{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9734939759036143,
  "eval_steps": 500,
  "global_step": 309,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00963855421686747,
      "grad_norm": 0.7321725047335572,
      "learning_rate": 3.2258064516129035e-07,
      "loss": 1.2207,
      "step": 1
    },
    {
      "epoch": 0.01927710843373494,
      "grad_norm": 0.6665264239249077,
      "learning_rate": 6.451612903225807e-07,
      "loss": 1.1371,
      "step": 2
    },
    {
      "epoch": 0.02891566265060241,
      "grad_norm": 0.7664904235774216,
      "learning_rate": 9.67741935483871e-07,
      "loss": 1.1605,
      "step": 3
    },
    {
      "epoch": 0.03855421686746988,
      "grad_norm": 0.8973293033776428,
      "learning_rate": 1.2903225806451614e-06,
      "loss": 1.297,
      "step": 4
    },
    {
      "epoch": 0.04819277108433735,
      "grad_norm": 0.7512289668671644,
      "learning_rate": 1.6129032258064516e-06,
      "loss": 1.2186,
      "step": 5
    },
    {
      "epoch": 0.05783132530120482,
      "grad_norm": 0.5784006362716563,
      "learning_rate": 1.935483870967742e-06,
      "loss": 1.1326,
      "step": 6
    },
    {
      "epoch": 0.06746987951807229,
      "grad_norm": 0.6067121698335344,
      "learning_rate": 2.2580645161290324e-06,
      "loss": 1.1416,
      "step": 7
    },
    {
      "epoch": 0.07710843373493977,
      "grad_norm": 0.5822439970894898,
      "learning_rate": 2.580645161290323e-06,
      "loss": 1.2283,
      "step": 8
    },
    {
      "epoch": 0.08674698795180723,
      "grad_norm": 0.43410811408302385,
      "learning_rate": 2.903225806451613e-06,
      "loss": 1.1301,
      "step": 9
    },
    {
      "epoch": 0.0963855421686747,
      "grad_norm": 0.44969321494204945,
      "learning_rate": 3.225806451612903e-06,
      "loss": 1.1998,
      "step": 10
    },
    {
      "epoch": 0.10602409638554217,
      "grad_norm": 0.40286147754479423,
      "learning_rate": 3.548387096774194e-06,
      "loss": 1.1185,
      "step": 11
    },
    {
      "epoch": 0.11566265060240964,
      "grad_norm": 0.5535764409411175,
      "learning_rate": 3.870967741935484e-06,
      "loss": 1.0715,
      "step": 12
    },
    {
      "epoch": 0.12530120481927712,
      "grad_norm": 0.48960428118771093,
      "learning_rate": 4.193548387096774e-06,
      "loss": 1.0387,
      "step": 13
    },
    {
      "epoch": 0.13493975903614458,
      "grad_norm": 0.38578430703859284,
      "learning_rate": 4.516129032258065e-06,
      "loss": 1.0138,
      "step": 14
    },
    {
      "epoch": 0.14457831325301204,
      "grad_norm": 0.36944902077760366,
      "learning_rate": 4.838709677419355e-06,
      "loss": 1.0815,
      "step": 15
    },
    {
      "epoch": 0.15421686746987953,
      "grad_norm": 0.22615568618733844,
      "learning_rate": 5.161290322580646e-06,
      "loss": 0.9954,
      "step": 16
    },
    {
      "epoch": 0.163855421686747,
      "grad_norm": 0.16924466530397725,
      "learning_rate": 5.483870967741935e-06,
      "loss": 0.9345,
      "step": 17
    },
    {
      "epoch": 0.17349397590361446,
      "grad_norm": 0.17506858257513364,
      "learning_rate": 5.806451612903226e-06,
      "loss": 1.0057,
      "step": 18
    },
    {
      "epoch": 0.18313253012048192,
      "grad_norm": 0.19256260025141983,
      "learning_rate": 6.129032258064517e-06,
      "loss": 0.961,
      "step": 19
    },
    {
      "epoch": 0.1927710843373494,
      "grad_norm": 0.20045422860718864,
      "learning_rate": 6.451612903225806e-06,
      "loss": 0.9732,
      "step": 20
    },
    {
      "epoch": 0.20240963855421687,
      "grad_norm": 0.20006737983005313,
      "learning_rate": 6.774193548387097e-06,
      "loss": 0.9114,
      "step": 21
    },
    {
      "epoch": 0.21204819277108433,
      "grad_norm": 0.1804762390245476,
      "learning_rate": 7.096774193548388e-06,
      "loss": 0.9238,
      "step": 22
    },
    {
      "epoch": 0.2216867469879518,
      "grad_norm": 0.19132885097999205,
      "learning_rate": 7.4193548387096784e-06,
      "loss": 0.9423,
      "step": 23
    },
    {
      "epoch": 0.23132530120481928,
      "grad_norm": 0.1946117127819535,
      "learning_rate": 7.741935483870968e-06,
      "loss": 0.8987,
      "step": 24
    },
    {
      "epoch": 0.24096385542168675,
      "grad_norm": 0.21044619139543286,
      "learning_rate": 8.064516129032258e-06,
      "loss": 0.923,
      "step": 25
    },
    {
      "epoch": 0.25060240963855424,
      "grad_norm": 0.1894635936827489,
      "learning_rate": 8.387096774193549e-06,
      "loss": 0.8432,
      "step": 26
    },
    {
      "epoch": 0.26024096385542167,
      "grad_norm": 0.2213813253619953,
      "learning_rate": 8.70967741935484e-06,
      "loss": 0.8921,
      "step": 27
    },
    {
      "epoch": 0.26987951807228916,
      "grad_norm": 0.16421014736345188,
      "learning_rate": 9.03225806451613e-06,
      "loss": 0.891,
      "step": 28
    },
    {
      "epoch": 0.27951807228915665,
      "grad_norm": 0.15888367691754784,
      "learning_rate": 9.35483870967742e-06,
      "loss": 0.8328,
      "step": 29
    },
    {
      "epoch": 0.2891566265060241,
      "grad_norm": 0.1218262296356629,
      "learning_rate": 9.67741935483871e-06,
      "loss": 0.8109,
      "step": 30
    },
    {
      "epoch": 0.2987951807228916,
      "grad_norm": 0.12772522304462855,
      "learning_rate": 1e-05,
      "loss": 0.8327,
      "step": 31
    },
    {
      "epoch": 0.30843373493975906,
      "grad_norm": 0.11662156818152736,
      "learning_rate": 9.999680739242022e-06,
      "loss": 0.8228,
      "step": 32
    },
    {
      "epoch": 0.3180722891566265,
      "grad_norm": 0.11423729898798743,
      "learning_rate": 9.99872299773906e-06,
      "loss": 0.8184,
      "step": 33
    },
    {
      "epoch": 0.327710843373494,
      "grad_norm": 0.10732392773580121,
      "learning_rate": 9.997126897798826e-06,
      "loss": 0.8431,
      "step": 34
    },
    {
      "epoch": 0.3373493975903614,
      "grad_norm": 0.10533931104652113,
      "learning_rate": 9.994892643250147e-06,
      "loss": 0.8436,
      "step": 35
    },
    {
      "epoch": 0.3469879518072289,
      "grad_norm": 0.10349233667045869,
      "learning_rate": 9.99202051941695e-06,
      "loss": 0.8124,
      "step": 36
