": 0.005988023952095809,
      "grad_norm": 1.183167965042078,
      "learning_rate": 9.999778821138357e-06,
      "loss": 0.1364,
      "step": 2
    },
    {
      "epoch": 0.008982035928143712,
      "grad_norm": 1.3390988282572531,
      "learning_rate": 9.999502352147583e-06,
      "loss": 0.1622,
      "step": 3
    },
    {
      "epoch": 0.011976047904191617,
      "grad_norm": 1.0512689771986117,
      "learning_rate": 9.999115304121459e-06,
      "loss": 0.137,
      "step": 4
    },
    {
      "epoch": 0.014970059880239521,
      "grad_norm": 1.045247208899895,
      "learning_rate": 9.998617685620715e-06,
      "loss": 0.1208,
      "step": 5
    },
    {
      "epoch": 0.017964071856287425,
      "grad_norm": 0.9798033591423202,
      "learning_rate": 9.998009507651683e-06,
      "loss": 0.1122,
      "step": 6
    },
    {
      "epoch": 0.020958083832335328,
      "grad_norm": 1.0207228000097202,
      "learning_rate": 9.997290783666048e-06,
      "loss": 0.1183,
      "step": 7
    },
    {
      "epoch": 0.023952095808383235,
      "grad_norm": 1.0101345879935173,
      "learning_rate": 9.996461529560553e-06,
      "loss": 0.1129,
      "step": 8
    },
    {
      "epoch": 0.02694610778443114,
      "grad_norm": 1.0367172230176434,
      "learning_rate": 9.995521763676645e-06,
      "loss": 0.1241,
      "step": 9
    },
    {
      "epoch": 0.029940119760479042,
      "grad_norm": 1.130152212247823,
      "learning_rate": 9.994471506800078e-06,
      "loss": 0.1249,
      "step": 10
    },
    {
      "epoch": 0.03293413173652695,
      "grad_norm": 1.0587430085413858,
      "learning_rate": 9.993310782160439e-06,
      "loss": 0.1285,
      "step": 11
    },
    {
      "epoch": 0.03592814371257485,
      "grad_norm": 1.0659967533822203,
      "learning_rate": 9.992039615430648e-06,
      "loss": 0.109,
      "step": 12
    },
    {
      "epoch": 0.038922155688622756,
      "grad_norm": 1.1275207164033734,
      "learning_rate": 9.99065803472638e-06,
      "loss": 0.1265,
      "step": 13
    },
    {
      "epoch": 0.041916167664670656,
      "grad_norm": 1.1649048885371962,
      "learning_rate": 9.989166070605447e-06,
      "loss": 0.1238,
      "step": 14
    },
    {
      "epoch": 0.04491017964071856,
      "grad_norm": 1.187652317279918,
      "learning_rate": 9.98756375606713e-06,
      "loss": 0.1079,
      "step": 15
    },
    {
      "epoch": 0.04790419161676647,
      "grad_norm": 1.2192695986992406,
      "learning_rate": 9.985851126551428e-06,
      "loss": 0.1275,
      "step": 16
    },
    {
      "epoch": 0.05089820359281437,
      "grad_norm": 0.9113435044539926,
      "learning_rate": 9.9840282199383e-06,
      "loss": 0.088,
      "step": 17
    },
    {
      "epoch": 0.05389221556886228,
      "grad_norm": 1.0834350443620688,
      "learning_rate": 9.982095076546806e-06,
      "loss": 0.1233,
      "step": 18
    },
    {
      "epoch": 0.05688622754491018,
      "grad_norm": 1.509180281705903,
      "learning_rate": 9.980051739134235e-06,
      "loss": 0.1455,
      "step": 19
    },
    {
      "epoch": 0.059880239520958084,
      "grad_norm": 1.1160626774320201,
      "learning_rate": 9.977898252895133e-06,
      "loss": 0.1282,
      "step": 20
    },
    {
      "epoch": 0.06287425149700598,
      "grad_norm": 1.0763564863366484,
      "learning_rate": 9.975634665460333e-06,
      "loss": 0.1232,
      "step": 21
    },
    {
      "epoch": 0.0658682634730539,
      "grad_norm": 1.0842486566157548,
      "learning_rate": 9.973261026895878e-06,
      "loss": 0.1272,
      "step": 22
    },
    {
      "epoch": 0.0688622754491018,
      "grad_norm": 1.0065287648343513,
      "learning_rate": 9.970777389701927e-06,
      "loss": 0.119,
      "step": 23
    },
    {
      "epoch": 0.0718562874251497,
      "grad_norm": 0.9454785274529562,
      "learning_rate": 9.968183808811586e-06,
      "loss": 0.1008,
      "step": 24
    },
    {
      "epoch": 0.0748502994011976,
      "grad_norm": 1.1242200061443328,
      "learning_rate": 9.965480341589702e-06,
      "loss": 0.1187,
      "step": 25
    },
    {
      "epoch": 0.07784431137724551,
      "grad_norm": 1.147073867847353,
      "learning_rate": 9.962667047831585e-06,
      "loss": 0.1214,
      "step": 26
    },
    {
      "epoch": 0.08083832335329341,
      "grad_norm": 1.010808606049857,
      "learning_rate": 9.95974398976169e-06,
      "loss": 0.1282,
      "step": 27
    },
    {
      "epoch": 0.08383233532934131,
      "grad_norm": 0.9694893331347081,
      "learning_rate": 9.95671123203224e-06,
      "loss": 0.1215,
      "step": 28
    },
    {
      "epoch": 0.08682634730538923,
      "grad_norm": 1.0736465370589077,
      "learning_rate": 9.953568841721796e-06,
      "loss": 0.1148,
      "step": 29
    },
    {
      "epoch": 0.08982035928143713,
      "grad_norm": 0.9595988006078322,
      "learning_rate": 9.950316888333775e-06,
      "loss": 0.1193,
      "step": 30
    },
    {
      "epoch": 0.09281437125748503,
      "grad_norm": 0.9620649224044232,
      "learning_rate": 9.946955443794908e-06,
      "loss": 0.1149,
      "step": 31
    },
    {
      "epoch": 0.09580838323353294,
      "grad_norm": 1.1567025313341026,
      "learning_rate": 9.943484582453653e-06,
      "loss": 0.1155,
      "step": 32
    },
    {
      "epoch": 0.09880239520958084,
      "grad_norm": 1.1408648365708318,
      "learning_rate": 9.939904381078553e-06,
      "loss": 0.1102,
      "step": 33
    },
    {
      "epoch": 0.10179640718562874,
      "grad_norm": 1.2170324761734317,
      "learning_rate": 9.93621491885653e-06,
      "loss": 0.1285,
      "step": 34
    },
    {
      "epoch": 0.10479041916167664,
      "grad_norm": 1.0450413690468956,
      "learning_rate": 9.932416277391144e-06,
      "loss": 0.1076,
      "step": 35
    },
    {
      "epoch": 0.10778443113772455,
      "grad_norm": 1.0856790690055609,
      "learning_rate": 9.928508540700775e-06,
      "loss": 0.1261,
      "step": 36
    },
    {
      "epoch": 0.11077844311377245,
      "grad_norm": 1.3238890044914724,
      "learning_rate": 9.924491795216777e-06,
      "loss": 0.1404,
      "step": 37
    },
    {
      "epoch": 0.11377245508982035,
      "grad_norm": 1.0955730348561874,
      "learning_rate": 9.920366129781564e-06,
      "loss": 0.1126,
      "step": 38
    },
    {
      "epoch": 0.11676646706586827,
      "grad_norm": 0.9469494313265368,
      "learning_rate": 9.916131635646635e-06,
      "loss": 0.0994,
      "step": 39
    },
    {
      "epoch": 0.11976047904191617,
      "grad_norm": 1.0865222762341304,
      "learning_rate": 9.91178840647057e-06,
      "loss": 0.1211,
      "step": 40
    },
    {
      "epoch": 0.12275449101796407,
      "grad_norm": 1.036649079559057,
      "learning_rate": 9.907336538316946e-06,
      "loss": 0.1315,
      "step": 41
    },
    {
      "epoch": 0.12574850299401197,
      "grad_norm": 0.7936120439677845,
      "learning_rate": 9.902776129652223e-06,
      "loss": 0.0823,
      "step": 42
    },
    {
      "epoch": 0.12874251497005987,
      "grad_norm": 1.0527901511318694,
      "learning_rate": 9.898107281343557e-06,
      "loss": 0.1134,
      "step": 43
    },
    {
      "epoch": 0.1317365269461078,
      "grad_norm": 1.4361570022204402,
      "learning_rate": 9.893330096656576e-06,
      "loss": 0.0956,
      "step": 44
    },
    {
      "epoch": 0.1347305389221557,
      "grad_norm": 1.0763053385857533,
      "learning_rate": 9.888444681253087e-06,
      "loss": 0.1138,
      "step": 45
    },
    {
      "epoch": 0.1377245508982036,
      "grad_norm": 1.1199723656690888,
      "learning_rate": 9.883451143188753e-06,
      "loss": 0.1232,
      "step": 46
    },
    {
      "epoch": 0.1407185628742515,
      "grad_norm": 0.9634543522044158,
      "learning_rate": 9.878349592910694e-06,
      "loss": 0.1083,
      "step": 47
    },
    {
      "epoch": 0.1437125748502994,
      "grad_norm": 1.1299050700662392,
      "learning_rate": 9.873140143255035e-06,
      "loss": 0.1315,
      "step": 48
    },
    {
      "epoch": 0.1467065868263473,
      "grad_norm": 1.0994022927849498,
      "learning_rate": 9.867822909444435e-06,
      "loss": 0.1305,
      "step": 49
    },
    {
      "epoch": 0.1497005988023952,
      "grad_norm": 1.0139674554360978,
      "learning_rate": 9.862398009085511e-06,
      "loss": 0.1103,
      "step": 50
    },
    {
      "epoch": 0.15269461077844312,
      "grad_norm": 1.0375287813360328,
      "learning_rate": 9.856865562166256e-06,
      "loss": 0.1236,
      "step": 51
    },
    {
      "epoch": 0.15568862275449102,
      "grad_norm": 0.9095710398476978,
      "learning_rate": 9.851225691053382e-06,
      "loss": 0.1003,
      "step": 52
    },
    {
      "epoch": 0.15868263473053892,
      "grad_norm": 1.1000563283808396,
      "learning_rate": 9.8454785204896e-06,
      "loss": 0.1435,
      "step": 53
    },
    {
      "epoch": 0.16167664670658682,
      "grad_norm": 0.9326192185773918,
      "learning_rate": 9.83962417759088e-06,
      "loss": 0.1082,
      "step": 54
    },
    {
      "epoch": 0.16467065868263472,
      "grad_norm": 1.0078544072281366,
      "learning_rate": 9.833662791843628e-06,
      "loss": 0.1152,
      "step": 55
    },
    {
      "epoch": 0.16766467065868262,
      "grad_norm": 0.9686081607085508,
      "learning_rate": 9.827594495101824e-06,
      "loss": 0.1152,
      "step": 56
    },
    {
      "epoch": 0.17065868263473055,
      "grad_norm": 1.0716294646801823,
      "learning_rate": 9.821419421584108e-06,
      "loss": 0.1225,
      "step": 57
    },
    {
      "epoch": 0.17365269461077845,
      "grad_norm": 1.1985712422019703,
      "learning_rate": 9.815137707870806e-06,
      "loss": 0.1677,
      "step": 58
    },
    {
      "epoch": 0.17664670658682635,
      "grad_norm": 1.0547712668426101,
      "learning_rate": 9.808749492900917e-06,
      "loss": 0.1217,
      "step": 59
    },
    {
      "epoch": 0.17964071856287425,
      "grad_norm": 1.0785984616133262,
      "learning_rate": 9.802254917969033e-06,
      "loss": 0.117,
      "step": 60
    },
    {
      "epoch": 0.18263473053892215,
      "grad_norm": 1.1545999083283733,
      "learning_rate": 9.795654126722218e-06,
      "loss": 0.1342,
      "step": 61
    },
    {
      "epoch": 0.18562874251497005,
      "grad_norm": 0.9915808193089987,
      "learning_rate": 9.788947265156828e-06,
      "loss": 0.1108,
      "step": 62
    },
    {
      "epoch": 0.18862275449101795,
      "grad_norm": 1.0019930068421536,
      "learning_rate": 9.782134481615282e-06,
      "loss": 0.0992,
      "step": 63
    },
    {
      "epoch": 0.19161676646706588,
      "grad_norm": 0.9914893952456957,
      "learning_rate": 9.775215926782788e-06,
      "loss": 0.1194,
      "step": 64
    },
    {
      "epoch": 0.19461077844311378,
      "grad_norm": 1.0943459239908753,
      "learning_rate": 9.768191753683997e-06,
      "loss": 0.1205,
      "step": 65
    },
    {
      "epoch": 0.19760479041916168,
      "grad_norm": 0.9564830427773949,
      "learning_rate": 9.761062117679632e-06,
      "loss": 0.1158,
      "step": 66
    },
    {
      "epoch": 0.20059880239520958,
      "grad_norm": 0.9743806827699126,
      "learning_rate": 9.75382717646304e-06,
      "loss": 0.1059,
      "step": 67
    },
    {
      "epoch": 0.20359281437125748,
      "grad_norm": 0.9795625090860776,
      "learning_rate": 9.746487090056712e-06,
      "loss": 0.105,
      "step": 68
    },
    {
      "epoch": 0.20658682634730538,
      "grad_norm": 0.9576197352568366,
      "learning_rate": 9.739042020808746e-06,
      "loss": 0.1059,
      "step": 69
    },
    {
      "epoch": 0.20958083832335328,
      "grad_norm": 1.0493648926443717,
      "learning_rate": 9.73149213338924e-06,
      "loss": 0.1238,
      "step": 70
    },
    {
      "epoch": 0.2125748502994012,
      "grad_norm": 1.1448437967273744,
      "learning_rate": 9.72383759478667e-06,
      "loss": 0.135,
      "step": 71
    },
    {
      "epoch": 0.2155688622754491,
      "grad_norm": 0.979501439472502,
      "learning_rate": 9.71607857430419e-06,
      "loss": 0.0877,
      "step": 72
    },
    {
      "epoch": 0.218562874251497,
      "grad_norm": 1.183654575618766,
      "learning_rate": 9.708215243555875e-06,
      "loss": 0.1353,
      "step": 73
    },
    {
      "epoch": 0.2215568862275449,
      "grad_norm": 1.0164597370301052,
      "learning_rate": 9.700247776462944e-06,
      "loss": 0.0998,
      "step": 74
    },
    {
      "epoch": 0.2245508982035928,
      "grad_norm": 1.083368147554738,
      "learning_rate": 9.6921763492499e-06,
      "loss": 0.113,
      "step": 75
    },
    {
      "epoch": 0.2275449101796407,
      "grad_norm": 0.8342061274177972,
      "learning_rate": 9.68400114044064e-06,
      "loss": 0.0763,
      "step": 76
    },
    {
      "epoch": 0.23053892215568864,
      "grad_norm": 1.1040905652766737,
      "learning_rate": 9.6757223308545e-06,
      "loss": 0.101,
      "step": 77
    },
    {
      "epoch": 0.23353293413173654,
      "grad_norm": 1.0879272643485558,
      "learning_rate": 9.667340103602263e-06,
      "loss": 0.1235,
      "step": 78
    },
    {
      "epoch": 0.23652694610778444,
      "grad_norm": 1.098637890385762,
      "learning_rate": 9.658854644082099e-06,
      "loss": 0.1174,
      "step": 79
    },
    {
      "epoch": 0.23952095808383234,
      "grad_norm": 1.3016810746558902,
      "learning_rate": 9.650266139975474e-06,
      "loss": 0.1505,
      "step": 80
    },
    {
      "epoch": 0.24251497005988024,
      "grad_norm": 0.9554741428098311,
      "learning_rate": 9.641574781242999e-06,
      "loss": 0.1085,
      "step": 81
    },
    {
      "epoch": 0.24550898203592814,
      "grad_norm": 0.9681073762322365,
      "learning_rate": 9.632780760120217e-06,
      "loss": 0.1039,
      "step": 82
    },
    {
      "epoch": 0.24850299401197604,
      "grad_norm": 0.9014333792133626,
      "learning_rate": 9.62388427111336e-06,
      "loss": 0.1035,
      "step": 83
    },
    {
      "epoch": 0.25149700598802394,
      "grad_norm": 0.8847294118382342,
      "learning_rate": 9.614885510995047e-06,
      "loss": 0.0967,
      "step": 84
    },
    {
      "epoch": 0.25449101796407186,
      "grad_norm": 0.9092977779848154,
      "learning_rate": 9.605784678799934e-06,
      "loss": 0.0998,
      "step": 85
    },
    {
      "epoch": 0.25748502994011974,