    },
    {
      "epoch": 0.3566265060240964,
      "grad_norm": 0.12234917986226827,
      "learning_rate": 9.9885108930818e-06,
      "loss": 0.8402,
      "step": 37
    },
    {
      "epoch": 0.36626506024096384,
      "grad_norm": 0.1084904559104385,
      "learning_rate": 9.984364212439089e-06,
      "loss": 0.7863,
      "step": 38
    },
    {
      "epoch": 0.3759036144578313,
      "grad_norm": 0.09315599863192536,
      "learning_rate": 9.979581007037776e-06,
      "loss": 0.7494,
      "step": 39
    },
    {
      "epoch": 0.3855421686746988,
      "grad_norm": 0.10455770281109837,
      "learning_rate": 9.974161887713775e-06,
      "loss": 0.7733,
      "step": 40
    },
    {
      "epoch": 0.39518072289156625,
      "grad_norm": 0.09713116343076955,
      "learning_rate": 9.968107546511942e-06,
      "loss": 0.8069,
      "step": 41
    },
    {
      "epoch": 0.40481927710843374,
      "grad_norm": 0.08994648197042518,
      "learning_rate": 9.961418756597703e-06,
      "loss": 0.7544,
      "step": 42
    },
    {
      "epoch": 0.41445783132530123,
      "grad_norm": 0.08619129061295391,
      "learning_rate": 9.95409637215831e-06,
      "loss": 0.8046,
      "step": 43
    },
    {
      "epoch": 0.42409638554216866,
      "grad_norm": 0.07881839593291304,
      "learning_rate": 9.94614132829377e-06,
      "loss": 0.7887,
      "step": 44
    },
    {
      "epoch": 0.43373493975903615,
      "grad_norm": 0.08828454670099575,
      "learning_rate": 9.937554640897414e-06,
      "loss": 0.806,
      "step": 45
    },
    {
      "epoch": 0.4433734939759036,
      "grad_norm": 0.08434767386597816,
      "learning_rate": 9.928337406526172e-06,
      "loss": 0.8187,
      "step": 46
    },
    {
      "epoch": 0.4530120481927711,
      "grad_norm": 0.07330130217165154,
      "learning_rate": 9.918490802260538e-06,
      "loss": 0.7442,
      "step": 47
    },
    {
      "epoch": 0.46265060240963857,
      "grad_norm": 0.08355675961610799,
      "learning_rate": 9.90801608555425e-06,
      "loss": 0.7288,
      "step": 48
    },
    {
      "epoch": 0.472289156626506,
      "grad_norm": 0.08550275063984396,
      "learning_rate": 9.896914594073703e-06,
      "loss": 0.7876,
      "step": 49
    },
    {
      "epoch": 0.4819277108433735,
      "grad_norm": 0.08574284988583232,
      "learning_rate": 9.885187745527132e-06,
      "loss": 0.7958,
      "step": 50
    },
    {
      "epoch": 0.491566265060241,
      "grad_norm": 0.07092583421564527,
      "learning_rate": 9.87283703748356e-06,
      "loss": 0.7526,
      "step": 51
    },
    {
      "epoch": 0.5012048192771085,
      "grad_norm": 0.08160685477674649,
      "learning_rate": 9.859864047181551e-06,
      "loss": 0.771,
      "step": 52
    },
    {
      "epoch": 0.5108433734939759,
      "grad_norm": 0.08388500089586182,
      "learning_rate": 9.846270431327793e-06,
      "loss": 0.7916,
      "step": 53
    },
    {
      "epoch": 0.5204819277108433,
      "grad_norm": 0.07274717865251835,
      "learning_rate": 9.832057925885526e-06,
      "loss": 0.7394,
      "step": 54
    },
    {
      "epoch": 0.5301204819277109,
      "grad_norm": 0.07939094338552345,
      "learning_rate": 9.817228345852853e-06,
      "loss": 0.7336,
      "step": 55
    },
    {
      "epoch": 0.5397590361445783,
      "grad_norm": 0.07845140622971908,
      "learning_rate": 9.801783585030959e-06,
      "loss": 0.7373,
      "step": 56
    },
    {
      "epoch": 0.5493975903614458,
      "grad_norm": 0.07825843281252011,
      "learning_rate": 9.785725615782262e-06,
      "loss": 0.7453,
      "step": 57
    },
    {
      "epoch": 0.5590361445783133,
      "grad_norm": 0.072056346411283,
      "learning_rate": 9.769056488778538e-06,
      "loss": 0.7468,
      "step": 58
    },
    {
      "epoch": 0.5686746987951807,
      "grad_norm": 0.06493580070744535,
      "learning_rate": 9.751778332739033e-06,
      "loss": 0.6937,
      "step": 59
    },
    {
      "epoch": 0.5783132530120482,
      "grad_norm": 0.06883589219210758,
      "learning_rate": 9.733893354158628e-06,
      "loss": 0.724,
      "step": 60
    },
    {
      "epoch": 0.5879518072289157,
      "grad_norm": 0.08127915681080482,
      "learning_rate": 9.715403837026046e-06,
      "loss": 0.7357,
      "step": 61
    },
    {
      "epoch": 0.5975903614457831,
      "grad_norm": 0.07532962426042335,
      "learning_rate": 9.69631214253219e-06,
      "loss": 0.7634,
      "step": 62
    },
    {
      "epoch": 0.6072289156626506,
      "grad_norm": 0.07103096817411828,
      "learning_rate": 9.676620708768608e-06,
      "loss": 0.7507,
      "step": 63
    },
    {
      "epoch": 0.6168674698795181,
      "grad_norm": 0.06730296630661546,
      "learning_rate": 9.656332050416118e-06,
      "loss": 0.7995,
      "step": 64
    },
    {
      "epoch": 0.6265060240963856,
      "grad_norm": 0.07088532082738806,
      "learning_rate": 9.635448758423703e-06,
      "loss": 0.7284,
      "step": 65
    },
    {
      "epoch": 0.636144578313253,
      "grad_norm": 0.07333112024028063,
      "learning_rate": 9.613973499677613e-06,
      "loss": 0.7757,
      "step": 66
    },
    {
      "epoch": 0.6457831325301204,
      "grad_norm": 0.08025501213483908,
      "learning_rate": 9.591909016660806e-06,
      "loss": 0.7109,
      "step": 67
    },
    {
      "epoch": 0.655421686746988,
      "grad_norm": 0.08343912298543543,
      "learning_rate": 9.569258127102708e-06,
      "loss": 0.6956,
      "step": 68
    },
    {
      "epoch": 0.6650602409638554,
      "grad_norm": 0.0702913605748641,
      "learning_rate": 9.546023723619387e-06,
      "loss": 0.6946,
      "step": 69
    },
    {
      "epoch": 0.6746987951807228,
      "grad_norm": 0.07487955447921284,
      "learning_rate": 9.522208773344147e-06,
      "loss": 0.6818,
      "step": 70
    },
    {
      "epoch": 0.6843373493975904,
      "grad_norm": 0.07183727481397417,
      "learning_rate": 9.497816317548625e-06,
      "loss": 0.7371,
      "step": 71
    },
    {
      "epoch": 0.6939759036144578,
      "grad_norm": 0.07423016586226384,
      "learning_rate": 9.472849471254386e-06,
      "loss": 0.7471,
      "step": 72
    },
    {
      "epoch": 0.7036144578313253,
      "grad_norm": 0.07862160976107511,
      "learning_rate": 9.447311422835141e-06,