      "grad_norm": 1.1650310802743007,
      "learning_rate": 9.596581975820304e-06,
      "loss": 0.1133,
      "step": 86
    },
    {
      "epoch": 0.26047904191616766,
      "grad_norm": 1.0182667693678538,
      "learning_rate": 9.587277605601617e-06,
      "loss": 0.1089,
      "step": 87
    },
    {
      "epoch": 0.2634730538922156,
      "grad_norm": 1.2161864912320763,
      "learning_rate": 9.577871773938013e-06,
      "loss": 0.1293,
      "step": 88
    },
    {
      "epoch": 0.26646706586826346,
      "grad_norm": 1.0901862296293205,
      "learning_rate": 9.568364688867757e-06,
      "loss": 0.1342,
      "step": 89
    },
    {
      "epoch": 0.2694610778443114,
      "grad_norm": 0.8526953110757662,
      "learning_rate": 9.558756560668637e-06,
      "loss": 0.0946,
      "step": 90
    },
    {
      "epoch": 0.27245508982035926,
      "grad_norm": 1.0463300859545857,
      "learning_rate": 9.549047601853313e-06,
      "loss": 0.1176,
      "step": 91
    },
    {
      "epoch": 0.2754491017964072,
      "grad_norm": 0.983674734438536,
      "learning_rate": 9.539238027164618e-06,
      "loss": 0.1113,
      "step": 92
    },
    {
      "epoch": 0.27844311377245506,
      "grad_norm": 0.9184212949483579,
      "learning_rate": 9.52932805357081e-06,
      "loss": 0.1027,
      "step": 93
    },
    {
      "epoch": 0.281437125748503,
      "grad_norm": 1.0973428051212963,
      "learning_rate": 9.519317900260769e-06,
      "loss": 0.1108,
      "step": 94
    },
    {
      "epoch": 0.2844311377245509,
      "grad_norm": 0.9436609528417036,
      "learning_rate": 9.509207788639148e-06,
      "loss": 0.1058,
      "step": 95
    },
    {
      "epoch": 0.2874251497005988,
      "grad_norm": 1.0137075719809734,
      "learning_rate": 9.498997942321484e-06,
      "loss": 0.1217,
      "step": 96
    },
    {
      "epoch": 0.2904191616766467,
      "grad_norm": 0.91644347984344,
      "learning_rate": 9.488688587129243e-06,
      "loss": 0.1056,
      "step": 97
    },
    {
      "epoch": 0.2934131736526946,
      "grad_norm": 0.9221589997690847,
      "learning_rate": 9.47827995108483e-06,
      "loss": 0.0897,
      "step": 98
    },
    {
      "epoch": 0.2964071856287425,
      "grad_norm": 1.09361970573248,
      "learning_rate": 9.467772264406545e-06,
      "loss": 0.1195,
      "step": 99
    },
    {
      "epoch": 0.2994011976047904,
      "grad_norm": 0.948150908471286,
      "learning_rate": 9.457165759503492e-06,
      "loss": 0.1016,
      "step": 100
    },
    {
      "epoch": 0.3023952095808383,
      "grad_norm": 0.9794533558441048,
      "learning_rate": 9.446460670970436e-06,
      "loss": 0.0913,
      "step": 101
    },
    {
      "epoch": 0.30538922155688625,
      "grad_norm": 0.9666370647661606,
      "learning_rate": 9.435657235582616e-06,
      "loss": 0.0895,
      "step": 102
    },
    {
      "epoch": 0.3083832335329341,
      "grad_norm": 1.2119140566811797,
      "learning_rate": 9.424755692290507e-06,
      "loss": 0.1195,
      "step": 103
    },
    {
      "epoch": 0.31137724550898205,
      "grad_norm": 1.0786819144691129,
      "learning_rate": 9.413756282214538e-06,
      "loss": 0.1125,
      "step": 104
    },
    {
      "epoch": 0.3143712574850299,
      "grad_norm": 0.9758053589028417,
      "learning_rate": 9.402659248639749e-06,
      "loss": 0.1102,
      "step": 105
    },
    {
      "epoch": 0.31736526946107785,
      "grad_norm": 1.1469397794163783,
      "learning_rate": 9.391464837010428e-06,
      "loss": 0.1173,
      "step": 106
    },
    {
      "epoch": 0.3203592814371258,
      "grad_norm": 1.1526787990107483,
      "learning_rate": 9.380173294924661e-06,
      "loss": 0.1152,
      "step": 107
    },
    {
      "epoch": 0.32335329341317365,
      "grad_norm": 1.0119658992992808,
      "learning_rate": 9.368784872128877e-06,
      "loss": 0.1186,
      "step": 108
    },
    {
      "epoch": 0.3263473053892216,
      "grad_norm": 1.084392932793967,
      "learning_rate": 9.357299820512305e-06,
      "loss": 0.1138,
      "step": 109
    },
    {
      "epoch": 0.32934131736526945,
      "grad_norm": 1.2080935258649363,
      "learning_rate": 9.345718394101412e-06,
      "loss": 0.1281,
      "step": 110
    },
    {
      "epoch": 0.3323353293413174,
      "grad_norm": 1.0686462512730956,
      "learning_rate": 9.334040849054288e-06,
      "loss": 0.1326,
      "step": 111
    },
    {
      "epoch": 0.33532934131736525,
      "grad_norm": 1.1763239967073622,
      "learning_rate": 9.322267443654974e-06,
      "loss": 0.1233,
      "step": 112
    },
    {
      "epoch": 0.3383233532934132,
      "grad_norm": 1.0669285308719239,
      "learning_rate": 9.310398438307747e-06,
      "loss": 0.1399,
      "step": 113
    },
    {
      "epoch": 0.3413173652694611,
      "grad_norm": 0.9622867137038135,
      "learning_rate": 9.29843409553137e-06,
      "loss": 0.1113,
      "step": 114
    },
    {
      "epoch": 0.344311377245509,
      "grad_norm": 0.9470980049143565,
      "learning_rate": 9.286374679953278e-06,
      "loss": 0.1094,
      "step": 115
    },
    {
      "epoch": 0.3473053892215569,
      "grad_norm": 0.9087824234238306,
      "learning_rate": 9.274220458303727e-06,
      "loss": 0.1005,
      "step": 116
    },
    {
      "epoch": 0.3502994011976048,
      "grad_norm": 1.0629362988116193,
      "learning_rate": 9.261971699409893e-06,
      "loss": 0.1225,
      "step": 117
    },
    {
      "epoch": 0.3532934131736527,
      "grad_norm": 1.07027198629993,
      "learning_rate": 9.249628674189928e-06,
      "loss": 0.1056,
      "step": 118
    },
    {
      "epoch": 0.3562874251497006,
      "grad_norm": 1.0443476283726563,
      "learning_rate": 9.237191655646972e-06,
      "loss": 0.1229,
      "step": 119
    },
    {
      "epoch": 0.3592814371257485,
      "grad_norm": 1.1201045391165656,
      "learning_rate": 9.224660918863104e-06,
      "loss": 0.1157,
      "step": 120
    },
    {
      "epoch": 0.36227544910179643,
      "grad_norm": 1.0241080437182333,
      "learning_rate": 9.212036740993265e-06,
      "loss": 0.1177,
      "step": 121
    },
    {
      "epoch": 0.3652694610778443,
      "grad_norm": 1.1349529960400069,
      "learning_rate": 9.199319401259132e-06,
      "loss": 0.1038,
      "step": 122
    },
    {
      "epoch": 0.36826347305389223,
      "grad_norm": 0.9926658634925102,
      "learning_rate": 9.186509180942928e-06,
      "loss": 0.0945,
      "step": 123
    },
    {
      "epoch": 0.3712574850299401,
      "grad_norm": 1.0436938345867341,
      "learning_rate": 9.173606363381218e-06,
      "loss": 0.1239,
      "step": 124
    },
    {
      "epoch": 0.37425149700598803,
      "grad_norm": 1.1043829371944485,
      "learning_rate": 9.16061123395863e-06,
      "loss": 0.1064,
      "step": 125
    },
    {
      "epoch": 0.3772455089820359,
      "grad_norm": 0.998543128875388,
      "learning_rate": 9.147524080101543e-06,
      "loss": 0.101,
      "step": 126
    },
    {
      "epoch": 0.38023952095808383,
      "grad_norm": 1.148348296809121,
      "learning_rate": 9.134345191271742e-06,
      "loss": 0.1239,
      "step": 127
    },
    {
      "epoch": 0.38323353293413176,
      "grad_norm": 1.0549601361780823,
      "learning_rate": 9.121074858959997e-06,
      "loss": 0.1224,
      "step": 128
    },
    {
      "epoch": 0.38622754491017963,
      "grad_norm": 1.0685269752609785,
      "learning_rate": 9.107713376679634e-06,
      "loss": 0.1161,
      "step": 129
    },
    {
      "epoch": 0.38922155688622756,
      "grad_norm": 1.0592159227486182,
      "learning_rate": 9.094261039960028e-06,
      "loss": 0.1082,
      "step": 130
    },
    {
      "epoch": 0.39221556886227543,
      "grad_norm": 1.104142960908252,
      "learning_rate": 9.08071814634008e-06,
      "loss": 0.1237,
      "step": 131
    },
    {
      "epoch": 0.39520958083832336,
      "grad_norm": 1.0110292896443436,
      "learning_rate": 9.067084995361623e-06,
      "loss": 0.1105,
      "step": 132
    },
    {
      "epoch": 0.39820359281437123,
      "grad_norm": 1.081705602875668,
      "learning_rate": 9.053361888562807e-06,
      "loss": 0.1338,
      "step": 133
    },
    {
      "epoch": 0.40119760479041916,
      "grad_norm": 1.0369384710858438,
      "learning_rate": 9.039549129471423e-06,
      "loss": 0.113,
      "step": 134
    },
    {
      "epoch": 0.4041916167664671,
      "grad_norm": 0.9226765243356801,
      "learning_rate": 9.025647023598196e-06,
      "loss": 0.0981,
      "step": 135
    },
    {
      "epoch": 0.40718562874251496,
      "grad_norm": 0.9340677171842876,
      "learning_rate": 9.011655878430018e-06,
      "loss": 0.0997,
      "step": 136
    },
    {
      "epoch": 0.4101796407185629,
      "grad_norm": 1.270724209221927,
      "learning_rate": 8.99757600342316e-06,
      "loss": 0.1447,
      "step": 137
    },
    {
      "epoch": 0.41317365269461076,
      "grad_norm": 1.067887314351562,
      "learning_rate": 8.983407709996415e-06,
      "loss": 0.106,
      "step": 138
    },
    {
      "epoch": 0.4161676646706587,
      "grad_norm": 1.0175694200798246,
      "learning_rate": 8.969151311524215e-06,
      "loss": 0.0991,
      "step": 139
    },
    {
      "epoch": 0.41916167664670656,
      "grad_norm": 1.1200183593585993,
      "learning_rate": 8.954807123329703e-06,
      "loss": 0.1273,
      "step": 140
    },
    {
      "epoch": 0.4221556886227545,
      "grad_norm": 1.0441084995718002,
      "learning_rate": 8.940375462677758e-06,
      "loss": 0.1105,
      "step": 141
    },
    {
      "epoch": 0.4251497005988024,
      "grad_norm": 1.042609903826092,
      "learning_rate": 8.92585664876797e-06,
      "loss": 0.1278,
      "step": 142
    },
    {
      "epoch": 0.4281437125748503,
      "grad_norm": 0.940143994553627,
      "learning_rate": 8.911251002727588e-06,
      "loss": 0.1021,
      "step": 143
    },
    {
      "epoch": 0.4311377245508982,
      "grad_norm": 0.8857544344326819,
      "learning_rate": 8.896558847604414e-06,
      "loss": 0.1017,
      "step": 144
    },
    {
      "epoch": 0.4341317365269461,
      "grad_norm": 1.103174234265638,
      "learning_rate": 8.881780508359661e-06,
      "loss": 0.1323,
      "step": 145
    },
    {
      "epoch": 0.437125748502994,
      "grad_norm": 1.102944185408898,
      "learning_rate": 8.86691631186076e-06,
      "loss": 0.1177,
      "step": 146
    },
    {
      "epoch": 0.44011976047904194,
      "grad_norm": 0.8662232623826593,
      "learning_rate": 8.851966586874138e-06,
      "loss": 0.0963,
      "step": 147
    },
    {
      "epoch": 0.4431137724550898,
      "grad_norm": 1.109309375524132,
      "learning_rate": 8.836931664057935e-06,
      "loss": 0.1251,
      "step": 148
    },
    {
      "epoch": 0.44610778443113774,
      "grad_norm": 1.0771699518191358,
      "learning_rate": 8.821811875954705e-06,
      "loss": 0.1088,
      "step": 149
    },
    {
      "epoch": 0.4491017964071856,
      "grad_norm": 0.9948286355545556,
      "learning_rate": 8.806607556984045e-06,
      "loss": 0.1163,
      "step": 150
    },
    {
      "epoch": 0.45209580838323354,
      "grad_norm": 1.091835951195638,
      "learning_rate": 8.791319043435213e-06,
      "loss": 0.1152,
      "step": 151
    },
    {
      "epoch": 0.4550898203592814,
      "grad_norm": 1.0431010872314552,
      "learning_rate": 8.775946673459682e-06,
      "loss": 0.1142,
      "step": 152
    },
    {
      "epoch": 0.45808383233532934,
      "grad_norm": 0.9564669470984967,
      "learning_rate": 8.76049078706366e-06,
      "loss": 0.0953,
      "step": 153
    },
    {
      "epoch": 0.46107784431137727,
      "grad_norm": 1.0599561894997371,
      "learning_rate": 8.744951726100572e-06,
      "loss": 0.1115,
      "step": 154
    },
    {
      "epoch": 0.46407185628742514,
      "grad_norm": 0.9135495134343601,
      "learning_rate": 8.729329834263503e-06,
      "loss": 0.0933,
      "step": 155
    },
    {
      "epoch": 0.46706586826347307,
      "grad_norm": 1.1143943845603541,
      "learning_rate": 8.713625457077585e-06,
      "loss": 0.1294,
      "step": 156
    },
    {
      "epoch": 0.47005988023952094,
      "grad_norm": 1.0786904427098798,
      "learning_rate": 8.697838941892371e-06,
      "loss": 0.1211,
      "step": 157
    },
    {
      "epoch": 0.47305389221556887,
      "grad_norm": 0.9737051150772273,
      "learning_rate": 8.681970637874131e-06,
      "loss": 0.1111,
      "step": 158
    },
    {
      "epoch": 0.47604790419161674,
      "grad_norm": 1.0843082377825357,
      "learning_rate": 8.666020895998154e-06,
      "loss": 0.1229,
      "step": 159
    },
    {
      "epoch": 0.47904191616766467,
      "grad_norm": 0.9538342484613052,
      "learning_rate": 8.64999006904096e-06,
      "loss": 0.0974,
      "step": 160
    },
    {
      "epoch": 0.4820359281437126,
      "grad_norm": 1.1737120964177783,
      "learning_rate": 8.63387851157252e-06,
      "loss": 0.125,
      "step": 161
    },
    {
      "epoch": 0.48502994011976047,
      "grad_norm": 1.0389874087016928,
      "learning_rate": 8.617686579948396e-06,
      "loss": 0.1068,
      "step": 162
    },
    {
      "epoch": 0.4880239520958084,
      "grad_norm": 1.2102635535549298,
      "learning_rate": 8.60141463230187e-06,
      "loss": 0.1173,
      "step": 163
    },
    {
      "epoch": 0.49101796407185627,
      "grad_norm": 1.139265592429286,
      "learning_rate": 8.585063028536015e-06,
      "loss": 0.1157,
      "step": 164
    },
    {
      "epoch": 0.4940119760479042,
      "grad_norm": 1.282263541531333,
      "learning_rate": 8.568632130315747e-06,
      "loss": 0.1385,
      "step": 165
    },
    {
      "epoch": 0.49700598802395207,
      "grad_norm": 1.0623386825329544,
      "learning_rate": 8.552122301059807e-06,
      "loss": 0.1028,
      "step": 166
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.954855262173226,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.0983,
      "step": 167
    },
    {
      "epoch": 0.5029940119760479,
      "grad_norm": 0.9579004232164834,
      "learning_rate": 8.518867311836808e-06,
      "loss": 0.1214,
      "step": 168
    },
    {
      "epoch": 0.5059880239520959,
      "grad_norm": 0.9935373897225895,
      "learning_rate": 8.502122887403882e-06,
      "loss": 0.0927,
      "step": 169
    },
    {
      "epoch": 0.5089820359281437,
      "grad_norm": 0.9997647527702106,
      "learning_rate": 8.485301002987285e-06,
      "loss": 0.0983,
      "step": 170
    },
    {
      "epoch": 0.5119760479041916,
      "grad_norm": 1.1310369225978931,
      "learning_rate": 8.468402030653598e-06,
      "loss": 0.1287,
      "step": 171
    },
    {
      "epoch": 0.5149700598802395,
      "grad_norm": 1.0063391488062465,
      "learning_rate": 8.451426344174433e-06,
      "loss": 0.1142,
      "step": 172
    },
    {
      "epoch": 0.5179640718562875,
      "grad_norm": 1.148675600762412,
      "learning_rate": 8.434374319018165e-06,
      "loss": 0.1134,
      "step": 173
    },
    {
      "epoch": 0.5209580838323353,
      "grad_norm": 0.9967497261825257,
      "learning_rate": 8.417246332341638e-06,
      "loss": 0.1001,
      "step": 174
    },
    {
      "epoch": 0.5239520958083832,
      "grad_norm": 1.0481845812586297,
      "learning_rate": 8.4000427629818e-06,
      "loss": 0.114,
      "step": 175
    },
    {
      "epoch": 0.5269461077844312,
      "grad_norm": 0.9372842280880583,
      "learning_rate": 8.382763991447344e-06,
      "loss": 0.1033,
      "step": 176
    },
    {
      "epoch": 0.5299401197604791,
      "grad_norm": 1.1307128148280887,
      "learning_rate": 8.365410399910287e-06,
      "loss": 0.1213,
      "step": 177
    },
    {
      "epoch": 0.5329341317365269,
      "grad_norm": 1.1680319124045293,
      "learning_rate": 8.347982372197515e-06,
      "loss": 0.121,
      "step": 178
    },
    {
      "epoch": 0.5359281437125748,
      "grad_norm": 1.0876863860091988,
      "learning_rate": 8.33048029378229e-06,
      "loss": 0.1107,
      "step": 179
    },
    {
      "epoch": 0.5389221556886228,
      "grad_norm": 1.0923534282965148,
      "learning_rate": 8.312904551775731e-06,
      "loss": 0.1315,
      "step": 180
    },
    {
      "epoch": 0.5419161676646707,
      "grad_norm": 1.094636782367177,
      "learning_rate": 8.295255534918249e-06,
      "loss": 0.1435,
      "step": 181
    },
    {
      "epoch": 0.5449101796407185,
      "grad_norm": 1.2079386786669601,
      "learning_rate": 8.277533633570948e-06,
      "loss": 0.1272,
      "step": 182
    },
    {
      "epoch": 0.5479041916167665,
      "grad_norm": 1.0693584372030143,
      "learning_rate": 8.25973923970699e-06,
      "loss": 0.1189,
      "step": 183
    },
    {
      "epoch": 0.5508982035928144,
      "grad_norm": 1.0658342004567258,
      "learning_rate": 8.241872746902934e-06,
      "loss": 0.1279,
      "step": 184
    },
    {
      "epoch": 0.5538922155688623,
      "grad_norm": 1.0405521185483353,
      "learning_rate": 8.223934550330015e-06,
      "loss": 0.1219,
      "step": 185
    },
    {
      "epoch": 0.5568862275449101,
      "grad_norm": 1.045840903250419,
      "learning_rate": 8.20592504674542e-06,
      "loss": 0.1221,
      "step": 186
    },
    {
      "epoch": 0.5598802395209581,
      "grad_norm": 1.2068937495703023,
      "learning_rate": 8.187844634483495e-06,
      "loss": 0.139,
      "step": 187
    },
    {
      "epoch": 0.562874251497006,
      "grad_norm": 1.1364781765467205,
      "learning_rate": 8.16969371344696e-06,
      "loss": 0.122,
      "step": 188
    },
    {
      "epoch": 0.5658682634730539,
      "grad_norm": 1.0388526660336852,
      "learning_rate": 8.151472685098037e-06,
      "loss": 0.1241,
      "step": 189
    },
    {
      "epoch": 0.5688622754491018,
      "grad_norm": 0.9480095062265537,
      "learning_rate": 8.13318195244958e-06,
      "loss": 0.1022,
      "step": 190
    },
    {
      "epoch": 0.5718562874251497,
      "grad_norm": 0.9896318927807755,
      "learning_rate": 8.114821920056177e-06,
      "loss": 0.103,
      "step": 191
    },
    {
      "epoch": 0.5748502994011976,
      "grad_norm": 1.1575058675823335,
      "learning_rate": 8.096392994005177e-06,
      "loss": 0.1317,
      "step": 192
    },
    {
      "epoch": 0.5778443113772455,
      "grad_norm": 0.948348969798947,
      "learning_rate": 8.077895581907719e-06,
      "loss": 0.1103,
      "step": 193
    },
    {
      "epoch": 0.5808383233532934,
      "grad_norm": 0.9763401317214996,
      "learning_rate": 8.059330092889724e-06,
      "loss": 0.1127,
      "step": 194
    },
    {
      "epoch": 0.5838323353293413,
      "grad_norm": 1.0049384892416582,
      "learning_rate": 8.040696937582833e-06,
      "loss": 0.1224,
      "step": 195
    },
    {
      "epoch": 0.5868263473053892,
      "grad_norm": 0.8980649572844476,
      "learning_rate": 8.021996528115335e-06,
      "loss": 0.0875,
      "step": 196
    },
    {
      "epoch": 0.5898203592814372,
      "grad_norm": 1.087962029666497,
      "learning_rate": 8.003229278103044e-06,
      "loss": 0.1078,
      "step": 197
    },
    {
      "epoch": 0.592814371257485,
      "grad_norm": 1.0377550987702207,
      "learning_rate": 7.984395602640153e-06,
      "loss": 0.114,
      "step": 198
    },
    {
      "epoch": 0.5958083832335329,
      "grad_norm": 1.0635839702202123,
      "learning_rate": 7.96549591829006e-06,
      "loss": 0.1103,
      "step": 199
    },
    {
      "epoch": 0.5988023952095808,
      "grad_norm": 0.9167338876578826,
      "learning_rate": 7.946530643076138e-06,
      "loss": 0.099,
      "step": 200
    },
    {
      "epoch": 0.5988023952095808,
      "eval_loss": 0.1256195604801178,
      "eval_runtime": 3.3785,
      "eval_samples_per_second": 15.983,
      "eval_steps_per_second": 4.144,
      "step": 200
    },
    {
      "epoch": 0.6017964071856288,
      "grad_norm": 0.8664511744045093,
      "learning_rate": 7.927500196472506e-06,
      "loss": 0.1034,
      "step": 201
    },
    {
      "epoch": 0.6047904191616766,
      "grad_norm": 1.0586315329378382,
      "learning_rate": 7.908404999394747e-06,
      "loss": 0.0996,
      "step": 202
    },
    {
      "epoch": 0.6077844311377245,
      "grad_norm": 1.0875849921751974,
      "learning_rate": 7.889245474190588e-06,
      "loss": 0.12,
      "step": 203
    },
    {
      "epoch": 0.6107784431137725,
      "grad_norm": 1.334416212835032,
      "learning_rate": 7.870022044630569e-06,
      "loss": 0.1448,
      "step": 204
    },
    {
      "epoch": 0.6137724550898204,
      "grad_norm": 1.1074861671140752,
      "learning_rate": 7.85073513589867e-06,
      "loss": 0.1203,
      "step": 205
    },
    {
      "epoch": 0.6167664670658682,
      "grad_norm": 1.0868962630101164,
      "learning_rate": 7.831385174582901e-06,
      "loss": 0.1075,
      "step": 206
    },
    {
      "epoch": 0.6197604790419161,
      "grad_norm": 0.9659015426541392,
      "learning_rate": 7.81197258866587e-06,
      "loss": 0.0961,
      "step": 207
    },
    {
      "epoch": 0.6227544910179641,
      "grad_norm": 1.033799032979925,
      "learning_rate": 7.792497807515317e-06,
      "loss": 0.1059,
      "step": 208
    },
    {
      "epoch": 0.625748502994012,
      "grad_norm": 1.0252774669182994,
      "learning_rate": 7.772961261874615e-06,
      "loss": 0.102,
      "step": 209
    },
    {
      "epoch": 0.6287425149700598,
      "grad_norm": 1.1383967475560077,
      "learning_rate": 7.75336338385325e-06,
      "loss": 0.1142,
      "step": 210
    },
    {
      "epoch": 0.6317365269461078,
      "grad_norm": 1.1491756158384079,
      "learning_rate": 7.733704606917248e-06,
      "loss": 0.1431,
      "step": 211
    },
    {
      "epoch": 0.6347305389221557,
      "grad_norm": 1.0742923214672528,
      "learning_rate": 7.713985365879607e-06,
      "loss": 0.0998,
      "step": 212
    },
    {
      "epoch": 0.6377245508982036,
      "grad_norm": 1.2501119023358702,
      "learning_rate": 7.694206096890667e-06,
      "loss": 0.1378,
      "step": 213
    },
    {
      "epoch": 0.6407185628742516,
      "grad_norm": 0.9964603921522615,
      "learning_rate": 7.674367237428467e-06,
      "loss": 0.1063,
      "step": 214
    },
    {
      "epoch": 0.6437125748502994,
      "grad_norm": 1.0314940005343953,
      "learning_rate": 7.654469226289068e-06,
      "loss": 0.1027,
      "step": 215
    },
    {
      "epoch": 0.6467065868263473,
      "grad_norm": 1.1265261165933578,
      "learning_rate": 7.63451250357685e-06,
      "loss": 0.1259,
      "step": 216
    },
    {
      "epoch": 0.6497005988023952,
      "grad_norm": 1.0170538886910856,
      "learning_rate": 7.614497510694774e-06,
      "loss": 0.1032,
      "step": 217
    },
    {
      "epoch": 0.6526946107784432,
      "grad_norm": 1.136674114591733,
      "learning_rate": 7.5944246903346204e-06,
      "loss": 0.1213,
      "step": 218
    },
    {
      "epoch": 0.655688622754491,
      "grad_norm": 0.8789875697749602,
      "learning_rate": 7.574294486467204e-06,
      "loss": 0.0837,
      "step": 219
    },
    {
      "epoch": 0.6586826347305389,
      "grad_norm": 0.9704801692308667,
      "learning_rate": 7.55410734433254e-06,
      "loss": 0.1022,
      "step": 220
    },
    {
      "epoch": 0.6616766467065869,
      "grad_norm": 1.1008196901458656,
      "learning_rate": 7.533863710430011e-06,
      "loss": 0.1247,
      "step": 221
    },
    {
      "epoch": 0.6646706586826348,
      "grad_norm": 0.7644935151701994,
      "learning_rate": 7.513564032508484e-06,
      "loss": 0.0847,
      "step": 222
    },
    {
      "epoch": 0.6676646706586826,
      "grad_norm": 1.150699339666722,
      "learning_rate": 7.493208759556406e-06,
      "loss": 0.117,
      "step": 223
    },
    {
      "epoch": 0.6706586826347305,
      "grad_norm": 1.085883024702509,
      "learning_rate": 7.472798341791877e-06,
      "loss": 0.1193,
      "step": 224
    },
    {
      "epoch": 0.6736526946107785,
      "grad_norm": 0.903487787259262,
      "learning_rate": 7.452333230652688e-06,
      "loss": 0.085,
      "step": 225
    },
    {
      "epoch": 0.6766467065868264,
      "grad_norm": 1.0130653556007712,
      "learning_rate": 7.431813878786343e-06,
      "loss": 0.1178,
      "step": 226
    },
    {
      "epoch": 0.6796407185628742,
      "grad_norm": 0.9755702371572815,
      "learning_rate": 7.4112407400400395e-06,
      "loss": 0.0917,
      "step": 227
    },
    {
      "epoch": 0.6826347305389222,
      "grad_norm": 0.8749466119747884,
      "learning_rate": 7.390614269450633e-06,
      "loss": 0.0875,
      "step": 228
    },
    {
      "epoch": 0.6856287425149701,
      "grad_norm": 1.031135610871034,
      "learning_rate": 7.369934923234577e-06,
      "loss": 0.101,
      "step": 229
    },
    {
      "epoch": 0.688622754491018,
      "grad_norm": 1.0690108023399494,
      "learning_rate": 7.349203158777826e-06,
      "loss": 0.1033,
      "step": 230
    },
    {
      "epoch": 0.6916167664670658,
      "grad_norm": 0.9708996480411576,
      "learning_rate": 7.32841943462572e-06,
      "loss": 0.1076,
      "step": 231
    },
    {
      "epoch": 0.6946107784431138,
      "grad_norm": 1.1269412575120337,
      "learning_rate": 7.3075842104728445e-06,
      "loss": 0.1139,
      "step": 232
    },
    {
      "epoch": 0.6976047904191617,
      "grad_norm": 1.1159837404230848,
      "learning_rate": 7.286697947152868e-06,
      "loss": 0.1271,
      "step": 233
    },
    {
      "epoch": 0.7005988023952096,
      "grad_norm": 0.8680430386244788,
      "learning_rate": 7.265761106628338e-06,
      "loss": 0.09,
      "step": 234
    },
    {
      "epoch": 0.7035928143712575,
      "grad_norm": 0.977272525168135,
      "learning_rate": 7.244774151980466e-06,
      "loss": 0.0986,
      "step": 235
    },
    {
      "epoch": 0.7065868263473054,
      "grad_norm": 1.1140799289466325,
      "learning_rate": 7.223737547398898e-06,
      "loss": 0.1106,
      "step": 236
    },
    {
      "epoch": 0.7095808383233533,
      "grad_norm": 0.9922672901193285,
      "learning_rate": 7.20265175817143e-06,
      "loss": 0.1089,
      "step": 237
    },
    {
      "epoch": 0.7125748502994012,
      "grad_norm": 1.4589131098129662,
      "learning_rate": 7.181517250673729e-06,
      "loss": 0.144,
      "step": 238
    },
    {
      "epoch": 0.7155688622754491,
      "grad_norm": 1.0069323431296096,
      "learning_rate": 7.1603344923590065e-06,
      "loss": 0.1092,
      "step": 239
    },
    {
      "epoch": 0.718562874251497,
      "grad_norm": 1.186965684505586,
      "learning_rate": 7.139103951747694e-06,
      "loss": 0.1389,
      "step": 240
    },
    {
      "epoch": 0.7215568862275449,
      "grad_norm": 0.9938171279154443,
      "learning_rate": 7.1178260984170675e-06,
      "loss": 0.1054,
      "step": 241
    },
    {
      "epoch": 0.7245508982035929,
      "grad_norm": 0.962440271792613,
      "learning_rate": 7.0965014029908654e-06,
      "loss": 0.0926,
      "step": 242
    },
    {
      "epoch": 0.7275449101796407,
      "grad_norm": 1.0791914167884313,
      "learning_rate": 7.075130337128883e-06,
      "loss": 0.1402,
      "step": 243
    },
    {
      "epoch": 0.7305389221556886,
      "grad_norm": 1.0968245158962497,
      "learning_rate": 7.053713373516538e-06,
      "loss": 0.1312,
      "step": 244
    },
    {
      "epoch": 0.7335329341317365,
      "grad_norm": 1.0284510184553015,
      "learning_rate": 7.03225098585441e-06,
      "loss": 0.1156,
      "step": 245
    },
    {
      "epoch": 0.7365269461077845,
      "grad_norm": 1.0389515559840237,
      "learning_rate": 7.0107436488477694e-06,
      "loss": 0.1286,
      "step": 246
    },
    {
      "epoch": 0.7395209580838323,
      "grad_norm": 1.1254734051202833,
      "learning_rate": 6.989191838196083e-06,
      "loss": 0.1191,
      "step": 247
    },
    {
      "epoch": 0.7425149700598802,
      "grad_norm": 0.9039234029643801,
      "learning_rate": 6.9675960305824785e-06,
      "loss": 0.0927,
      "step": 248
    },
    {
      "epoch": 0.7455089820359282,
      "grad_norm": 1.1785172208605987,
      "learning_rate": 6.945956703663212e-06,
      "loss": 0.1212,
      "step": 249
    },
    {
      "epoch": 0.7485029940119761,
      "grad_norm": 1.0821709166255045,
      "learning_rate": 6.9242743360570985e-06,
      "loss": 0.1134,
      "step": 250
    },
    {
      "epoch": 0.7514970059880239,
      "grad_norm": 1.118572535436202,
      "learning_rate": 6.9025494073349284e-06,
      "loss": 0.1173,
      "step": 251
    },
    {
      "epoch": 0.7544910179640718,
      "grad_norm": 0.9927944497825903,
      "learning_rate": 6.880782398008862e-06,
      "loss": 0.1138,
      "step": 252
    },
    {
      "epoch": 0.7574850299401198,
      "grad_norm": 0.9642461902275179,