      "loss": 0.7603,
      "step": 73
    },
    {
      "epoch": 0.7132530120481928,
      "grad_norm": 0.06406902535918113,
      "learning_rate": 9.421205433609568e-06,
      "loss": 0.6906,
      "step": 74
    },
    {
      "epoch": 0.7228915662650602,
      "grad_norm": 0.06529603874190637,
      "learning_rate": 9.39453483742483e-06,
      "loss": 0.688,
      "step": 75
    },
    {
      "epoch": 0.7325301204819277,
      "grad_norm": 0.0715253101992791,
      "learning_rate": 9.367303040230828e-06,
      "loss": 0.6933,
      "step": 76
    },
    {
      "epoch": 0.7421686746987952,
      "grad_norm": 0.07138315068754539,
      "learning_rate": 9.33951351964525e-06,
      "loss": 0.7064,
      "step": 77
    },
    {
      "epoch": 0.7518072289156627,
      "grad_norm": 0.06646488529034193,
      "learning_rate": 9.311169824509454e-06,
      "loss": 0.7174,
      "step": 78
    },
    {
      "epoch": 0.7614457831325301,
      "grad_norm": 0.07024176035919157,
      "learning_rate": 9.28227557443528e-06,
      "loss": 0.7226,
      "step": 79
    },
    {
      "epoch": 0.7710843373493976,
      "grad_norm": 0.062310962213733945,
      "learning_rate": 9.252834459342801e-06,
      "loss": 0.7238,
      "step": 80
    },
    {
      "epoch": 0.7807228915662651,
      "grad_norm": 0.07322177936729748,
      "learning_rate": 9.222850238989104e-06,
      "loss": 0.706,
      "step": 81
    },
    {
      "epoch": 0.7903614457831325,
      "grad_norm": 0.06195276994822025,
      "learning_rate": 9.192326742488153e-06,
      "loss": 0.7107,
      "step": 82
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.07022979646591547,
      "learning_rate": 9.161267867821802e-06,
      "loss": 0.678,
      "step": 83
    },
    {
      "epoch": 0.8096385542168675,
      "grad_norm": 0.0757230181696156,
      "learning_rate": 9.129677581342e-06,
      "loss": 0.7487,
      "step": 84
    },
    {
      "epoch": 0.8192771084337349,
      "grad_norm": 0.06342455930199235,
      "learning_rate": 9.097559917264268e-06,
      "loss": 0.6984,
      "step": 85
    },
    {
      "epoch": 0.8289156626506025,
      "grad_norm": 0.06113100771712842,
      "learning_rate": 9.064918977152517e-06,
      "loss": 0.72,
      "step": 86
    },
    {
      "epoch": 0.8385542168674699,
      "grad_norm": 0.06646354277708373,
      "learning_rate": 9.031758929395259e-06,
      "loss": 0.6858,
      "step": 87
    },
    {
      "epoch": 0.8481927710843373,
      "grad_norm": 0.06414930650101153,
      "learning_rate": 8.998084008673284e-06,
      "loss": 0.7008,
      "step": 88
    },
    {
      "epoch": 0.8578313253012049,
      "grad_norm": 0.06669153411991571,
      "learning_rate": 8.963898515418885e-06,
      "loss": 0.7093,
      "step": 89
    },
    {
      "epoch": 0.8674698795180723,
      "grad_norm": 0.07412062045012623,
      "learning_rate": 8.929206815266653e-06,
      "loss": 0.6888,
      "step": 90
    },
    {
      "epoch": 0.8771084337349397,
      "grad_norm": 0.07202834958462946,
      "learning_rate": 8.89401333849598e-06,
      "loss": 0.6948,
      "step": 91
    },
    {
      "epoch": 0.8867469879518072,
      "grad_norm": 0.06873056807112735,
      "learning_rate": 8.8583225794653e-06,
      "loss": 0.7512,
      "step": 92
    },
    {
      "epoch": 0.8963855421686747,
      "grad_norm": 0.06679929866753545,
      "learning_rate": 8.82213909603812e-06,
      "loss": 0.6964,
      "step": 93
    },
    {
      "epoch": 0.9060240963855422,
      "grad_norm": 0.07323069668965577,
      "learning_rate": 8.78546750900098e-06,
      "loss": 0.7281,
      "step": 94
    },
    {
      "epoch": 0.9156626506024096,
      "grad_norm": 0.07083868471644728,
      "learning_rate": 8.748312501473351e-06,
      "loss": 0.6991,
      "step": 95
    },
    {
      "epoch": 0.9253012048192771,
      "grad_norm": 0.05937570692801768,
      "learning_rate": 8.710678818309576e-06,
      "loss": 0.6241,
      "step": 96
    },
    {
      "epoch": 0.9349397590361446,
      "grad_norm": 0.061050430884171045,
      "learning_rate": 8.672571265492944e-06,
      "loss": 0.6176,
      "step": 97
    },
    {
      "epoch": 0.944578313253012,
      "grad_norm": 0.0658465310603122,
      "learning_rate": 8.63399470952193e-06,
      "loss": 0.672,
      "step": 98
    },
    {
      "epoch": 0.9542168674698795,
      "grad_norm": 0.06755179095814319,
      "learning_rate": 8.594954076788736e-06,
      "loss": 0.7228,
      "step": 99
    },
    {
      "epoch": 0.963855421686747,
      "grad_norm": 0.07262602835248963,
      "learning_rate": 8.555454352950161e-06,
      "loss": 0.7101,
      "step": 100
    },
    {
      "epoch": 0.9734939759036144,
      "grad_norm": 0.07303010678022298,
      "learning_rate": 8.515500582290914e-06,
      "loss": 0.6865,
      "step": 101
    },
    {
      "epoch": 0.983132530120482,
      "grad_norm": 0.08187664550907275,
      "learning_rate": 8.475097867079437e-06,
      "loss": 0.6994,
      "step": 102
    },
    {
      "epoch": 0.9927710843373494,
      "grad_norm": 0.06244683599564814,
      "learning_rate": 8.434251366916323e-06,
      "loss": 0.7047,
      "step": 103
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.06244683599564814,
      "learning_rate": 8.392966298075413e-06,
      "loss": 0.6859,
      "step": 104
    },
    {
      "epoch": 1.0096385542168675,
      "grad_norm": 0.07725262140434404,
      "learning_rate": 8.351247932837655e-06,
      "loss": 0.6524,
      "step": 105
    },
    {
      "epoch": 1.0192771084337349,
      "grad_norm": 0.05782618669079852,
      "learning_rate": 8.309101598817812e-06,
      "loss": 0.6007,
      "step": 106
    },
    {
      "epoch": 1.0289156626506024,
      "grad_norm": 0.05932365316594588,
      "learning_rate": 8.266532678284103e-06,
      "loss": 0.6718,
      "step": 107
    },
    {
      "epoch": 1.03855421686747,
      "grad_norm": 0.06611899088620404,
      "learning_rate": 8.223546607470863e-06,
      "loss": 0.62,
      "step": 108
    },
    {
      "epoch": 1.0481927710843373,
      "grad_norm": 0.07180268403968028,
      "learning_rate": 8.18014887588431e-06,
      "loss": 0.5977,
      "step": 109
    },
    {
      "epoch": 1.0578313253012048,
      "grad_norm": 0.061881962840029656,
      "learning_rate": 8.13634502560152e-06,