      "learning_rate": 6.858973789521792e-06,
      "loss": 0.095,
      "step": 253
    },
    {
      "epoch": 0.7604790419161677,
      "grad_norm": 1.0086007012394462,
      "learning_rate": 6.837124064236709e-06,
      "loss": 0.1027,
      "step": 254
    },
    {
      "epoch": 0.7634730538922155,
      "grad_norm": 0.8797110400475275,
      "learning_rate": 6.815233705426019e-06,
      "loss": 0.0942,
      "step": 255
    },
    {
      "epoch": 0.7664670658682635,
      "grad_norm": 1.0319920107601583,
      "learning_rate": 6.7933031972608644e-06,
      "loss": 0.105,
      "step": 256
    },
    {
      "epoch": 0.7694610778443114,
      "grad_norm": 1.1523453051478574,
      "learning_rate": 6.771333024800411e-06,
      "loss": 0.1185,
      "step": 257
    },
    {
      "epoch": 0.7724550898203593,
      "grad_norm": 1.1681299495998154,
      "learning_rate": 6.74932367398112e-06,
      "loss": 0.1185,
      "step": 258
    },
    {
      "epoch": 0.7754491017964071,
      "grad_norm": 0.9112813647284537,
      "learning_rate": 6.727275631605996e-06,
      "loss": 0.0935,
      "step": 259
    },
    {
      "epoch": 0.7784431137724551,
      "grad_norm": 1.0802578221361256,
      "learning_rate": 6.70518938533383e-06,
      "loss": 0.1121,
      "step": 260
    },
    {
      "epoch": 0.781437125748503,
      "grad_norm": 0.9871659331759355,
      "learning_rate": 6.683065423668403e-06,
      "loss": 0.0934,
      "step": 261
    },
    {
      "epoch": 0.7844311377245509,
      "grad_norm": 1.0661501271120575,
      "learning_rate": 6.660904235947687e-06,
      "loss": 0.1154,
      "step": 262
    },
    {
      "epoch": 0.7874251497005988,
      "grad_norm": 0.9705254665567562,
      "learning_rate": 6.638706312333018e-06,
      "loss": 0.0986,
      "step": 263
    },
    {
      "epoch": 0.7904191616766467,
      "grad_norm": 1.1320298305968832,
      "learning_rate": 6.61647214379826e-06,
      "loss": 0.1157,
      "step": 264
    },
    {
      "epoch": 0.7934131736526946,
      "grad_norm": 1.2289557476542912,
      "learning_rate": 6.594202222118941e-06,
      "loss": 0.1335,
      "step": 265
    },
    {
      "epoch": 0.7964071856287425,
      "grad_norm": 0.9723456121342514,
      "learning_rate": 6.571897039861377e-06,
      "loss": 0.1008,
      "step": 266
    },
    {
      "epoch": 0.7994011976047904,
      "grad_norm": 1.0266061119235725,
      "learning_rate": 6.549557090371775e-06,
      "loss": 0.118,
      "step": 267
    },
    {
      "epoch": 0.8023952095808383,
      "grad_norm": 0.8587114235980591,
      "learning_rate": 6.527182867765333e-06,
      "loss": 0.0846,
      "step": 268
    },
    {
      "epoch": 0.8053892215568862,
      "grad_norm": 1.3291534596454142,
      "learning_rate": 6.504774866915291e-06,
      "loss": 0.134,
      "step": 269
    },
    {
      "epoch": 0.8083832335329342,
      "grad_norm": 1.0006056006169444,
      "learning_rate": 6.482333583442002e-06,
      "loss": 0.095,
      "step": 270
    },
    {
      "epoch": 0.811377245508982,
      "grad_norm": 1.0774270468937779,
      "learning_rate": 6.459859513701967e-06,
      "loss": 0.1094,
      "step": 271
    },
    {
      "epoch": 0.8143712574850299,
      "grad_norm": 1.3951116103918044,
      "learning_rate": 6.437353154776848e-06,
      "loss": 0.1194,
      "step": 272
    },
    {
      "epoch": 0.8173652694610778,
      "grad_norm": 1.021341198815532,
      "learning_rate": 6.414815004462483e-06,
      "loss": 0.0901,
      "step": 273
    },
    {
      "epoch": 0.8203592814371258,
      "grad_norm": 1.1010030821343109,
      "learning_rate": 6.3922455612578715e-06,
      "loss": 0.1228,
      "step": 274
    },
    {
      "epoch": 0.8233532934131736,
      "grad_norm": 0.9511866880333804,
      "learning_rate": 6.369645324354149e-06,
      "loss": 0.0899,
      "step": 275
    },
    {
      "epoch": 0.8263473053892215,
      "grad_norm": 1.250465741565522,
      "learning_rate": 6.3470147936235485e-06,
      "loss": 0.1422,
      "step": 276
    },
    {
      "epoch": 0.8293413173652695,
      "grad_norm": 0.8178696842079152,
      "learning_rate": 6.3243544696083355e-06,
      "loss": 0.0892,
      "step": 277
    },
    {
      "epoch": 0.8323353293413174,
      "grad_norm": 1.099159320502425,
      "learning_rate": 6.301664853509755e-06,
      "loss": 0.1235,
      "step": 278
    },
    {
      "epoch": 0.8353293413173652,
      "grad_norm": 0.9704378354393064,
      "learning_rate": 6.278946447176924e-06,
      "loss": 0.09,
      "step": 279
    },
    {
      "epoch": 0.8383233532934131,
      "grad_norm": 1.060279189122283,
      "learning_rate": 6.256199753095745e-06,
      "loss": 0.111,
      "step": 280
    },
    {
      "epoch": 0.8413173652694611,
      "grad_norm": 0.9431430257691026,
      "learning_rate": 6.233425274377793e-06,
      "loss": 0.0888,
      "step": 281
    },
    {
      "epoch": 0.844311377245509,
      "grad_norm": 0.9972149946167769,
      "learning_rate": 6.21062351474918e-06,
      "loss": 0.0992,
      "step": 282
    },
    {
      "epoch": 0.8473053892215568,
      "grad_norm": 0.9999944513851664,
      "learning_rate": 6.18779497853942e-06,
      "loss": 0.1069,
      "step": 283
    },
    {
      "epoch": 0.8502994011976048,
      "grad_norm": 1.0777833873698162,
      "learning_rate": 6.164940170670266e-06,
      "loss": 0.1277,
      "step": 284
    },
    {
      "epoch": 0.8532934131736527,
      "grad_norm": 0.944059780515698,
      "learning_rate": 6.142059596644557e-06,
      "loss": 0.093,
      "step": 285
    },
    {
      "epoch": 0.8562874251497006,
      "grad_norm": 0.9406475847456754,
      "learning_rate": 6.11915376253502e-06,
      "loss": 0.0878,
      "step": 286
    },
    {
      "epoch": 0.8592814371257484,
      "grad_norm": 1.0803299716258454,
      "learning_rate": 6.096223174973091e-06,
      "loss": 0.1084,
      "step": 287
    },
    {
      "epoch": 0.8622754491017964,
      "grad_norm": 0.9625467482296507,
      "learning_rate": 6.073268341137694e-06,
      "loss": 0.0948,
      "step": 288
    },
    {
      "epoch": 0.8652694610778443,
      "grad_norm": 1.0169387123195366,
      "learning_rate": 6.050289768744042e-06,
      "loss": 0.1019,
      "step": 289
    },
    {
      "epoch": 0.8682634730538922,
      "grad_norm": 1.1145749757386088,
      "learning_rate": 6.0272879660323936e-06,
      "loss": 0.1162,
      "step": 290
    },
    {
      "epoch": 0.8712574850299402,
      "grad_norm": 1.0824577316733177,
      "learning_rate": 6.004263441756815e-06,
      "loss": 0.1116,
      "step": 291
    },
    {
      "epoch": 0.874251497005988,
      "grad_norm": 1.126245497081582,
      "learning_rate": 5.98121670517393e-06,
      "loss": 0.1095,
      "step": 292
    },
    {
      "epoch": 0.8772455089820359,
      "grad_norm": 0.9303100051388207,
      "learning_rate": 5.958148266031654e-06,
      "loss": 0.1055,
      "step": 293
    },
    {
      "epoch": 0.8802395209580839,
      "grad_norm": 0.9539346533573443,
      "learning_rate": 5.935058634557917e-06,
      "loss": 0.0921,
      "step": 294
    },
    {
      "epoch": 0.8832335329341318,
      "grad_norm": 0.9424827977582004,
      "learning_rate": 5.911948321449384e-06,
      "loss": 0.0977,
      "step": 295
    },
    {
      "epoch": 0.8862275449101796,
      "grad_norm": 1.1061979437887834,
      "learning_rate": 5.8888178378601565e-06,
      "loss": 0.1162,
      "step": 296
    },
    {
      "epoch": 0.8892215568862275,
      "grad_norm": 1.0755508915639524,
      "learning_rate": 5.865667695390468e-06,
      "loss": 0.1077,
      "step": 297
    },
    {
      "epoch": 0.8922155688622755,
      "grad_norm": 0.8957267256285629,
      "learning_rate": 5.842498406075363e-06,
      "loss": 0.0854,
      "step": 298
    },
    {
      "epoch": 0.8952095808383234,
      "grad_norm": 0.9020454729309785,
      "learning_rate": 5.819310482373381e-06,
      "loss": 0.0792,
      "step": 299
    },
    {
      "epoch": 0.8982035928143712,
      "grad_norm": 1.1316190695023067,
      "learning_rate": 5.796104437155213e-06,
      "loss": 0.1172,
      "step": 300
    },
    {
      "epoch": 0.9011976047904192,
      "grad_norm": 0.9809059160582403,
      "learning_rate": 5.772880783692363e-06,
      "loss": 0.0907,
      "step": 301
    },
    {
      "epoch": 0.9041916167664671,
      "grad_norm": 1.1266734978429434,
      "learning_rate": 5.749640035645798e-06,
      "loss": 0.1103,
      "step": 302
    },
    {
      "epoch": 0.907185628742515,
      "grad_norm": 0.9700961485467914,
      "learning_rate": 5.726382707054578e-06,
      "loss": 0.0914,
      "step": 303
    },
    {
      "epoch": 0.9101796407185628,
      "grad_norm": 1.1022339605321532,
      "learning_rate": 5.703109312324493e-06,
      "loss": 0.1063,
      "step": 304
    },
    {
      "epoch": 0.9131736526946108,
      "grad_norm": 1.1131505992017734,
      "learning_rate": 5.679820366216684e-06,
      "loss": 0.1255,
      "step": 305
    },
    {
      "epoch": 0.9161676646706587,
      "grad_norm": 0.9103007173820205,
      "learning_rate": 5.656516383836263e-06,
      "loss": 0.0835,
      "step": 306
    },
    {
      "epoch": 0.9191616766467066,
      "grad_norm": 1.0253402587442022,
      "learning_rate": 5.6331978806209044e-06,
      "loss": 0.1167,
      "step": 307
    },
    {
      "epoch": 0.9221556886227545,
      "grad_norm": 0.9983205063399545,
      "learning_rate": 5.609865372329461e-06,
      "loss": 0.0979,
      "step": 308
    },
    {
      "epoch": 0.9251497005988024,
      "grad_norm": 1.062384203615879,
      "learning_rate": 5.586519375030549e-06,
      "loss": 0.11,
      "step": 309
    },
    {
      "epoch": 0.9281437125748503,
      "grad_norm": 1.0238885150771553,
      "learning_rate": 5.5631604050911354e-06,
      "loss": 0.1015,
      "step": 310
    },
    {
      "epoch": 0.9311377245508982,
      "grad_norm": 1.081717697880171,
      "learning_rate": 5.539788979165115e-06,
      "loss": 0.1085,
      "step": 311
    },
    {
      "epoch": 0.9341317365269461,
      "grad_norm": 0.9454557767735384,
      "learning_rate": 5.516405614181883e-06,
      "loss": 0.0972,
      "step": 312
    },
    {
      "epoch": 0.937125748502994,
      "grad_norm": 0.8642255056444359,
      "learning_rate": 5.4930108273349034e-06,
      "loss": 0.0838,
      "step": 313
    },
    {
      "epoch": 0.9401197604790419,
      "grad_norm": 1.0742451238313808,
      "learning_rate": 5.4696051360702725e-06,
      "loss": 0.1004,
      "step": 314
    },
    {
      "epoch": 0.9431137724550899,
      "grad_norm": 1.2427104390543382,
      "learning_rate": 5.446189058075265e-06,
      "loss": 0.125,
      "step": 315
    },
    {
      "epoch": 0.9461077844311377,
      "grad_norm": 1.1547232598958863,
      "learning_rate": 5.4227631112668955e-06,
      "loss": 0.1172,
      "step": 316
    },
    {
      "epoch": 0.9491017964071856,
      "grad_norm": 1.0713588681149866,
      "learning_rate": 5.39932781378045e-06,
      "loss": 0.0913,
      "step": 317
    },
    {
      "epoch": 0.9520958083832335,
      "grad_norm": 1.2655230572225311,
      "learning_rate": 5.375883683958041e-06,
      "loss": 0.1146,
      "step": 318
    },
    {
      "epoch": 0.9550898203592815,
      "grad_norm": 1.02738422625948,
      "learning_rate": 5.3524312403371255e-06,
      "loss": 0.0983,
      "step": 319
    },
    {
      "epoch": 0.9580838323353293,
      "grad_norm": 1.1158253492067602,
      "learning_rate": 5.328971001639054e-06,
      "loss": 0.1025,
      "step": 320
    },
    {
      "epoch": 0.9610778443113772,
      "grad_norm": 1.1563466876550246,
      "learning_rate": 5.3055034867575825e-06,
      "loss": 0.1209,
      "step": 321
    },
    {
      "epoch": 0.9640718562874252,
      "grad_norm": 1.1624400725956934,
      "learning_rate": 5.282029214747404e-06,
      "loss": 0.1149,
      "step": 322
    },
    {
      "epoch": 0.9670658682634731,
      "grad_norm": 1.094983393860188,
      "learning_rate": 5.258548704812667e-06,
      "loss": 0.1068,
      "step": 323
    },
    {
      "epoch": 0.9700598802395209,
      "grad_norm": 1.2424605367207802,
      "learning_rate": 5.235062476295488e-06,
      "loss": 0.1131,
      "step": 324
    },
    {
      "epoch": 0.9730538922155688,
      "grad_norm": 1.165669123959319,
      "learning_rate": 5.211571048664469e-06,
      "loss": 0.105,
      "step": 325
    },
    {
      "epoch": 0.9760479041916168,
      "grad_norm": 0.9826900693069529,
      "learning_rate": 5.188074941503203e-06,
      "loss": 0.1004,
      "step": 326
    },
    {
      "epoch": 0.9790419161676647,
      "grad_norm": 1.1607668382762542,
      "learning_rate": 5.164574674498788e-06,
      "loss": 0.1106,
      "step": 327
    },
    {
      "epoch": 0.9820359281437125,
      "grad_norm": 1.041407049081876,
      "learning_rate": 5.141070767430331e-06,
      "loss": 0.0958,
      "step": 328
    },
    {
      "epoch": 0.9850299401197605,
      "grad_norm": 1.0608075207765482,
      "learning_rate": 5.117563740157444e-06,
      "loss": 0.109,
      "step": 329
    },
    {
      "epoch": 0.9880239520958084,
      "grad_norm": 1.1798838372371454,
      "learning_rate": 5.094054112608758e-06,
      "loss": 0.1061,
      "step": 330
    },
    {
      "epoch": 0.9910179640718563,
      "grad_norm": 0.9776367931266164,
      "learning_rate": 5.070542404770413e-06,
      "loss": 0.1006,
      "step": 331
    },
    {
      "epoch": 0.9940119760479041,
      "grad_norm": 0.949049338086999,
      "learning_rate": 5.047029136674563e-06,
      "loss": 0.0932,
      "step": 332
    },
    {
      "epoch": 0.9970059880239521,
      "grad_norm": 1.3590598281232353,
      "learning_rate": 5.023514828387868e-06,
      "loss": 0.1381,
      "step": 333
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8195006736867005,
      "learning_rate": 5e-06,
      "loss": 0.0603,
      "step": 334
    },
    {
      "epoch": 1.0029940119760479,
      "grad_norm": 0.7022250508096024,
      "learning_rate": 4.976485171612134e-06,
      "loss": 0.0444,
      "step": 335
    },
    {
      "epoch": 1.0059880239520957,
      "grad_norm": 0.8400384422589814,
      "learning_rate": 4.95297086332544e-06,
      "loss": 0.0539,
      "step": 336
    },
    {
      "epoch": 1.0089820359281436,
      "grad_norm": 0.6845296248630369,