      "loss": 0.624,
      "step": 110
    },
    {
      "epoch": 1.0674698795180724,
      "grad_norm": 0.06247008666424044,
      "learning_rate": 8.092140650562665e-06,
      "loss": 0.6297,
      "step": 111
    },
    {
      "epoch": 1.0771084337349397,
      "grad_norm": 0.06286105854890581,
      "learning_rate": 8.047541395856661e-06,
      "loss": 0.628,
      "step": 112
    },
    {
      "epoch": 1.0867469879518072,
      "grad_norm": 0.06730512890818845,
      "learning_rate": 8.002552957000254e-06,
      "loss": 0.6064,
      "step": 113
    },
    {
      "epoch": 1.0963855421686748,
      "grad_norm": 0.06403895444236096,
      "learning_rate": 7.957181079210676e-06,
      "loss": 0.6034,
      "step": 114
    },
    {
      "epoch": 1.106024096385542,
      "grad_norm": 0.05995505967096142,
      "learning_rate": 7.911431556671967e-06,
      "loss": 0.5827,
      "step": 115
    },
    {
      "epoch": 1.1156626506024097,
      "grad_norm": 0.06106916100219134,
      "learning_rate": 7.865310231795026e-06,
      "loss": 0.6235,
      "step": 116
    },
    {
      "epoch": 1.1253012048192772,
      "grad_norm": 0.0727899167728142,
      "learning_rate": 7.818822994471504e-06,
      "loss": 0.6082,
      "step": 117
    },
    {
      "epoch": 1.1349397590361445,
      "grad_norm": 0.06595393108950595,
      "learning_rate": 7.771975781321655e-06,
      "loss": 0.6335,
      "step": 118
    },
    {
      "epoch": 1.144578313253012,
      "grad_norm": 0.07220179641493146,
      "learning_rate": 7.72477457493619e-06,
      "loss": 0.6208,
      "step": 119
    },
    {
      "epoch": 1.1542168674698796,
      "grad_norm": 0.053152296862338196,
      "learning_rate": 7.677225403112277e-06,
      "loss": 0.6314,
      "step": 120
    },
    {
      "epoch": 1.163855421686747,
      "grad_norm": 0.06908539751424997,
      "learning_rate": 7.629334338083774e-06,
      "loss": 0.618,
      "step": 121
    },
    {
      "epoch": 1.1734939759036145,
      "grad_norm": 0.0607666575336561,
      "learning_rate": 7.58110749574577e-06,
      "loss": 0.6127,
      "step": 122
    },
    {
      "epoch": 1.1831325301204818,
      "grad_norm": 0.05725193856979287,
      "learning_rate": 7.532551034873558e-06,
      "loss": 0.6414,
      "step": 123
    },
    {
      "epoch": 1.1927710843373494,
      "grad_norm": 0.06466648989573874,
      "learning_rate": 7.483671156336142e-06,
      "loss": 0.5967,
      "step": 124
    },
    {
      "epoch": 1.202409638554217,
      "grad_norm": 0.0653325240595629,
      "learning_rate": 7.43447410230435e-06,
      "loss": 0.6185,
      "step": 125
    },
    {
      "epoch": 1.2120481927710842,
      "grad_norm": 0.07538141090987788,
      "learning_rate": 7.384966155453686e-06,
      "loss": 0.6412,
      "step": 126
    },
    {
      "epoch": 1.2216867469879518,
      "grad_norm": 0.07040259282115024,
      "learning_rate": 7.335153638162005e-06,
      "loss": 0.6173,
      "step": 127
    },
    {
      "epoch": 1.2313253012048193,
      "grad_norm": 0.060551841344496767,
      "learning_rate": 7.285042911702116e-06,
      "loss": 0.5867,
      "step": 128
    },
    {
      "epoch": 1.2409638554216866,
      "grad_norm": 0.060059786172390106,
      "learning_rate": 7.234640375429427e-06,
      "loss": 0.617,
      "step": 129
    },
    {
      "epoch": 1.2506024096385542,
      "grad_norm": 0.06747743957174769,
      "learning_rate": 7.183952465964711e-06,
      "loss": 0.6193,
      "step": 130
    },
    {
      "epoch": 1.2602409638554217,
      "grad_norm": 0.06828017138762647,
      "learning_rate": 7.132985656372126e-06,
      "loss": 0.6423,
      "step": 131
    },
    {
      "epoch": 1.269879518072289,
      "grad_norm": 0.05392382216193354,
      "learning_rate": 7.0817464553325764e-06,
      "loss": 0.5984,
      "step": 132
    },
    {
      "epoch": 1.2795180722891566,
      "grad_norm": 0.06241321312010483,
      "learning_rate": 7.030241406312528e-06,
      "loss": 0.6179,
      "step": 133
    },
    {
      "epoch": 1.2891566265060241,
      "grad_norm": 0.06613062850109816,
      "learning_rate": 6.978477086728375e-06,
      "loss": 0.6758,
      "step": 134
    },
    {
      "epoch": 1.2987951807228915,
      "grad_norm": 0.06410748091385764,
      "learning_rate": 6.926460107106483e-06,
      "loss": 0.6083,
      "step": 135
    },
    {
      "epoch": 1.308433734939759,
      "grad_norm": 0.0608689353711739,
      "learning_rate": 6.874197110238986e-06,
      "loss": 0.621,
      "step": 136
    },
    {
      "epoch": 1.3180722891566266,
      "grad_norm": 0.06714353872124008,
      "learning_rate": 6.8216947703354815e-06,
      "loss": 0.602,
      "step": 137
    },
    {
      "epoch": 1.3277108433734939,
      "grad_norm": 0.0762090276876796,
      "learning_rate": 6.7689597921707065e-06,
      "loss": 0.6078,
      "step": 138
    },
    {
      "epoch": 1.3373493975903614,
      "grad_norm": 0.06482193036022396,
      "learning_rate": 6.715998910228296e-06,
      "loss": 0.6012,
      "step": 139
    },
    {
      "epoch": 1.346987951807229,
      "grad_norm": 0.06505573004876232,
      "learning_rate": 6.6628188878407806e-06,
      "loss": 0.5436,
      "step": 140
    },
    {
      "epoch": 1.3566265060240963,
      "grad_norm": 0.062063475413644965,
      "learning_rate": 6.609426516325859e-06,
      "loss": 0.6328,
      "step": 141
    },
    {
      "epoch": 1.3662650602409638,
      "grad_norm": 0.05934060841850848,
      "learning_rate": 6.555828614119132e-06,
      "loss": 0.6157,
      "step": 142
    },
    {
      "epoch": 1.3759036144578314,
      "grad_norm": 0.06467965488879493,
      "learning_rate": 6.502032025903356e-06,
      "loss": 0.668,
      "step": 143
    },
    {
      "epoch": 1.3855421686746987,
      "grad_norm": 0.06412287381702947,
      "learning_rate": 6.4480436217343366e-06,
      "loss": 0.6411,
      "step": 144
    },
    {
      "epoch": 1.3951807228915662,
      "grad_norm": 0.06321881583648611,
      "learning_rate": 6.393870296163616e-06,
      "loss": 0.6235,
      "step": 145
    },
    {
      "epoch": 1.4048192771084338,
      "grad_norm": 0.06334369747563459,
      "learning_rate": 6.339518967357985e-06,
      "loss": 0.6281,
      "step": 146
    },
    {
      "epoch": 1.4144578313253011,
      "grad_norm": 0.06504837084952969,
      "learning_rate": 6.284996576216014e-06,
      "loss": 0.5896,
      "step": 147
    },
    {
      "epoch": 1.4240963855421687,
      "grad_norm": 0.05937400854730589,
      "learning_rate": 6.230310085481677e-06,
      "loss": 0.6225,
      "step": 148
    },
    {
      "epoch": 1.4337349397590362,
      "grad_norm": 0.07525329007434042,
      "learning_rate": 6.175466478855161e-06,
      "loss": 0.6683,
      "step": 149
    },
    {
      "epoch": 1.4433734939759035,
      "grad_norm": 0.0692386238529192,
      "learning_rate": 6.1204727601010396e-06,
      "loss": 0.665,
      "step": 150
    },
    {
      "epoch": 1.453012048192771,
      "grad_norm": 0.06422226013162163,
      "learning_rate": 6.065335952153846e-06,
      "loss": 0.6313,
      "step": 151
    },
    {
      "epoch": 1.4626506024096386,
      "grad_norm": 0.06068901796091081,
      "learning_rate": 6.010063096221215e-06,
      "loss": 0.6139,
      "step": 152
    },
    {
      "epoch": 1.472289156626506,
      "grad_norm": 0.06840718236821038,
      "learning_rate": 5.954661250884704e-06,
      "loss": 0.6263,
      "step": 153
    },
    {
      "epoch": 1.4819277108433735,
      "grad_norm": 0.06212232268715233,
      "learning_rate": 5.899137491198364e-06,
      "loss": 0.6635,
      "step": 154
    },
    {
      "epoch": 1.491566265060241,
      "grad_norm": 0.07029083873722003,
      "learning_rate": 5.843498907785236e-06,
      "loss": 0.5506,
      "step": 155
    },
    {
      "epoch": 1.5012048192771084,
      "grad_norm": 0.06346720559167707,
      "learning_rate": 5.78775260593185e-06,
      "loss": 0.6193,
      "step": 156
    },
    {
      "epoch": 1.510843373493976,
      "grad_norm": 0.07558970902676904,
      "learning_rate": 5.731905704680834e-06,
      "loss": 0.6251,
      "step": 157
    },
    {
      "epoch": 1.5204819277108435,
      "grad_norm": 0.05842912970122868,
      "learning_rate": 5.6759653359218e-06,
      "loss": 0.6761,
      "step": 158
    },
    {
      "epoch": 1.5301204819277108,
      "grad_norm": 0.05813865281679669,
      "learning_rate": 5.6199386434805615e-06,
      "loss": 0.5842,
      "step": 159
    },
    {
      "epoch": 1.5397590361445783,
      "grad_norm": 0.05744456895698129,
      "learning_rate": 5.563832782206835e-06,
      "loss": 0.6087,
      "step": 160
    },
    {
      "epoch": 1.5493975903614459,
      "grad_norm": 0.06419357719628659,
      "learning_rate": 5.507654917060541e-06,
      "loss": 0.6059,
      "step": 161
    },
    {
      "epoch": 1.5590361445783132,
      "grad_norm": 0.06081199348431484,
      "learning_rate": 5.451412222196801e-06,
      "loss": 0.5965,
      "step": 162
    },
    {
      "epoch": 1.5686746987951807,
      "grad_norm": 0.06588220582304925,
      "learning_rate": 5.395111880049775e-06,
      "loss": 0.5697,
      "step": 163
    },
    {
      "epoch": 1.5783132530120483,
      "grad_norm": 0.06367402188772205,
      "learning_rate": 5.338761080415425e-06,
      "loss": 0.6026,
      "step": 164
    },
    {
      "epoch": 1.5879518072289156,
      "grad_norm": 0.06057405284771154,
      "learning_rate": 5.28236701953335e-06,
      "loss": 0.5812,
      "step": 165
    },
    {
      "epoch": 1.5975903614457831,
      "grad_norm": 0.053482867334479156,
      "learning_rate": 5.225936899167803e-06,
      "loss": 0.6347,
      "step": 166
    },
    {
      "epoch": 1.6072289156626507,
      "grad_norm": 0.06071788010807735,
      "learning_rate": 5.169477925687981e-06,
      "loss": 0.61,
      "step": 167
    },
    {
      "epoch": 1.616867469879518,
      "grad_norm": 0.04803665860432399,
      "learning_rate": 5.112997309147753e-06,
      "loss": 0.6102,
      "step": 168
    },
    {
      "epoch": 1.6265060240963856,
      "grad_norm": 0.07383190175754212,
      "learning_rate": 5.0565022623649e-06,
      "loss": 0.6634,
      "step": 169
    },
    {
      "epoch": 1.636144578313253,
      "grad_norm": 0.06343533190073457,
      "learning_rate": 5e-06,
      "loss": 0.6221,
      "step": 170
    },
    {
      "epoch": 1.6457831325301204,
      "grad_norm": 0.05067311084373564,
      "learning_rate": 4.943497737635103e-06,
      "loss": 0.576,
      "step": 171
    },
    {
      "epoch": 1.655421686746988,
      "grad_norm": 0.07430397087873777,
      "learning_rate": 4.887002690852249e-06,
      "loss": 0.5677,
      "step": 172
    },
    {
      "epoch": 1.6650602409638555,
      "grad_norm": 0.0573447314824425,
      "learning_rate": 4.830522074312019e-06,
      "loss": 0.6066,
      "step": 173
    },
    {
      "epoch": 1.6746987951807228,
      "grad_norm": 0.0663766323629017,
      "learning_rate": 4.774063100832199e-06,
      "loss": 0.6569,
      "step": 174
    },
    {
      "epoch": 1.6843373493975904,
      "grad_norm": 0.06715405212610676,
      "learning_rate": 4.717632980466652e-06,
      "loss": 0.6171,
      "step": 175
    },
    {
      "epoch": 1.693975903614458,
      "grad_norm": 0.06792888500682888,
      "learning_rate": 4.661238919584578e-06,
      "loss": 0.6257,
      "step": 176
    },
    {
      "epoch": 1.7036144578313253,
      "grad_norm": 0.06352940843145351,
      "learning_rate": 4.6048881199502265e-06,
      "loss": 0.5744,
      "step": 177
    },
    {
      "epoch": 1.7132530120481928,
      "grad_norm": 0.06122504041319759,
      "learning_rate": 4.548587777803198e-06,
      "loss": 0.6479,
      "step": 178
    },
    {
      "epoch": 1.7228915662650603,
      "grad_norm": 0.06714121511192614,
      "learning_rate": 4.49234508293946e-06,
      "loss": 0.5905,
      "step": 179
    },
    {
      "epoch": 1.7325301204819277,
      "grad_norm": 0.06504871151522112,
      "learning_rate": 4.436167217793167e-06,
      "loss": 0.6377,
      "step": 180
    },
    {
      "epoch": 1.7421686746987952,
      "grad_norm": 0.05746737774784524,
      "learning_rate": 4.38006135651944e-06,
      "loss": 0.6343,
      "step": 181
    },
    {
      "epoch": 1.7518072289156628,
      "grad_norm": 0.0742332284296601,
      "learning_rate": 4.3240346640782014e-06,
      "loss": 0.604,
      "step": 182
    },
    {
      "epoch": 1.76144578313253,
      "grad_norm": 0.07199906822994642,
      "learning_rate": 4.268094295319167e-06,
      "loss": 0.5884,