      "learning_rate": 4.9294575952295896e-06,
      "loss": 0.0444,
      "step": 337
    },
    {
      "epoch": 1.0119760479041917,
      "grad_norm": 0.6905446250214292,
      "learning_rate": 4.905945887391242e-06,
      "loss": 0.0466,
      "step": 338
    },
    {
      "epoch": 1.0149700598802396,
      "grad_norm": 0.7099200734930969,
      "learning_rate": 4.882436259842556e-06,
      "loss": 0.0371,
      "step": 339
    },
    {
      "epoch": 1.0179640718562875,
      "grad_norm": 0.6846237770550866,
      "learning_rate": 4.858929232569671e-06,
      "loss": 0.0423,
      "step": 340
    },
    {
      "epoch": 1.0209580838323353,
      "grad_norm": 0.8422212258193694,
      "learning_rate": 4.835425325501214e-06,
      "loss": 0.0584,
      "step": 341
    },
    {
      "epoch": 1.0239520958083832,
      "grad_norm": 0.6718418934985309,
      "learning_rate": 4.811925058496799e-06,
      "loss": 0.0425,
      "step": 342
    },
    {
      "epoch": 1.026946107784431,
      "grad_norm": 0.7866219711881971,
      "learning_rate": 4.788428951335534e-06,
      "loss": 0.0429,
      "step": 343
    },
    {
      "epoch": 1.029940119760479,
      "grad_norm": 0.6744798090833058,
      "learning_rate": 4.7649375237045135e-06,
      "loss": 0.0427,
      "step": 344
    },
    {
      "epoch": 1.032934131736527,
      "grad_norm": 0.6129564687517989,
      "learning_rate": 4.741451295187333e-06,
      "loss": 0.0307,
      "step": 345
    },
    {
      "epoch": 1.035928143712575,
      "grad_norm": 0.7597310180785687,
      "learning_rate": 4.717970785252596e-06,
      "loss": 0.0428,
      "step": 346
    },
    {
      "epoch": 1.0389221556886228,
      "grad_norm": 0.8159453183034721,
      "learning_rate": 4.694496513242418e-06,
      "loss": 0.0453,
      "step": 347
    },
    {
      "epoch": 1.0419161676646707,
      "grad_norm": 0.7899612435212686,
      "learning_rate": 4.671028998360947e-06,
      "loss": 0.0389,
      "step": 348
    },
    {
      "epoch": 1.0449101796407185,
      "grad_norm": 0.7844445993080502,
      "learning_rate": 4.647568759662876e-06,
      "loss": 0.0341,
      "step": 349
    },
    {
      "epoch": 1.0479041916167664,
      "grad_norm": 0.9902015125369322,
      "learning_rate": 4.624116316041962e-06,
      "loss": 0.0459,
      "step": 350
    },
    {
      "epoch": 1.0508982035928143,
      "grad_norm": 0.7288614709902538,
      "learning_rate": 4.600672186219551e-06,
      "loss": 0.0325,
      "step": 351
    },
    {
      "epoch": 1.0538922155688624,
      "grad_norm": 1.0379948328322208,
      "learning_rate": 4.5772368887331044e-06,
      "loss": 0.0392,
      "step": 352
    },
    {
      "epoch": 1.0568862275449102,
      "grad_norm": 0.9525464167880752,
      "learning_rate": 4.553810941924735e-06,
      "loss": 0.0472,
      "step": 353
    },
    {
      "epoch": 1.0598802395209581,
      "grad_norm": 0.9908063099250911,
      "learning_rate": 4.530394863929728e-06,
      "loss": 0.0466,
      "step": 354
    },
    {
      "epoch": 1.062874251497006,
      "grad_norm": 0.890148196124663,
      "learning_rate": 4.506989172665097e-06,
      "loss": 0.0409,
      "step": 355
    },
    {
      "epoch": 1.0658682634730539,
      "grad_norm": 0.8588250415633526,
      "learning_rate": 4.483594385818119e-06,
      "loss": 0.0255,
      "step": 356
    },
    {
      "epoch": 1.0688622754491017,
      "grad_norm": 1.4436171796111879,
      "learning_rate": 4.460211020834887e-06,
      "loss": 0.0509,
      "step": 357
    },
    {
      "epoch": 1.0718562874251496,
      "grad_norm": 1.0453025017183444,
      "learning_rate": 4.436839594908866e-06,
      "loss": 0.0396,
      "step": 358
    },
    {
      "epoch": 1.0748502994011977,
      "grad_norm": 0.9000241982502061,
      "learning_rate": 4.4134806249694514e-06,
      "loss": 0.033,
      "step": 359
    },
    {
      "epoch": 1.0778443113772456,
      "grad_norm": 1.4778591988780183,
      "learning_rate": 4.39013462767054e-06,
      "loss": 0.0571,
      "step": 360
    },
    {
      "epoch": 1.0808383233532934,
      "grad_norm": 0.9120237966074196,
      "learning_rate": 4.366802119379098e-06,
      "loss": 0.0359,
      "step": 361
    },
    {
      "epoch": 1.0838323353293413,
      "grad_norm": 0.861035210399316,
      "learning_rate": 4.34348361616374e-06,
      "loss": 0.0343,
      "step": 362
    },
    {
      "epoch": 1.0868263473053892,
      "grad_norm": 1.2449114641135295,
      "learning_rate": 4.3201796337833165e-06,
      "loss": 0.0421,
      "step": 363
    },
    {
      "epoch": 1.089820359281437,
      "grad_norm": 1.0254909246402393,
      "learning_rate": 4.29689068767551e-06,
      "loss": 0.0383,
      "step": 364
    },
    {
      "epoch": 1.092814371257485,
      "grad_norm": 1.0322528091999164,
      "learning_rate": 4.273617292945425e-06,
      "loss": 0.0418,
      "step": 365
    },
    {
      "epoch": 1.095808383233533,
      "grad_norm": 0.8771453681004516,
      "learning_rate": 4.250359964354203e-06,
      "loss": 0.0357,
      "step": 366
    },
    {
      "epoch": 1.098802395209581,
      "grad_norm": 0.9241347977702095,
      "learning_rate": 4.227119216307637e-06,
      "loss": 0.0312,
      "step": 367
    },
    {
      "epoch": 1.1017964071856288,
      "grad_norm": 0.9312614404924237,
      "learning_rate": 4.203895562844789e-06,
      "loss": 0.0321,
      "step": 368
    },
    {
      "epoch": 1.1047904191616766,
      "grad_norm": 0.9299612677491177,
      "learning_rate": 4.18068951762662e-06,
      "loss": 0.0314,
      "step": 369
    },
    {
      "epoch": 1.1077844311377245,
      "grad_norm": 1.0755481573764654,
      "learning_rate": 4.157501593924638e-06,
      "loss": 0.0397,
      "step": 370
    },
    {
      "epoch": 1.1107784431137724,
      "grad_norm": 1.2817493788111067,
      "learning_rate": 4.134332304609533e-06,
      "loss": 0.0512,
      "step": 371
    },
    {
      "epoch": 1.1137724550898203,
      "grad_norm": 0.7831865066541385,
      "learning_rate": 4.111182162139844e-06,
      "loss": 0.0287,
      "step": 372
    },
    {
      "epoch": 1.1167664670658684,
      "grad_norm": 0.9104483556890546,
      "learning_rate": 4.088051678550617e-06,
      "loss": 0.029,
      "step": 373
    },
    {
      "epoch": 1.1197604790419162,
      "grad_norm": 1.0727534306651358,
      "learning_rate": 4.064941365442084e-06,
      "loss": 0.0401,
      "step": 374
    },
    {
      "epoch": 1.122754491017964,
      "grad_norm": 1.2791014951497919,
      "learning_rate": 4.041851733968348e-06,
      "loss": 0.0523,
      "step": 375
    },
    {
      "epoch": 1.125748502994012,
      "grad_norm": 1.156575196796342,
      "learning_rate": 4.018783294826071e-06,
      "loss": 0.0489,
      "step": 376
    },
    {
      "epoch": 1.1287425149700598,
      "grad_norm": 1.1315916978871787,
      "learning_rate": 3.995736558243186e-06,
      "loss": 0.0417,
      "step": 377
    },
    {
      "epoch": 1.1317365269461077,
      "grad_norm": 0.8825597316061771,
      "learning_rate": 3.972712033967608e-06,
      "loss": 0.039,
      "step": 378
    },
    {
      "epoch": 1.1347305389221556,
      "grad_norm": 0.9683200779006866,
      "learning_rate": 3.949710231255961e-06,
      "loss": 0.0408,
      "step": 379
    },
    {
      "epoch": 1.1377245508982037,
      "grad_norm": 0.9712527531198679,
      "learning_rate": 3.926731658862307e-06,
      "loss": 0.0417,
      "step": 380
    },
    {
      "epoch": 1.1407185628742516,
      "grad_norm": 0.8350218043206888,
      "learning_rate": 3.903776825026912e-06,
      "loss": 0.0354,
      "step": 381
    },
    {
      "epoch": 1.1437125748502994,
      "grad_norm": 0.9210222426166789,
      "learning_rate": 3.8808462374649805e-06,
      "loss": 0.0469,
      "step": 382
    },
    {
      "epoch": 1.1467065868263473,
      "grad_norm": 0.83057947119231,
      "learning_rate": 3.857940403355444e-06,
      "loss": 0.0332,
      "step": 383
    },
    {
      "epoch": 1.1497005988023952,
      "grad_norm": 0.9369072838558526,
      "learning_rate": 3.8350598293297345e-06,
      "loss": 0.0477,
      "step": 384
    },
    {
      "epoch": 1.152694610778443,
      "grad_norm": 1.0204151040644838,
      "learning_rate": 3.8122050214605822e-06,
      "loss": 0.0421,
      "step": 385
    },
    {
      "epoch": 1.1556886227544911,
      "grad_norm": 0.9429121227055448,
      "learning_rate": 3.7893764852508207e-06,
      "loss": 0.0468,
      "step": 386
    },
    {
      "epoch": 1.158682634730539,
      "grad_norm": 1.1341086066262012,
      "learning_rate": 3.766574725622208e-06,
      "loss": 0.0354,
      "step": 387
    },
    {
      "epoch": 1.1616766467065869,
      "grad_norm": 0.896449256938119,
      "learning_rate": 3.7438002469042567e-06,
      "loss": 0.0359,
      "step": 388
    },
    {
      "epoch": 1.1646706586826348,
      "grad_norm": 0.9947399393955884,
      "learning_rate": 3.721053552823078e-06,
      "loss": 0.0399,
      "step": 389
    },
    {
      "epoch": 1.1676646706586826,
      "grad_norm": 0.8833187563777827,
      "learning_rate": 3.698335146490246e-06,
      "loss": 0.0402,
      "step": 390
    },
    {
      "epoch": 1.1706586826347305,
      "grad_norm": 0.9911011239135578,
      "learning_rate": 3.675645530391665e-06,
      "loss": 0.0365,
      "step": 391
    },
    {
      "epoch": 1.1736526946107784,
      "grad_norm": 1.0840176172640383,
      "learning_rate": 3.652985206376455e-06,
      "loss": 0.0406,
      "step": 392
    },
    {
      "epoch": 1.1766467065868262,
      "grad_norm": 0.8760812579100875,
      "learning_rate": 3.630354675645853e-06,
      "loss": 0.0417,
      "step": 393
    },
    {
      "epoch": 1.1796407185628743,
      "grad_norm": 1.1143345805853575,
      "learning_rate": 3.6077544387421293e-06,
      "loss": 0.0433,
      "step": 394
    },
    {
      "epoch": 1.1826347305389222,
      "grad_norm": 0.8075479023380153,
      "learning_rate": 3.5851849955375177e-06,
      "loss": 0.0338,
      "step": 395
    },
    {
      "epoch": 1.18562874251497,
      "grad_norm": 0.7628895299170928,
      "learning_rate": 3.5626468452231534e-06,
      "loss": 0.0296,
      "step": 396
    },
    {
      "epoch": 1.188622754491018,
      "grad_norm": 0.796450630478703,
      "learning_rate": 3.540140486298035e-06,
      "loss": 0.0322,
      "step": 397
    },
    {
      "epoch": 1.1916167664670658,
      "grad_norm": 0.9289430207715151,
      "learning_rate": 3.517666416557999e-06,
      "loss": 0.0442,
      "step": 398
    },
    {
      "epoch": 1.1946107784431137,
      "grad_norm": 0.821485237499904,
      "learning_rate": 3.495225133084712e-06,
      "loss": 0.0305,
      "step": 399
    },
    {
      "epoch": 1.1976047904191618,
      "grad_norm": 0.7668009339297708,
      "learning_rate": 3.472817132234669e-06,
      "loss": 0.0321,
      "step": 400
    },
    {
      "epoch": 1.1976047904191618,
      "eval_loss": 0.13780812919139862,
      "eval_runtime": 3.3372,
      "eval_samples_per_second": 16.181,
      "eval_steps_per_second": 4.195,
      "step": 400
    },
    {
      "epoch": 1.2005988023952097,
      "grad_norm": 0.9271611220363486,
      "learning_rate": 3.4504429096282246e-06,
      "loss": 0.0349,
      "step": 401
    },
    {
      "epoch": 1.2035928143712575,
      "grad_norm": 0.98300005474815,
      "learning_rate": 3.428102960138625e-06,
      "loss": 0.0426,
      "step": 402
    },
    {
      "epoch": 1.2065868263473054,
      "grad_norm": 1.1938347371909825,
      "learning_rate": 3.405797777881059e-06,
      "loss": 0.0463,
      "step": 403
    },
    {
      "epoch": 1.2095808383233533,
      "grad_norm": 0.9291923066749767,
      "learning_rate": 3.3835278562017405e-06,
      "loss": 0.0344,
      "step": 404
    },
    {
      "epoch": 1.2125748502994012,
      "grad_norm": 0.7700342942383323,
      "learning_rate": 3.3612936876669834e-06,
      "loss": 0.0317,
      "step": 405
    },
    {
      "epoch": 1.215568862275449,
      "grad_norm": 0.8834942586025577,
      "learning_rate": 3.3390957640523147e-06,
      "loss": 0.0398,
      "step": 406
    },
    {
      "epoch": 1.218562874251497,
      "grad_norm": 0.8667383335758857,
      "learning_rate": 3.3169345763315986e-06,
      "loss": 0.0331,
      "step": 407
    },
    {
      "epoch": 1.221556886227545,
      "grad_norm": 0.9737301322940288,
      "learning_rate": 3.29481061466617e-06,
      "loss": 0.042,
      "step": 408
    },
    {
      "epoch": 1.2245508982035929,
      "grad_norm": 1.0693430967635835,
      "learning_rate": 3.2727243683940045e-06,
      "loss": 0.044,
      "step": 409
    },
    {
      "epoch": 1.2275449101796407,
      "grad_norm": 0.9646423394804879,
      "learning_rate": 3.2506763260188824e-06,
      "loss": 0.0354,
      "step": 410
    },
    {
      "epoch": 1.2305389221556886,
      "grad_norm": 0.9645068212307435,
      "learning_rate": 3.2286669751995905e-06,
      "loss": 0.0337,
      "step": 411
    },
    {
      "epoch": 1.2335329341317365,
      "grad_norm": 1.0521076649648817,
      "learning_rate": 3.2066968027391377e-06,
      "loss": 0.0362,
      "step": 412
    },
    {
      "epoch": 1.2365269461077844,
      "grad_norm": 0.8720712848707571,
      "learning_rate": 3.1847662945739833e-06,
      "loss": 0.0357,
      "step": 413
    },
    {
      "epoch": 1.2395209580838324,
      "grad_norm": 0.8779999441612727,
      "learning_rate": 3.1628759357632943e-06,
      "loss": 0.0337,
      "step": 414
    },
    {
      "epoch": 1.2425149700598803,
      "grad_norm": 0.867901355187604,
      "learning_rate": 3.1410262104782086e-06,
      "loss": 0.0361,
      "step": 415
    },
    {
      "epoch": 1.2455089820359282,
      "grad_norm": 0.9871812750329428,
      "learning_rate": 3.119217601991139e-06,
      "loss": 0.0352,
      "step": 416
    },
    {
      "epoch": 1.248502994011976,
      "grad_norm": 0.9895019085971923,
      "learning_rate": 3.0974505926650724e-06,
      "loss": 0.0408,
      "step": 417
    },
    {
      "epoch": 1.251497005988024,
      "grad_norm": 0.9870216405705468,
      "learning_rate": 3.0757256639429027e-06,
      "loss": 0.0366,
      "step": 418
    },
    {
      "epoch": 1.2544910179640718,
      "grad_norm": 0.8780513766002908,
      "learning_rate": 3.0540432963367907e-06,
      "loss": 0.0366,
      "step": 419
    },
    {
      "epoch": 1.2574850299401197,