      "step": 183
    },
    {
      "epoch": 1.7710843373493976,
      "grad_norm": 0.06146250524213322,
      "learning_rate": 4.212247394068151e-06,
      "loss": 0.5786,
      "step": 184
    },
    {
      "epoch": 1.7807228915662652,
      "grad_norm": 0.06678830744721113,
      "learning_rate": 4.1565010922147644e-06,
      "loss": 0.568,
      "step": 185
    },
    {
      "epoch": 1.7903614457831325,
      "grad_norm": 0.06225018767217524,
      "learning_rate": 4.100862508801639e-06,
      "loss": 0.6216,
      "step": 186
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.06390858568136344,
      "learning_rate": 4.045338749115299e-06,
      "loss": 0.58,
      "step": 187
    },
    {
      "epoch": 1.8096385542168676,
      "grad_norm": 0.06358790048246866,
      "learning_rate": 3.989936903778785e-06,
      "loss": 0.6225,
      "step": 188
    },
    {
      "epoch": 1.819277108433735,
      "grad_norm": 0.07051504136128764,
      "learning_rate": 3.934664047846157e-06,
      "loss": 0.5685,
      "step": 189
    },
    {
      "epoch": 1.8289156626506025,
      "grad_norm": 0.05730039419508112,
      "learning_rate": 3.879527239898962e-06,
      "loss": 0.5875,
      "step": 190
    },
    {
      "epoch": 1.83855421686747,
      "grad_norm": 0.0578158196725697,
      "learning_rate": 3.8245335211448404e-06,
      "loss": 0.5635,
      "step": 191
    },
    {
      "epoch": 1.8481927710843373,
      "grad_norm": 0.07225016497571735,
      "learning_rate": 3.769689914518326e-06,
      "loss": 0.6155,
      "step": 192
    },
    {
      "epoch": 1.8578313253012049,
      "grad_norm": 0.0549804131963855,
      "learning_rate": 3.715003423783986e-06,
      "loss": 0.624,
      "step": 193
    },
    {
      "epoch": 1.8674698795180724,
      "grad_norm": 0.06298514519756715,
      "learning_rate": 3.660481032642016e-06,
      "loss": 0.6506,
      "step": 194
    },
    {
      "epoch": 1.8771084337349397,
      "grad_norm": 0.0686500519258368,
      "learning_rate": 3.6061297038363853e-06,
      "loss": 0.602,
      "step": 195
    },
    {
      "epoch": 1.886746987951807,
      "grad_norm": 0.06368295604671866,
      "learning_rate": 3.5519563782656642e-06,
      "loss": 0.6721,
      "step": 196
    },
    {
      "epoch": 1.8963855421686748,
      "grad_norm": 0.06096396381363934,
      "learning_rate": 3.497967974096647e-06,
      "loss": 0.6107,
      "step": 197
    },
    {
      "epoch": 1.9060240963855422,
      "grad_norm": 0.06013183394905736,
      "learning_rate": 3.4441713858808684e-06,
      "loss": 0.582,
      "step": 198
    },
    {
      "epoch": 1.9156626506024095,
      "grad_norm": 0.05869263235641963,
      "learning_rate": 3.3905734836741415e-06,
      "loss": 0.5896,
      "step": 199
    },
    {
      "epoch": 1.9253012048192772,
      "grad_norm": 0.06144735421932478,
      "learning_rate": 3.3371811121592203e-06,
      "loss": 0.551,
      "step": 200
    },
    {
      "epoch": 1.9349397590361446,
      "grad_norm": 0.07092447055189571,
      "learning_rate": 3.2840010897717045e-06,
      "loss": 0.603,
      "step": 201
    },
    {
      "epoch": 1.944578313253012,
      "grad_norm": 0.0608882656847333,
      "learning_rate": 3.2310402078292956e-06,
      "loss": 0.6072,
      "step": 202
    },
    {
      "epoch": 1.9542168674698797,
      "grad_norm": 0.06498984091309146,
      "learning_rate": 3.178305229664519e-06,
      "loss": 0.5848,
      "step": 203
    },
    {
      "epoch": 1.963855421686747,
      "grad_norm": 0.06597558210840206,
      "learning_rate": 3.125802889761016e-06,
      "loss": 0.6262,
      "step": 204
    },
    {
      "epoch": 1.9734939759036143,
      "grad_norm": 0.05675086313446132,
      "learning_rate": 3.073539892893519e-06,
      "loss": 0.5991,
      "step": 205
    },
    {
      "epoch": 1.983132530120482,
      "grad_norm": 0.062196722016115806,
      "learning_rate": 3.021522913271627e-06,
      "loss": 0.601,
      "step": 206
    },
    {
      "epoch": 1.9927710843373494,
      "grad_norm": 0.054930108170390506,
      "learning_rate": 2.969758593687475e-06,
      "loss": 0.5751,
      "step": 207
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.08111810625267253,
      "learning_rate": 2.9182535446674244e-06,
      "loss": 0.5888,
      "step": 208
    },
    {
      "epoch": 2.0096385542168673,
      "grad_norm": 0.08388412876294439,
      "learning_rate": 2.8670143436278757e-06,
      "loss": 0.5468,
      "step": 209
    },
    {
      "epoch": 2.019277108433735,
      "grad_norm": 0.05934240328760545,
      "learning_rate": 2.8160475340352913e-06,
      "loss": 0.5644,
      "step": 210
    },
    {
      "epoch": 2.0289156626506024,
      "grad_norm": 0.06787576120463162,
      "learning_rate": 2.765359624570574e-06,
      "loss": 0.5886,
      "step": 211
    },
    {
      "epoch": 2.0385542168674697,
      "grad_norm": 0.0633591218554768,
      "learning_rate": 2.714957088297886e-06,
      "loss": 0.527,
      "step": 212
    },
    {
      "epoch": 2.0481927710843375,
      "grad_norm": 0.06017979720395379,
      "learning_rate": 2.664846361837997e-06,
      "loss": 0.5562,
      "step": 213
    },
    {
      "epoch": 2.057831325301205,
      "grad_norm": 0.0680365289599211,
      "learning_rate": 2.6150338445463146e-06,
      "loss": 0.558,
      "step": 214
    },
    {
      "epoch": 2.067469879518072,
      "grad_norm": 0.05321779366151024,
      "learning_rate": 2.565525897695651e-06,
      "loss": 0.5618,
      "step": 215
    },
    {
      "epoch": 2.07710843373494,
      "grad_norm": 0.06537975532750284,
      "learning_rate": 2.51632884366386e-06,
      "loss": 0.5539,
      "step": 216
    },
    {
      "epoch": 2.0867469879518072,
      "grad_norm": 0.0611607010768512,
      "learning_rate": 2.4674489651264433e-06,
      "loss": 0.5085,
      "step": 217
    },
    {
      "epoch": 2.0963855421686746,
      "grad_norm": 0.05053608558411659,
      "learning_rate": 2.418892504254231e-06,
      "loss": 0.5368,
      "step": 218
    },
    {
      "epoch": 2.1060240963855423,
      "grad_norm": 0.06189373201667931,
      "learning_rate": 2.3706656619162278e-06,
      "loss": 0.5227,
      "step": 219
    },
    {
      "epoch": 2.1156626506024097,
      "grad_norm": 0.061459908604694274,