      "grad_norm": 0.8956725818447345,
      "learning_rate": 3.032403969417523e-06,
      "loss": 0.0336,
      "step": 420
    },
    {
      "epoch": 1.2604790419161676,
      "grad_norm": 0.8378149694900093,
      "learning_rate": 3.010808161803917e-06,
      "loss": 0.0361,
      "step": 421
    },
    {
      "epoch": 1.2634730538922156,
      "grad_norm": 1.119672198263993,
      "learning_rate": 2.9892563511522305e-06,
      "loss": 0.0488,
      "step": 422
    },
    {
      "epoch": 1.2664670658682635,
      "grad_norm": 0.9267554877923985,
      "learning_rate": 2.9677490141455915e-06,
      "loss": 0.0362,
      "step": 423
    },
    {
      "epoch": 1.2694610778443114,
      "grad_norm": 0.9254019394598606,
      "learning_rate": 2.946286626483463e-06,
      "loss": 0.0426,
      "step": 424
    },
    {
      "epoch": 1.2724550898203593,
      "grad_norm": 0.9314120465339711,
      "learning_rate": 2.924869662871117e-06,
      "loss": 0.0409,
      "step": 425
    },
    {
      "epoch": 1.2754491017964071,
      "grad_norm": 0.9358364788002177,
      "learning_rate": 2.903498597009136e-06,
      "loss": 0.0472,
      "step": 426
    },
    {
      "epoch": 1.278443113772455,
      "grad_norm": 1.0158323127919096,
      "learning_rate": 2.8821739015829338e-06,
      "loss": 0.0407,
      "step": 427
    },
    {
      "epoch": 1.281437125748503,
      "grad_norm": 0.6684116644857107,
      "learning_rate": 2.8608960482523058e-06,
      "loss": 0.0266,
      "step": 428
    },
    {
      "epoch": 1.284431137724551,
      "grad_norm": 0.8629783713968431,
      "learning_rate": 2.839665507640992e-06,
      "loss": 0.0334,
      "step": 429
    },
    {
      "epoch": 1.2874251497005988,
      "grad_norm": 0.9542038899534631,
      "learning_rate": 2.818482749326272e-06,
      "loss": 0.0371,
      "step": 430
    },
    {
      "epoch": 1.2904191616766467,
      "grad_norm": 0.8672086239938103,
      "learning_rate": 2.797348241828569e-06,
      "loss": 0.0343,
      "step": 431
    },
    {
      "epoch": 1.2934131736526946,
      "grad_norm": 0.7488187996340843,
      "learning_rate": 2.776262452601104e-06,
      "loss": 0.0307,
      "step": 432
    },
    {
      "epoch": 1.2964071856287425,
      "grad_norm": 0.9513722695609567,
      "learning_rate": 2.7552258480195348e-06,
      "loss": 0.037,
      "step": 433
    },
    {
      "epoch": 1.2994011976047903,
      "grad_norm": 0.8253486792915236,
      "learning_rate": 2.734238893371667e-06,
      "loss": 0.0306,
      "step": 434
    },
    {
      "epoch": 1.3023952095808382,
      "grad_norm": 0.9929470155180504,
      "learning_rate": 2.7133020528471322e-06,
      "loss": 0.0386,
      "step": 435
    },
    {
      "epoch": 1.3053892215568863,
      "grad_norm": 0.9147342783007899,
      "learning_rate": 2.6924157895271563e-06,
      "loss": 0.0323,
      "step": 436
    },
    {
      "epoch": 1.3083832335329342,
      "grad_norm": 0.8762640635908312,
      "learning_rate": 2.671580565374282e-06,
      "loss": 0.0418,
      "step": 437
    },
    {
      "epoch": 1.311377245508982,
      "grad_norm": 1.03228121413644,
      "learning_rate": 2.6507968412221763e-06,
      "loss": 0.0456,
      "step": 438
    },
    {
      "epoch": 1.31437125748503,
      "grad_norm": 0.9236414417724473,
      "learning_rate": 2.6300650767654234e-06,
      "loss": 0.0333,
      "step": 439
    },
    {
      "epoch": 1.3173652694610778,
      "grad_norm": 0.9657694751827307,
      "learning_rate": 2.6093857305493666e-06,
      "loss": 0.0419,
      "step": 440
    },
    {
      "epoch": 1.3203592814371259,
      "grad_norm": 0.9019012968388108,
      "learning_rate": 2.588759259959962e-06,
      "loss": 0.0398,
      "step": 441
    },
    {
      "epoch": 1.3233532934131738,
      "grad_norm": 0.9298261278356583,
      "learning_rate": 2.568186121213658e-06,
      "loss": 0.0363,
      "step": 442
    },
    {
      "epoch": 1.3263473053892216,
      "grad_norm": 1.2240905923431065,
      "learning_rate": 2.547666769347312e-06,
      "loss": 0.0488,
      "step": 443
    },
    {
      "epoch": 1.3293413173652695,
      "grad_norm": 1.0889035979064696,
      "learning_rate": 2.5272016582081236e-06,
      "loss": 0.0426,
      "step": 444
    },
    {
      "epoch": 1.3323353293413174,
      "grad_norm": 0.925889881165544,
      "learning_rate": 2.5067912404435952e-06,
      "loss": 0.0359,
      "step": 445
    },
    {
      "epoch": 1.3353293413173652,
      "grad_norm": 0.9150893063037066,
      "learning_rate": 2.486435967491516e-06,
      "loss": 0.0403,
      "step": 446
    },
    {
      "epoch": 1.3383233532934131,
      "grad_norm": 0.9794856179588529,
      "learning_rate": 2.4661362895699903e-06,
      "loss": 0.0403,
      "step": 447
    },
    {
      "epoch": 1.341317365269461,
      "grad_norm": 1.0125071969005328,
      "learning_rate": 2.445892655667462e-06,
      "loss": 0.0441,
      "step": 448
    },
    {
      "epoch": 1.3443113772455089,
      "grad_norm": 0.9049114175072965,
      "learning_rate": 2.425705513532798e-06,
      "loss": 0.0362,
      "step": 449
    },
    {
      "epoch": 1.347305389221557,
      "grad_norm": 0.9526170637604479,
      "learning_rate": 2.4055753096653795e-06,
      "loss": 0.0435,
      "step": 450
    },
    {
      "epoch": 1.3502994011976048,
      "grad_norm": 0.9813475965414489,
      "learning_rate": 2.3855024893052286e-06,
      "loss": 0.0369,
      "step": 451
    },
    {
      "epoch": 1.3532934131736527,
      "grad_norm": 0.9018380127385193,
      "learning_rate": 2.365487496423152e-06,
      "loss": 0.0409,
      "step": 452
    },
    {
      "epoch": 1.3562874251497006,
      "grad_norm": 1.0068696888567505,
      "learning_rate": 2.3455307737109338e-06,
      "loss": 0.0372,
      "step": 453
    },
    {
      "epoch": 1.3592814371257484,
      "grad_norm": 0.8992549250693872,
      "learning_rate": 2.3256327625715345e-06,
      "loss": 0.0362,
      "step": 454
    },
    {
      "epoch": 1.3622754491017965,
      "grad_norm": 0.8179812090436723,
      "learning_rate": 2.3057939031093346e-06,
      "loss": 0.0303,
      "step": 455
    },
    {
      "epoch": 1.3652694610778444,
      "grad_norm": 0.7436963483344036,
      "learning_rate": 2.2860146341203936e-06,
      "loss": 0.0305,
      "step": 456
    },
    {
      "epoch": 1.3682634730538923,
      "grad_norm": 0.9178145503515854,
      "learning_rate": 2.2662953930827546e-06,
      "loss": 0.0445,
      "step": 457
    },
    {
      "epoch": 1.3712574850299402,
      "grad_norm": 1.0682235826302338,
      "learning_rate": 2.2466366161467528e-06,
      "loss": 0.0283,
      "step": 458
    },
    {
      "epoch": 1.374251497005988,
      "grad_norm": 0.9909161612586675,
      "learning_rate": 2.227038738125385e-06,
      "loss": 0.0379,
      "step": 459
    },
    {
      "epoch": 1.377245508982036,
      "grad_norm": 1.0489147208906358,
      "learning_rate": 2.207502192484685e-06,
      "loss": 0.0377,
      "step": 460
    },
    {
      "epoch": 1.3802395209580838,
      "grad_norm": 0.8240521658886,
      "learning_rate": 2.188027411334131e-06,
      "loss": 0.0297,
      "step": 461
    },
    {
      "epoch": 1.3832335329341316,
      "grad_norm": 1.012425630814328,
      "learning_rate": 2.1686148254171012e-06,
      "loss": 0.0428,
      "step": 462
    },
    {
      "epoch": 1.3862275449101795,
      "grad_norm": 0.9133537060947438,
      "learning_rate": 2.1492648641013305e-06,
      "loss": 0.0322,
      "step": 463
    },
    {
      "epoch": 1.3892215568862276,
      "grad_norm": 0.7328334518432168,
      "learning_rate": 2.1299779553694323e-06,
      "loss": 0.0278,
      "step": 464
    },
    {
      "epoch": 1.3922155688622755,
      "grad_norm": 0.9760436430695598,
      "learning_rate": 2.1107545258094135e-06,
      "loss": 0.038,
      "step": 465
    },
    {
      "epoch": 1.3952095808383234,
      "grad_norm": 0.8767138969331109,
      "learning_rate": 2.0915950006052555e-06,
      "loss": 0.0298,
      "step": 466
    },
    {
      "epoch": 1.3982035928143712,
      "grad_norm": 0.7963562407443522,
      "learning_rate": 2.0724998035274947e-06,
      "loss": 0.0275,
      "step": 467
    },
    {
      "epoch": 1.401197604790419,
      "grad_norm": 0.815240498472856,
      "learning_rate": 2.053469356923865e-06,
      "loss": 0.0302,
      "step": 468
    },
    {
      "epoch": 1.4041916167664672,
      "grad_norm": 0.8250436472251806,
      "learning_rate": 2.0345040817099433e-06,
      "loss": 0.0265,
      "step": 469
    },
    {
      "epoch": 1.407185628742515,
      "grad_norm": 0.6931371543506848,
      "learning_rate": 2.0156043973598475e-06,
      "loss": 0.0276,
      "step": 470
    },
    {
      "epoch": 1.410179640718563,
      "grad_norm": 1.1066743785923727,
      "learning_rate": 1.996770721896957e-06,
      "loss": 0.037,
      "step": 471
    },
    {
      "epoch": 1.4131736526946108,
      "grad_norm": 0.7845958476199715,
      "learning_rate": 1.9780034718846653e-06,
      "loss": 0.0281,
      "step": 472
    },
    {
      "epoch": 1.4161676646706587,
      "grad_norm": 1.092505161128868,
      "learning_rate": 1.9593030624171683e-06,
      "loss": 0.043,
      "step": 473
    },
    {
      "epoch": 1.4191616766467066,
      "grad_norm": 0.9968098388138645,
      "learning_rate": 1.9406699071102774e-06,
      "loss": 0.0411,
      "step": 474
    },
    {
      "epoch": 1.4221556886227544,
      "grad_norm": 0.7625327735153937,
      "learning_rate": 1.9221044180922833e-06,
      "loss": 0.0286,
      "step": 475
    },
    {
      "epoch": 1.4251497005988023,
      "grad_norm": 0.9561990252612784,
      "learning_rate": 1.9036070059948253e-06,
      "loss": 0.0372,
      "step": 476
    },
    {
      "epoch": 1.4281437125748502,
      "grad_norm": 1.1381730797979581,
      "learning_rate": 1.885178079943823e-06,
      "loss": 0.044,
      "step": 477
    },
    {
      "epoch": 1.4311377245508983,
      "grad_norm": 1.1421220738835591,
      "learning_rate": 1.866818047550419e-06,
      "loss": 0.0481,
      "step": 478
    },
    {
      "epoch": 1.4341317365269461,
      "grad_norm": 0.8091431255907582,
      "learning_rate": 1.8485273149019655e-06,
      "loss": 0.0327,
      "step": 479
    },
    {
      "epoch": 1.437125748502994,
      "grad_norm": 1.0302486203137764,
      "learning_rate": 1.8303062865530407e-06,
      "loss": 0.029,
      "step": 480
    },
    {
      "epoch": 1.4401197604790419,
      "grad_norm": 0.9385305161033584,
      "learning_rate": 1.8121553655165058e-06,
      "loss": 0.0362,
      "step": 481
    },
    {
      "epoch": 1.4431137724550898,
      "grad_norm": 1.2125617669933204,
      "learning_rate": 1.7940749532545832e-06,
      "loss": 0.0367,
      "step": 482
    },
    {
      "epoch": 1.4461077844311379,
      "grad_norm": 1.0336840509717438,
      "learning_rate": 1.7760654496699876e-06,
      "loss": 0.0336,
      "step": 483
    },
    {
      "epoch": 1.4491017964071857,
      "grad_norm": 0.9153272128900931,
      "learning_rate": 1.7581272530970666e-06,
      "loss": 0.04,
      "step": 484
    },
    {
      "epoch": 1.4520958083832336,
      "grad_norm": 0.8901653350755709,
      "learning_rate": 1.7402607602930106e-06,
      "loss": 0.0317,
      "step": 485
    },
    {
      "epoch": 1.4550898203592815,
      "grad_norm": 0.9239872375706311,
      "learning_rate": 1.7224663664290537e-06,
      "loss": 0.0341,
      "step": 486
    },
    {
      "epoch": 1.4580838323353293,
      "grad_norm": 0.856622304612412,
      "learning_rate": 1.7047444650817518e-06,
      "loss": 0.0338,
      "step": 487
    },
    {
      "epoch": 1.4610778443113772,
      "grad_norm": 0.7649462285585268,
      "learning_rate": 1.6870954482242707e-06,
      "loss": 0.0267,
      "step": 488
    },
    {
      "epoch": 1.464071856287425,
      "grad_norm": 0.8399844122361557,
      "learning_rate": 1.669519706217711e-06,
      "loss": 0.0315,
      "step": 489
    },
    {
      "epoch": 1.467065868263473,
      "grad_norm": 1.1346555295276777,
      "learning_rate": 1.652017627802487e-06,
      "loss": 0.0412,
      "step": 490
    },
    {
      "epoch": 1.4700598802395208,
      "grad_norm": 0.7898462844513765,
      "learning_rate": 1.6345896000897122e-06,
      "loss": 0.0283,
      "step": 491
    },
    {
      "epoch": 1.473053892215569,
      "grad_norm": 1.071158052676053,
      "learning_rate": 1.6172360085526567e-06,
      "loss": 0.0439,
      "step": 492
    },
    {
      "epoch": 1.4760479041916168,
      "grad_norm": 1.1086667455086867,
      "learning_rate": 1.5999572370182016e-06,
      "loss": 0.0458,
      "step": 493
    },
    {
      "epoch": 1.4790419161676647,
      "grad_norm": 0.9342746712943463,
      "learning_rate": 1.5827536676583643e-06,
      "loss": 0.0363,
      "step": 494
    },
    {
      "epoch": 1.4820359281437125,
      "grad_norm": 0.894292057582835,
      "learning_rate": 1.5656256809818343e-06,
      "loss": 0.0336,
      "step": 495
    },
    {
      "epoch": 1.4850299401197604,
      "grad_norm": 0.7975914735077538,
      "learning_rate": 1.54857365582557e-06,
      "loss": 0.0262,
      "step": 496
    },
    {
      "epoch": 1.4880239520958085,
      "grad_norm": 0.9107586601688882,
      "learning_rate": 1.5315979693464039e-06,
      "loss": 0.0343,
      "step": 497
    },
    {
      "epoch": 1.4910179640718564,
      "grad_norm": 0.9256926766092709,
      "learning_rate": 1.5146989970127158e-06,
      "loss": 0.0311,
      "step": 498
    },
    {
      "epoch": 1.4940119760479043,
      "grad_norm": 1.1305167840468748,
      "learning_rate": 1.4978771125961177e-06,
      "loss": 0.0395,
      "step": 499
    },
    {
      "epoch": 1.4970059880239521,
      "grad_norm": 0.9965666026077408,
      "learning_rate": 1.4811326881631937e-06,
      "loss": 0.0436,
      "step": 500
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.892816539298654,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.0358,
      "step": 501
    },
    {
      "epoch": 1.5029940119760479,
      "grad_norm": 1.22206150108301,
      "learning_rate": 1.4478776989401949e-06,
      "loss": 0.0423,
      "step": 502
    },
    {
      "epoch": 1.5059880239520957,
      "grad_norm": 1.031632565991146,
      "learning_rate": 1.4313678696842559e-06,
      "loss": 0.0379,
      "step": 503
    },
    {
      "epoch": 1.5089820359281436,
      "grad_norm": 0.912492546356356,
      "learning_rate": 1.4149369714639856e-06,
      "loss": 0.0359,
      "step": 504
    },
    {
      "epoch": 1.5119760479041915,
      "grad_norm": 1.056683411611693,
      "learning_rate": 1.3985853676981316e-06,
      "loss": 0.0323,
      "step": 505
    },
    {
      "epoch": 1.5149700598802394,
      "grad_norm": 0.9459183995084929,
      "learning_rate": 1.3823134200516043e-06,
      "loss": 0.0289,
      "step": 506
    },
    {
      "epoch": 1.5179640718562875,
      "grad_norm": 0.9791717664239262,
      "learning_rate": 1.366121488427481e-06,
      "loss": 0.0369,
      "step": 507
    },
    {
      "epoch": 1.5209580838323353,
      "grad_norm": 0.8448327976535539,
      "learning_rate": 1.3500099309590397e-06,
      "loss": 0.0281,
      "step": 508
    },
    {
      "epoch": 1.5239520958083832,
      "grad_norm": 0.9589443249740586,
      "learning_rate": 1.3339791040018479e-06,
      "loss": 0.0306,
      "step": 509
    },
    {
      "epoch": 1.5269461077844313,
      "grad_norm": 0.9974377296637968,
      "learning_rate": 1.3180293621258694e-06,
      "loss": 0.0359,
      "step": 510
    },
    {
      "epoch": 1.5299401197604792,
      "grad_norm": 0.8130843099875799,
      "learning_rate": 1.3021610581076316e-06,
      "loss": 0.0296,
      "step": 511
    },
    {
      "epoch": 1.532934131736527,
      "grad_norm": 1.055372433696626,
      "learning_rate": 1.2863745429224145e-06,
      "loss": 0.0288,
      "step": 512
    },
    {
      "epoch": 1.535928143712575,
      "grad_norm": 0.8574697830122907,
      "learning_rate": 1.270670165736499e-06,
      "loss": 0.03,
      "step": 513
    },
    {
      "epoch": 1.5389221556886228,
      "grad_norm": 1.0219464655901311,
      "learning_rate": 1.2550482738994284e-06,
      "loss": 0.0393,
      "step": 514
    },
    {
      "epoch": 1.5419161676646707,
      "grad_norm": 0.9435907783382437,
      "learning_rate": 1.239509212936343e-06,
      "loss": 0.0419,
      "step": 515
    },
    {
      "epoch": 1.5449101796407185,
      "grad_norm": 0.8952249916322559,
      "learning_rate": 1.22405332654032e-06,
      "loss": 0.0347,
      "step": 516
    },
    {
      "epoch": 1.5479041916167664,
      "grad_norm": 0.8719234069171748,
      "learning_rate": 1.2086809565647877e-06,
      "loss": 0.0364,
      "step": 517
    },
    {
      "epoch": 1.5508982035928143,
      "grad_norm": 0.9067688648717696,
      "learning_rate": 1.1933924430159571e-06,
      "loss": 0.0345,
      "step": 518
    },
    {
      "epoch": 1.5538922155688621,
      "grad_norm": 0.974900328573073,
      "learning_rate": 1.1781881240452958e-06,
      "loss": 0.0344,
      "step": 519
    },
    {
      "epoch": 1.55688622754491,
      "grad_norm": 1.0799192861247542,
      "learning_rate": 1.1630683359420653e-06,
      "loss": 0.0445,
      "step": 520
    },
    {
      "epoch": 1.5598802395209581,
      "grad_norm": 1.1462018547308261,
      "learning_rate": 1.1480334131258626e-06,
      "loss": 0.032,
      "step": 521
    },
    {
      "epoch": 1.562874251497006,
      "grad_norm": 0.984295462219783,
      "learning_rate": 1.1330836881392405e-06,
      "loss": 0.0343,
      "step": 522
    },
    {
      "epoch": 1.5658682634730539,
      "grad_norm": 0.8244354713062851,
      "learning_rate": 1.11821949164034e-06,
      "loss": 0.029,
      "step": 523
    },
    {
      "epoch": 1.568862275449102,
      "grad_norm": 0.7613320589661253,
      "learning_rate": 1.103441152395588e-06,
      "loss": 0.0271,
      "step": 524
    },
    {
      "epoch": 1.5718562874251498,
      "grad_norm": 0.9662360897276236,
      "learning_rate": 1.088748997272414e-06,
      "loss": 0.0371,
      "step": 525
    },
    {
      "epoch": 1.5748502994011977,
      "grad_norm": 1.0343246806449573,
      "learning_rate": 1.0741433512320316e-06,
      "loss": 0.0338,
      "step": 526
    },
    {
      "epoch": 1.5778443113772456,
      "grad_norm": 1.157736929654406,
      "learning_rate": 1.0596245373222424e-06,
      "loss": 0.0321,
      "step": 527
    },
    {
      "epoch": 1.5808383233532934,
      "grad_norm": 1.0373822014760812,
      "learning_rate": 1.045192876670298e-06,
      "loss": 0.0372,
      "step": 528
    },
    {
      "epoch": 1.5838323353293413,
      "grad_norm": 0.9909651388443135,
      "learning_rate": 1.0308486884757868e-06,
      "loss": 0.0333,
      "step": 529
    },
    {
      "epoch": 1.5868263473053892,
      "grad_norm": 1.0379982375258758,
      "learning_rate": 1.0165922900035886e-06,
      "loss": 0.0351,
      "step": 530
    },
    {
      "epoch": 1.589820359281437,
      "grad_norm": 0.6992586348320551,
      "learning_rate": 1.0024239965768417e-06,
      "loss": 0.0262,
      "step": 531
    },
    {
      "epoch": 1.592814371257485,
      "grad_norm": 0.8421730639029076,
      "learning_rate": 9.883441215699824e-07,
      "loss": 0.0263,
      "step": 532
    },
    {
      "epoch": 1.5958083832335328,
      "grad_norm": 0.9589895089553961,
      "learning_rate": 9.74352976401805e-07,
      "loss": 0.0324,
      "step": 533
    },
    {
      "epoch": 1.5988023952095807,
      "grad_norm": 1.042854272088062,
      "learning_rate": 9.604508705285765e-07,
      "loss": 0.0416,
      "step": 534
    },
    {
      "epoch": 1.6017964071856288,
      "grad_norm": 0.8894391680252648,
      "learning_rate": 9.466381114371942e-07,
      "loss": 0.0359,
      "step": 535
    },
    {
      "epoch": 1.6047904191616766,
      "grad_norm": 1.0166265256585618,
      "learning_rate": 9.329150046383773e-07,
      "loss": 0.0423,
      "step": 536
    },
    {
      "epoch": 1.6077844311377245,
      "grad_norm": 0.7996635216062675,
      "learning_rate": 9.192818536599213e-07,
      "loss": 0.0324,
      "step": 537
    },
    {
      "epoch": 1.6107784431137726,
      "grad_norm": 0.9546392882250924,
      "learning_rate": 9.057389600399719e-07,
      "loss": 0.0378,
      "step": 538
    },
    {
      "epoch": 1.6137724550898205,
      "grad_norm": 0.894516590426351,
      "learning_rate": 8.922866233203681e-07,
      "loss": 0.0418,
      "step": 539
    },
    {
      "epoch": 1.6167664670658684,
      "grad_norm": 0.7960780180162699,
      "learning_rate": 8.789251410400024e-07,
      "loss": 0.0305,
      "step": 540
    },
    {
      "epoch": 1.6197604790419162,
      "grad_norm": 0.8300234901707579,
      "learning_rate": 8.65654808728259e-07,
      "loss": 0.0354,
      "step": 541
    },
    {
      "epoch": 1.622754491017964,
      "grad_norm": 1.0205604631616625,
      "learning_rate": 8.524759198984567e-07,
      "loss": 0.0361,
      "step": 542
    },
    {
      "epoch": 1.625748502994012,
      "grad_norm": 0.9936148863087529,
      "learning_rate": 8.393887660413719e-07,
      "loss": 0.0409,
      "step": 543
    },
    {
      "epoch": 1.6287425149700598,
      "grad_norm": 1.0698907340443171,
      "learning_rate": 8.263936366187825e-07,
      "loss": 0.0398,
      "step": 544
    },
    {
      "epoch": 1.6317365269461077,
      "grad_norm": 0.7855987754368887,
      "learning_rate": 8.134908190570723e-07,
      "loss": 0.0315,
      "step": 545
    },
    {
      "epoch": 1.6347305389221556,
      "grad_norm": 1.020509696045261,
      "learning_rate": 8.006805987408705e-07,
      "loss": 0.0374,
      "step": 546
    },
    {
      "epoch": 1.6377245508982035,
      "grad_norm": 0.8915003246811914,
      "learning_rate": 7.879632590067354e-07,
      "loss": 0.0364,
      "step": 547
    },
    {
      "epoch": 1.6407185628742516,
      "grad_norm": 0.7662888515803956,
      "learning_rate": 7.753390811368972e-07,
      "loss": 0.0301,
      "step": 548
    },
    {
      "epoch": 1.6437125748502994,
      "grad_norm": 1.106800491593679,
      "learning_rate": 7.628083443530287e-07,
      "loss": 0.0397,
      "step": 549
    },
    {
      "epoch": 1.6467065868263473,
      "grad_norm": 1.0183820794919076,
      "learning_rate": 7.503713258100726e-07,
      "loss": 0.0352,
      "step": 550
    },
    {
      "epoch": 1.6497005988023952,
      "grad_norm": 0.9777665223339046,
      "learning_rate": 7.380283005901084e-07,
      "loss": 0.0416,
      "step": 551
    },
    {
      "epoch": 1.6526946107784433,
      "grad_norm": 0.8767222130909182,
      "learning_rate": 7.257795416962754e-07,
      "loss": 0.0314,
      "step": 552
    },
    {
      "epoch": 1.6556886227544911,
      "grad_norm": 0.737287272693723,
      "learning_rate": 7.136253200467231e-07,
      "loss": 0.0198,
      "step": 553
    },
    {
      "epoch": 1.658682634730539,
      "grad_norm": 0.7835491381802023,
      "learning_rate": 7.015659044686307e-07,
      "loss": 0.0246,
      "step": 554
    },
    {
      "epoch": 1.6616766467065869,
      "grad_norm": 0.8598495214934521,
      "learning_rate": 6.896015616922535e-07,
      "loss": 0.0341,
      "step": 555
    },
    {
      "epoch": 1.6646706586826348,
      "grad_norm": 1.025286450985692,
      "learning_rate": 6.777325563450282e-07,
      "loss": 0.0378,
      "step": 556
    },
    {
      "epoch": 1.6676646706586826,
      "grad_norm": 0.9327848641100186,
      "learning_rate": 6.659591509457125e-07,
      "loss": 0.0379,
      "step": 557
    },
    {
      "epoch": 1.6706586826347305,
      "grad_norm": 0.7618682808268273,
      "learning_rate": 6.542816058985896e-07,
      "loss": 0.0228,
      "step": 558
    },
    {
      "epoch": 1.6736526946107784,
      "grad_norm": 0.9284164548910486,
      "learning_rate": 6.427001794876974e-07,
      "loss": 0.0333,
      "step": 559
    },
    {
      "epoch": 1.6766467065868262,
      "grad_norm": 0.8004605526189144,
      "learning_rate": 6.312151278711237e-07,
      "loss": 0.024,
      "step": 560
    },
    {
      "epoch": 1.6796407185628741,
      "grad_norm": 0.8379010268641507,
      "learning_rate": 6.198267050753387e-07,
      "loss": 0.0304,
      "step": 561
    },
    {
      "epoch": 1.6826347305389222,
      "grad_norm": 0.9707432222175602,
      "learning_rate": 6.085351629895736e-07,
      "loss": 0.0359,
      "step": 562
    },
    {
      "epoch": 1.68562874251497,
      "grad_norm": 0.9631945134924423,
      "learning_rate": 5.973407513602514e-07,
      "loss": 0.0351,
      "step": 563
    },
    {
      "epoch": 1.688622754491018,
      "grad_norm": 0.9983692507383731,
      "learning_rate": 5.862437177854629e-07,
      "loss": 0.0387,
      "step": 564
    },
    {
      "epoch": 1.6916167664670658,
      "grad_norm": 1.0296716244745396,
      "learning_rate": 5.752443077094927e-07,
      "loss": 0.0339,
      "step": 565
    },
    {
      "epoch": 1.694610778443114,
      "grad_norm": 0.8656177042157595,
      "learning_rate": 5.643427644173838e-07,
      "loss": 0.0288,
      "step": 566
    },
    {
      "epoch": 1.6976047904191618,
      "grad_norm": 1.0750573274794817,
      "learning_rate": 5.535393290295643e-07,
      "loss": 0.04,
      "step": 567
    },
    {
      "epoch": 1.7005988023952097,
      "grad_norm": 0.9210171117202798,
      "learning_rate": 5.428342404965076e-07,
      "loss": 0.0366,
      "step": 568
    },
    {
      "epoch": 1.7035928143712575,
      "grad_norm": 0.960309870629056,
      "learning_rate": 5.322277355934557e-07,
      "loss": 0.0337,
      "step": 569
    },
    {
      "epoch": 1.7065868263473054,
      "grad_norm": 0.8679487391126443,
      "learning_rate": 5.217200489151714e-07,
      "loss": 0.0294,
      "step": 570
    },
    {
      "epoch": 1.7095808383233533,
      "grad_norm": 0.890094418447598,
      "learning_rate": 5.113114128707592e-07,
      "loss": 0.0306,
      "step": 571
    },
    {
      "epoch": 1.7125748502994012,
      "grad_norm": 0.9155220410515212,
      "learning_rate": 5.010020576785174e-07,
      "loss": 0.0318,
      "step": 572
    },
    {
      "epoch": 1.715568862275449,
      "grad_norm": 0.8104746870230927,
      "learning_rate": 4.907922113608532e-07,
      "loss": 0.0269,
      "step": 573
    },
    {
      "epoch": 1.718562874251497,
      "grad_norm": 0.9157292905853648,
      "learning_rate": 4.806820997392325e-07,
      "loss": 0.038,
      "step": 574
    },
    {
      "epoch": 1.7215568862275448,
      "grad_norm": 0.6691444205807425,
      "learning_rate": 4.7067194642919036e-07,
      "loss": 0.0227,
      "step": 575
    },
    {
      "epoch": 1.7245508982035929,
      "grad_norm": 0.9425706995318925,
      "learning_rate": 4.607619728353818e-07,
      "loss": 0.0302,
      "step": 576
    },
    {
      "epoch": 1.7275449101796407,
      "grad_norm": 0.9147873238989198,
      "learning_rate": 4.50952398146689e-07,
      "loss": 0.032,
      "step": 577
    },
    {
      "epoch": 1.7305389221556886,
      "grad_norm": 0.787164386303056,
      "learning_rate": 4.4124343933136525e-07,
      "loss": 0.0262,
      "step": 578
    },
    {
      "epoch": 1.7335329341317365,
      "grad_norm": 0.9199516480637124,
      "learning_rate": 4.3163531113224466e-07,
      "loss": 0.0331,
      "step": 579
    },
    {
      "epoch": 1.7365269461077846,
      "grad_norm": 1.1044273534432067,
      "learning_rate": 4.221282260619891e-07,
      "loss": 0.0426,
      "step": 580
    },
    {
      "epoch": 1.7395209580838324,
      "grad_norm": 0.7937167954640157,
      "learning_rate": 4.127223943983849e-07,
      "loss": 0.0286,
      "step": 581
    },
    {
      "epoch": 1.7425149700598803,
      "grad_norm": 0.8762851988093123,
      "learning_rate": 4.03418024179697e-07,
      "loss": 0.0283,
      "step": 582
    },
    {
      "epoch": 1.7455089820359282,
      "grad_norm": 1.0817141631161138,
      "learning_rate": 3.9421532120006544e-07,
      "loss": 0.0438,
      "step": 583
    },
    {
      "epoch": 1.748502994011976,
      "grad_norm": 0.8365384535655536,
      "learning_rate": 3.851144890049535e-07,
      "loss": 0.0321,
      "step": 584
    },
    {
      "epoch": 1.751497005988024,
      "grad_norm": 0.911256784116802,
      "learning_rate": 3.761157288866418e-07,
      "loss": 0.0386,
      "step": 585
    },
    {
      "epoch": 1.7544910179640718,
      "grad_norm": 1.1656282957154653,
      "learning_rate": 3.672192398797858e-07,
      "loss": 0.045,
      "step": 586
    },
    {
      "epoch": 1.7574850299401197,
      "grad_norm": 0.9018546582132964,
      "learning_rate": 3.58425218757002e-07,
      "loss": 0.0349,
      "step": 587
    },
    {
      "epoch": 1.7604790419161676,