      "learning_rate": 2.322774596887726e-06,
      "loss": 0.6102,
      "step": 220
    },
    {
      "epoch": 2.125301204819277,
      "grad_norm": 0.07071845948302857,
      "learning_rate": 2.275225425063813e-06,
      "loss": 0.5673,
      "step": 221
    },
    {
      "epoch": 2.1349397590361447,
      "grad_norm": 0.059265035233080764,
      "learning_rate": 2.2280242186783473e-06,
      "loss": 0.5475,
      "step": 222
    },
    {
      "epoch": 2.144578313253012,
      "grad_norm": 0.05927854864357284,
      "learning_rate": 2.1811770055284968e-06,
      "loss": 0.566,
      "step": 223
    },
    {
      "epoch": 2.1542168674698794,
      "grad_norm": 0.05843386609477244,
      "learning_rate": 2.134689768204975e-06,
      "loss": 0.5263,
      "step": 224
    },
    {
      "epoch": 2.163855421686747,
      "grad_norm": 0.054515077941895744,
      "learning_rate": 2.0885684433280336e-06,
      "loss": 0.5706,
      "step": 225
    },
    {
      "epoch": 2.1734939759036145,
      "grad_norm": 0.060254043551932875,
      "learning_rate": 2.042818920789326e-06,
      "loss": 0.4764,
      "step": 226
    },
    {
      "epoch": 2.183132530120482,
      "grad_norm": 0.0570869413051247,
      "learning_rate": 1.9974470429997482e-06,
      "loss": 0.5465,
      "step": 227
    },
    {
      "epoch": 2.1927710843373496,
      "grad_norm": 0.055562628054217,
      "learning_rate": 1.9524586041433393e-06,
      "loss": 0.5472,
      "step": 228
    },
    {
      "epoch": 2.202409638554217,
      "grad_norm": 0.06059513133262611,
      "learning_rate": 1.907859349437336e-06,
      "loss": 0.5484,
      "step": 229
    },
    {
      "epoch": 2.212048192771084,
      "grad_norm": 0.06123920490330747,
      "learning_rate": 1.8636549743984815e-06,
      "loss": 0.5507,
      "step": 230
    },
    {
      "epoch": 2.221686746987952,
      "grad_norm": 0.05837754169240895,
      "learning_rate": 1.8198511241156902e-06,
      "loss": 0.5172,
      "step": 231
    },
    {
      "epoch": 2.2313253012048193,
      "grad_norm": 0.06444730736010255,
      "learning_rate": 1.776453392529139e-06,
      "loss": 0.5473,
      "step": 232
    },
    {
      "epoch": 2.2409638554216866,
      "grad_norm": 0.057036642865459466,
      "learning_rate": 1.7334673217158976e-06,
      "loss": 0.5243,
      "step": 233
    },
    {
      "epoch": 2.2506024096385544,
      "grad_norm": 0.05541742317802861,
      "learning_rate": 1.6908984011821883e-06,
      "loss": 0.5755,
      "step": 234
    },
    {
      "epoch": 2.2602409638554217,
      "grad_norm": 0.059707658549255194,
      "learning_rate": 1.6487520671623469e-06,
      "loss": 0.5463,
      "step": 235
    },
    {
      "epoch": 2.269879518072289,
      "grad_norm": 0.0652488770688145,
      "learning_rate": 1.6070337019245896e-06,
      "loss": 0.5434,
      "step": 236
    },
    {
      "epoch": 2.279518072289157,
      "grad_norm": 0.062427610958618734,
      "learning_rate": 1.5657486330836786e-06,
      "loss": 0.5404,
      "step": 237
    },
    {
      "epoch": 2.289156626506024,
      "grad_norm": 0.066481990615362,
      "learning_rate": 1.5249021329205638e-06,
      "loss": 0.5886,
      "step": 238
    },
    {
      "epoch": 2.2987951807228915,
      "grad_norm": 0.06075798833103501,
      "learning_rate": 1.4844994177090871e-06,
      "loss": 0.5906,
      "step": 239
    },
    {
      "epoch": 2.3084337349397592,
      "grad_norm": 0.060457408593263186,
      "learning_rate": 1.4445456470498392e-06,
      "loss": 0.5201,
      "step": 240
    },
    {
      "epoch": 2.3180722891566266,
      "grad_norm": 0.06189290727406132,
      "learning_rate": 1.4050459232112652e-06,
      "loss": 0.5486,
      "step": 241
    },
    {
      "epoch": 2.327710843373494,
      "grad_norm": 0.05446525602984187,
      "learning_rate": 1.3660052904780707e-06,
      "loss": 0.5667,
      "step": 242
    },
    {
      "epoch": 2.337349397590361,
      "grad_norm": 0.05680759175644314,
      "learning_rate": 1.3274287345070564e-06,
      "loss": 0.4991,
      "step": 243
    },
    {
      "epoch": 2.346987951807229,
      "grad_norm": 0.07369531885761765,
      "learning_rate": 1.2893211816904243e-06,
      "loss": 0.5635,
      "step": 244
    },
    {
      "epoch": 2.3566265060240963,
      "grad_norm": 0.047802386111405644,
      "learning_rate": 1.2516874985266508e-06,
      "loss": 0.5458,
      "step": 245
    },
    {
      "epoch": 2.3662650602409636,
      "grad_norm": 0.056195785484295456,
      "learning_rate": 1.2145324909990202e-06,
      "loss": 0.5458,
      "step": 246
    },
    {
      "epoch": 2.3759036144578314,
      "grad_norm": 0.05563528805443362,
      "learning_rate": 1.1778609039618804e-06,
      "loss": 0.4986,
      "step": 247
    },
    {
      "epoch": 2.3855421686746987,
      "grad_norm": 0.06250095823475356,
      "learning_rate": 1.1416774205347015e-06,
      "loss": 0.5358,
      "step": 248
    },
    {
      "epoch": 2.395180722891566,
      "grad_norm": 0.05734992662747759,
      "learning_rate": 1.1059866615040205e-06,
      "loss": 0.5315,
      "step": 249
    },
    {
      "epoch": 2.404819277108434,
      "grad_norm": 0.06007313643233539,
      "learning_rate": 1.0707931847333487e-06,
      "loss": 0.5265,
      "step": 250
    },
    {
      "epoch": 2.414457831325301,
      "grad_norm": 0.055511713433456615,
      "learning_rate": 1.036101484581117e-06,
      "loss": 0.5207,
      "step": 251
    },
    {
      "epoch": 2.4240963855421684,
      "grad_norm": 0.06459562675579453,
      "learning_rate": 1.0019159913267156e-06,
      "loss": 0.5467,
      "step": 252
    },
    {
      "epoch": 2.433734939759036,
      "grad_norm": 0.0616817710533823,
      "learning_rate": 9.68241070604743e-07,
      "loss": 0.5487,
      "step": 253
    },
    {
      "epoch": 2.4433734939759035,
      "grad_norm": 0.05982170620400226,
      "learning_rate": 9.350810228474855e-07,
      "loss": 0.5894,
      "step": 254
    },
    {
      "epoch": 2.453012048192771,
      "grad_norm": 0.06404688237048371,
      "learning_rate": 9.024400827357344e-07,
      "loss": 0.5273,
      "step": 255
    },
    {
      "epoch": 2.4626506024096386,
      "grad_norm": 0.055361287946098334,
      "learning_rate": 8.703224186580012e-07,
      "loss": 0.5576,
      "step": 256
    },