      "grad_norm": 0.9492299947761066,
      "learning_rate": 3.497338600245254e-07,
      "loss": 0.0269,
      "step": 588
    },
    {
      "epoch": 1.7634730538922154,
      "grad_norm": 0.8198456203128308,
      "learning_rate": 3.4114535591790233e-07,
      "loss": 0.0267,
      "step": 589
    },
    {
      "epoch": 1.7664670658682635,
      "grad_norm": 0.9112940051841174,
      "learning_rate": 3.326598963977395e-07,
      "loss": 0.0274,
      "step": 590
    },
    {
      "epoch": 1.7694610778443114,
      "grad_norm": 1.0327312258181385,
      "learning_rate": 3.242776691455013e-07,
      "loss": 0.0363,
      "step": 591
    },
    {
      "epoch": 1.7724550898203593,
      "grad_norm": 0.9957551221120605,
      "learning_rate": 3.159988595593616e-07,
      "loss": 0.0332,
      "step": 592
    },
    {
      "epoch": 1.7754491017964071,
      "grad_norm": 0.82728590031526,
      "learning_rate": 3.078236507501015e-07,
      "loss": 0.0317,
      "step": 593
    },
    {
      "epoch": 1.7784431137724552,
      "grad_norm": 0.8968464525097516,
      "learning_rate": 2.9975222353705757e-07,
      "loss": 0.034,
      "step": 594
    },
    {
      "epoch": 1.781437125748503,
      "grad_norm": 0.9377205077516637,
      "learning_rate": 2.917847564441256e-07,
      "loss": 0.0331,
      "step": 595
    },
    {
      "epoch": 1.784431137724551,
      "grad_norm": 0.8665273534022907,
      "learning_rate": 2.839214256958106e-07,
      "loss": 0.0378,
      "step": 596
    },
    {
      "epoch": 1.7874251497005988,
      "grad_norm": 0.8257179290605647,
      "learning_rate": 2.7616240521332884e-07,
      "loss": 0.0303,
      "step": 597
    },
    {
      "epoch": 1.7904191616766467,
      "grad_norm": 1.1306970186188008,
      "learning_rate": 2.6850786661076047e-07,
      "loss": 0.0361,
      "step": 598
    },
    {
      "epoch": 1.7934131736526946,
      "grad_norm": 1.0092679266996627,
      "learning_rate": 2.6095797919125533e-07,
      "loss": 0.0302,
      "step": 599
    },
    {
      "epoch": 1.7964071856287425,
      "grad_norm": 0.8951248327569581,
      "learning_rate": 2.5351290994328703e-07,
      "loss": 0.0307,
      "step": 600
    },
    {
      "epoch": 1.7964071856287425,
      "eval_loss": 0.13603587448596954,
      "eval_runtime": 3.3465,
      "eval_samples_per_second": 16.136,
      "eval_steps_per_second": 4.183,
      "step": 600
    },
    {
      "epoch": 1.7994011976047903,
      "grad_norm": 1.0151980592476524,
      "learning_rate": 2.4617282353696093e-07,
      "loss": 0.0344,
      "step": 601
    },
    {
      "epoch": 1.8023952095808382,
      "grad_norm": 1.1052408841197836,
      "learning_rate": 2.3893788232036807e-07,
      "loss": 0.0393,
      "step": 602
    },
    {
      "epoch": 1.805389221556886,
      "grad_norm": 0.9072023909933259,
      "learning_rate": 2.318082463160032e-07,
      "loss": 0.0367,
      "step": 603
    },
    {
      "epoch": 1.8083832335329342,
      "grad_norm": 0.8741407229337965,
      "learning_rate": 2.2478407321721295e-07,
      "loss": 0.033,
      "step": 604
    },
    {
      "epoch": 1.811377245508982,
      "grad_norm": 1.0423052115599827,
      "learning_rate": 2.1786551838471892e-07,
      "loss": 0.041,
      "step": 605
    },
    {
      "epoch": 1.81437125748503,
      "grad_norm": 0.9178025937169011,
      "learning_rate": 2.1105273484317402e-07,
      "loss": 0.0295,
      "step": 606
    },
    {
      "epoch": 1.8173652694610778,
      "grad_norm": 0.6910732944767047,
      "learning_rate": 2.043458732777831e-07,
      "loss": 0.023,
      "step": 607
    },
    {
      "epoch": 1.8203592814371259,
      "grad_norm": 1.1262295342973343,
      "learning_rate": 1.9774508203096843e-07,
      "loss": 0.0381,
      "step": 608
    },
    {
      "epoch": 1.8233532934131738,
      "grad_norm": 0.7756040211730907,
      "learning_rate": 1.9125050709908388e-07,
      "loss": 0.024,
      "step": 609
    },
    {
      "epoch": 1.8263473053892216,
      "grad_norm": 1.0385200457199717,
      "learning_rate": 1.8486229212919482e-07,
      "loss": 0.0373,
      "step": 610
    },
    {
      "epoch": 1.8293413173652695,
      "grad_norm": 0.8186709978978579,
      "learning_rate": 1.7858057841589281e-07,
      "loss": 0.0282,
      "step": 611
    },
    {
      "epoch": 1.8323353293413174,
      "grad_norm": 0.8430090256482891,
      "learning_rate": 1.7240550489817652e-07,
      "loss": 0.0276,
      "step": 612
    },
    {
      "epoch": 1.8353293413173652,
      "grad_norm": 1.0176443399790533,
      "learning_rate": 1.66337208156373e-07,
      "loss": 0.0342,
      "step": 613
    },
    {
      "epoch": 1.8383233532934131,
      "grad_norm": 1.0821364283109127,
      "learning_rate": 1.6037582240912175e-07,
      "loss": 0.0391,
      "step": 614
    },
    {
      "epoch": 1.841317365269461,
      "grad_norm": 0.9772939807227844,
      "learning_rate": 1.5452147951040165e-07,
      "loss": 0.0378,
      "step": 615
    },
    {
      "epoch": 1.8443113772455089,
      "grad_norm": 1.0361685598370454,
      "learning_rate": 1.4877430894662037e-07,
      "loss": 0.0336,
      "step": 616
    },
    {
      "epoch": 1.8473053892215567,
      "grad_norm": 0.8253724574892053,
      "learning_rate": 1.4313443783374405e-07,
      "loss": 0.0311,
      "step": 617
    },
    {
      "epoch": 1.8502994011976048,
      "grad_norm": 0.8716149729451586,
      "learning_rate": 1.3760199091449045e-07,
      "loss": 0.0284,
      "step": 618
    },
    {
      "epoch": 1.8532934131736527,
      "grad_norm": 0.9885390026531578,
      "learning_rate": 1.3217709055556638e-07,
      "loss": 0.0366,
      "step": 619
    },
    {
      "epoch": 1.8562874251497006,
      "grad_norm": 0.9837223721509337,
      "learning_rate": 1.268598567449647e-07,
      "loss": 0.0338,
      "step": 620
    },
    {
      "epoch": 1.8592814371257484,
      "grad_norm": 0.9331185727946466,
      "learning_rate": 1.2165040708930763e-07,
      "loss": 0.0335,
      "step": 621
    },
    {
      "epoch": 1.8622754491017965,
      "grad_norm": 0.6593203199388472,
      "learning_rate": 1.1654885681124661e-07,
      "loss": 0.0284,
      "step": 622
    },
    {
      "epoch": 1.8652694610778444,
      "grad_norm": 0.9485397691527565,
      "learning_rate": 1.1155531874691372e-07,
      "loss": 0.0342,
      "step": 623
    },
    {
      "epoch": 1.8682634730538923,
      "grad_norm": 0.8799238219833875,
      "learning_rate": 1.0666990334342708e-07,
      "loss": 0.028,
      "step": 624
    },
    {
      "epoch": 1.8712574850299402,
      "grad_norm": 0.9495394462700678,
      "learning_rate": 1.0189271865644445e-07,
      "loss": 0.0332,
      "step": 625
    },
    {
      "epoch": 1.874251497005988,
      "grad_norm": 1.1019991417583253,
      "learning_rate": 9.722387034777847e-08,
      "loss": 0.0384,
      "step": 626
    },
    {
      "epoch": 1.877245508982036,
      "grad_norm": 0.8775253683539328,
      "learning_rate": 9.266346168305518e-08,
      "loss": 0.0406,
      "step": 627
    },
    {
      "epoch": 1.8802395209580838,
      "grad_norm": 0.7906028947762891,
      "learning_rate": 8.821159352943142e-08,
      "loss": 0.0276,
      "step": 628
    },
    {
      "epoch": 1.8832335329341316,
      "grad_norm": 0.8195621149974784,
      "learning_rate": 8.38683643533661e-08,
      "loss": 0.0278,
      "step": 629
    },
    {
      "epoch": 1.8862275449101795,
      "grad_norm": 0.8399557593025311,
      "learning_rate": 7.963387021843683e-08,
      "loss": 0.0251,
      "step": 630
    },
    {
      "epoch": 1.8892215568862274,
      "grad_norm": 1.0637811605024765,
      "learning_rate": 7.550820478322285e-08,
      "loss": 0.0386,
      "step": 631
    },
    {
      "epoch": 1.8922155688622755,
      "grad_norm": 0.9403156124852005,
      "learning_rate": 7.149145929922607e-08,
      "loss": 0.0434,
      "step": 632
    },
    {
      "epoch": 1.8952095808383234,
      "grad_norm": 1.1693945887368118,
      "learning_rate": 6.758372260885714e-08,
      "loss": 0.0371,
      "step": 633
    },
    {
      "epoch": 1.8982035928143712,
      "grad_norm": 1.107074739505125,
      "learning_rate": 6.378508114346982e-08,
      "loss": 0.0433,
      "step": 634
    },
    {
      "epoch": 1.9011976047904193,
      "grad_norm": 0.8071962038303815,
      "learning_rate": 6.009561892144744e-08,
      "loss": 0.0274,
      "step": 635
    },
    {
      "epoch": 1.9041916167664672,
      "grad_norm": 0.92253988620008,
      "learning_rate": 5.651541754634726e-08,
      "loss": 0.0264,
      "step": 636
    },
    {
      "epoch": 1.907185628742515,
      "grad_norm": 0.9569167295082278,
      "learning_rate": 5.304455620509297e-08,
      "loss": 0.0344,
      "step": 637
    },
    {
      "epoch": 1.910179640718563,
      "grad_norm": 1.032106554513129,
      "learning_rate": 4.968311166622553e-08,
      "loss": 0.0316,
      "step": 638
    },
    {
      "epoch": 1.9131736526946108,
      "grad_norm": 1.0791945091607877,
      "learning_rate": 4.643115827820399e-08,
      "loss": 0.0407,
      "step": 639
    },
    {
      "epoch": 1.9161676646706587,
      "grad_norm": 0.9447935733472742,
      "learning_rate": 4.328876796776071e-08,
      "loss": 0.0289,
      "step": 640
    },
    {
      "epoch": 1.9191616766467066,
      "grad_norm": 0.903012748217911,
      "learning_rate": 4.0256010238310936e-08,
      "loss": 0.0347,
      "step": 641
    },
    {
      "epoch": 1.9221556886227544,
      "grad_norm": 0.9160983682975663,
      "learning_rate": 3.733295216841626e-08,
      "loss": 0.0317,
      "step": 642
    },
    {
      "epoch": 1.9251497005988023,
      "grad_norm": 1.0724780870724786,
      "learning_rate": 3.451965841029914e-08,
      "loss": 0.0365,
      "step": 643
    },
    {
      "epoch": 1.9281437125748502,
      "grad_norm": 0.7446247129335253,
      "learning_rate": 3.181619118841517e-08,
      "loss": 0.0218,
      "step": 644
    },
    {
      "epoch": 1.931137724550898,
      "grad_norm": 0.7748088900862936,
      "learning_rate": 2.9222610298074717e-08,
      "loss": 0.0236,
      "step": 645
    },
    {
      "epoch": 1.9341317365269461,
      "grad_norm": 0.9108670005189082,
      "learning_rate": 2.673897310412288e-08,
      "loss": 0.0367,
      "step": 646
    },
    {
      "epoch": 1.937125748502994,
      "grad_norm": 0.7295769821012579,
      "learning_rate": 2.4365334539667717e-08,
      "loss": 0.024,
      "step": 647
    },
    {
      "epoch": 1.9401197604790419,
      "grad_norm": 0.954838353986231,
      "learning_rate": 2.210174710486679e-08,
      "loss": 0.0367,
      "step": 648
    },
    {
      "epoch": 1.94311377245509,
      "grad_norm": 0.7323795466807219,
      "learning_rate": 1.99482608657664e-08,
      "loss": 0.0269,
      "step": 649
    },
    {
      "epoch": 1.9461077844311379,
      "grad_norm": 0.8061066446238484,
      "learning_rate": 1.7904923453193056e-08,
      "loss": 0.0283,
      "step": 650
    },
    {
      "epoch": 1.9491017964071857,
      "grad_norm": 1.0369116800719558,
      "learning_rate": 1.5971780061701524e-08,
      "loss": 0.0341,
      "step": 651
    },
    {
      "epoch": 1.9520958083832336,
      "grad_norm": 0.7014387629697275,
      "learning_rate": 1.4148873448573408e-08,
      "loss": 0.0237,
      "step": 652
    },
    {
      "epoch": 1.9550898203592815,
      "grad_norm": 0.8904022538460379,
      "learning_rate": 1.2436243932872349e-08,
      "loss": 0.0399,
      "step": 653
    },
    {
      "epoch": 1.9580838323353293,
      "grad_norm": 0.908942954528451,
      "learning_rate": 1.0833929394552523e-08,
      "loss": 0.0292,
      "step": 654
    },
    {
      "epoch": 1.9610778443113772,
      "grad_norm": 0.7651593283878957,
      "learning_rate": 9.341965273621522e-09,
      "loss": 0.0268,
      "step": 655
    },
    {
      "epoch": 1.964071856287425,
      "grad_norm": 1.1609375557337298,
      "learning_rate": 7.96038456935322e-09,
      "loss": 0.0448,
      "step": 656
    },
    {
      "epoch": 1.967065868263473,
      "grad_norm": 1.1485353834530596,
      "learning_rate": 6.6892178395611125e-09,
      "loss": 0.0418,
      "step": 657
    },
    {
      "epoch": 1.9700598802395208,
      "grad_norm": 1.1215592554398053,
      "learning_rate": 5.528493199922769e-09,
      "loss": 0.0336,
      "step": 658
    },
    {
      "epoch": 1.9730538922155687,
      "grad_norm": 0.8464519640849736,
      "learning_rate": 4.478236323355312e-09,
      "loss": 0.0316,
      "step": 659
    },
    {
      "epoch": 1.9760479041916168,
      "grad_norm": 0.9035220077836342,
      "learning_rate": 3.538470439448105e-09,
      "loss": 0.0323,
      "step": 660
    },
    {
      "epoch": 1.9790419161676647,
      "grad_norm": 0.9860038452035425,
      "learning_rate": 2.709216333952602e-09,
      "loss": 0.0378,
      "step": 661
    },
    {
      "epoch": 1.9820359281437125,
      "grad_norm": 0.7720852780845451,
      "learning_rate": 1.9904923483171632e-09,
      "loss": 0.0321,
      "step": 662
    },
    {
      "epoch": 1.9850299401197606,
      "grad_norm": 1.0972720970039895,
      "learning_rate": 1.3823143792851545e-09,
      "loss": 0.0406,
      "step": 663
    },
    {
      "epoch": 1.9880239520958085,
      "grad_norm": 1.105002813112153,
      "learning_rate": 8.846958785418969e-10,
      "loss": 0.0417,
      "step": 664
    },
    {
      "epoch": 1.9910179640718564,
      "grad_norm": 0.8375778942299396,
      "learning_rate": 4.97647852417682e-10,
      "loss": 0.0295,
      "step": 665
    },
    {
      "epoch": 1.9940119760479043,
      "grad_norm": 0.9200127903581127,
      "learning_rate": 2.2117886164407797e-10,
      "loss": 0.03,
      "step": 666
    },
    {
      "epoch": 1.9970059880239521,
      "grad_norm": 0.7565807823421462,
      "learning_rate": 5.529502116519148e-11,
      "loss": 0.028,
      "step": 667
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.48533526453125986,
      "learning_rate": 0.0,
      "loss": 0.0124,
      "step": 668
    },
    {
      "epoch": 2.0,
      "step": 668,
      "total_flos": 12178659213312.0,
      "train_loss": 0.07412140557826859,
      "train_runtime": 870.6282,
      "train_samples_per_second": 12.242,
      "train_steps_per_second": 0.767
    }
  ],
  "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": 12178659213312.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}