    {
      "epoch": 2.472289156626506,
      "grad_norm": 0.0563022702076071,
      "learning_rate": 8.387321321781977e-07,
      "loss": 0.5291,
      "step": 257
    },
    {
      "epoch": 2.4819277108433733,
      "grad_norm": 0.06713390666603862,
      "learning_rate": 8.07673257511849e-07,
      "loss": 0.5465,
      "step": 258
    },
    {
      "epoch": 2.491566265060241,
      "grad_norm": 0.049523569414613715,
      "learning_rate": 7.771497610108981e-07,
      "loss": 0.5626,
      "step": 259
    },
    {
      "epoch": 2.5012048192771084,
      "grad_norm": 0.06250857829748814,
      "learning_rate": 7.471655406572003e-07,
      "loss": 0.5076,
      "step": 260
    },
    {
      "epoch": 2.5108433734939757,
      "grad_norm": 0.05831361066226844,
      "learning_rate": 7.177244255647209e-07,
      "loss": 0.5583,
      "step": 261
    },
    {
      "epoch": 2.5204819277108435,
      "grad_norm": 0.06401512676119323,
      "learning_rate": 6.888301754905469e-07,
      "loss": 0.5406,
      "step": 262
    },
    {
      "epoch": 2.5301204819277108,
      "grad_norm": 0.05000225920367488,
      "learning_rate": 6.604864803547511e-07,
      "loss": 0.5643,
      "step": 263
    },
    {
      "epoch": 2.539759036144578,
      "grad_norm": 0.06148033172818441,
      "learning_rate": 6.326969597691724e-07,
      "loss": 0.5538,
      "step": 264
    },
    {
      "epoch": 2.549397590361446,
      "grad_norm": 0.06706799429689687,
      "learning_rate": 6.054651625751717e-07,
      "loss": 0.5409,
      "step": 265
    },
    {
      "epoch": 2.559036144578313,
      "grad_norm": 0.060888258525890235,
      "learning_rate": 5.787945663904332e-07,
      "loss": 0.5391,
      "step": 266
    },
    {
      "epoch": 2.5686746987951805,
      "grad_norm": 0.06253452716963322,
      "learning_rate": 5.526885771648599e-07,
      "loss": 0.545,
      "step": 267
    },
    {
      "epoch": 2.5783132530120483,
      "grad_norm": 0.06343440419288733,
      "learning_rate": 5.271505287456153e-07,
      "loss": 0.5832,
      "step": 268
    },
    {
      "epoch": 2.5879518072289156,
      "grad_norm": 0.05543612918268993,
      "learning_rate": 5.021836824513759e-07,
      "loss": 0.5834,
      "step": 269
    },
    {
      "epoch": 2.597590361445783,
      "grad_norm": 0.058093169583808696,
      "learning_rate": 4.777912266558532e-07,
      "loss": 0.5272,
      "step": 270
    },
    {
      "epoch": 2.6072289156626507,
      "grad_norm": 0.057594604691082484,
      "learning_rate": 4.5397627638061604e-07,
      "loss": 0.5327,
      "step": 271
    },
    {
      "epoch": 2.616867469879518,
      "grad_norm": 0.05690543891633744,
      "learning_rate": 4.307418728972934e-07,
      "loss": 0.5503,
      "step": 272
    },
    {
      "epoch": 2.6265060240963853,
      "grad_norm": 0.06235909893309931,
      "learning_rate": 4.080909833391944e-07,
      "loss": 0.529,
      "step": 273
    },
    {
      "epoch": 2.636144578313253,
      "grad_norm": 0.05905101458291052,
      "learning_rate": 3.8602650032238675e-07,
      "loss": 0.5298,
      "step": 274
    },
    {
      "epoch": 2.6457831325301204,
      "grad_norm": 0.06563516042706921,
      "learning_rate": 3.6455124157629805e-07,
      "loss": 0.5732,
      "step": 275
    },
    {
      "epoch": 2.6554216867469878,
      "grad_norm": 0.056133471314432144,
      "learning_rate": 3.436679495838835e-07,
      "loss": 0.5369,
      "step": 276
    },
    {
      "epoch": 2.6650602409638555,
      "grad_norm": 0.07043862639489791,
      "learning_rate": 3.233792912313943e-07,
      "loss": 0.5324,
      "step": 277
    },
    {
      "epoch": 2.674698795180723,
      "grad_norm": 0.05655745002115051,
      "learning_rate": 3.0368785746780925e-07,
      "loss": 0.5449,
      "step": 278
    },
    {
      "epoch": 2.68433734939759,
      "grad_norm": 0.06000478684007564,
      "learning_rate": 2.8459616297395464e-07,
      "loss": 0.5378,
      "step": 279
    },
    {
      "epoch": 2.693975903614458,
      "grad_norm": 0.06143997342419319,
      "learning_rate": 2.6610664584137413e-07,
      "loss": 0.5496,
      "step": 280
    },
    {
      "epoch": 2.7036144578313253,
      "grad_norm": 0.0665566566789724,
      "learning_rate": 2.482216672609677e-07,
      "loss": 0.5115,
      "step": 281
    },
    {
      "epoch": 2.7132530120481926,
      "grad_norm": 0.06198933331334414,
      "learning_rate": 2.3094351122146307e-07,
      "loss": 0.5563,
      "step": 282
    },
    {
      "epoch": 2.7228915662650603,
      "grad_norm": 0.06294960442358041,
      "learning_rate": 2.142743842177386e-07,
      "loss": 0.5321,
      "step": 283
    },
    {
      "epoch": 2.7325301204819277,
      "grad_norm": 0.059684494964788816,
      "learning_rate": 1.98216414969043e-07,
      "loss": 0.49,
      "step": 284
    },
    {
      "epoch": 2.742168674698795,
      "grad_norm": 0.05972378233034468,
      "learning_rate": 1.8277165414714858e-07,
      "loss": 0.5189,
      "step": 285
    },
    {
      "epoch": 2.7518072289156628,
      "grad_norm": 0.06503508352439365,
      "learning_rate": 1.6794207411447548e-07,
      "loss": 0.5701,
      "step": 286
    },
    {
      "epoch": 2.76144578313253,
      "grad_norm": 0.05713319020861861,
      "learning_rate": 1.5372956867220678e-07,
      "loss": 0.576,
      "step": 287
    },
    {
      "epoch": 2.7710843373493974,
      "grad_norm": 0.06601360939759876,
      "learning_rate": 1.4013595281844872e-07,
      "loss": 0.5726,
      "step": 288
    },
    {
      "epoch": 2.780722891566265,
      "grad_norm": 0.052203739381335726,
      "learning_rate": 1.2716296251644e-07,
      "loss": 0.5328,
      "step": 289
    },
    {
      "epoch": 2.7903614457831325,
      "grad_norm": 0.06980077697889618,
      "learning_rate": 1.1481225447286803e-07,
      "loss": 0.5655,
      "step": 290
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.055551268853618295,
      "learning_rate": 1.0308540592629756e-07,
      "loss": 0.5792,
      "step": 291
    },
    {
      "epoch": 2.8096385542168676,
      "grad_norm": 0.06646819340479895,
      "learning_rate": 9.198391444575072e-08,
      "loss": 0.548,
      "step": 292
    },
    {
      "epoch": 2.819277108433735,
      "grad_norm": 0.06410763337476913,
      "learning_rate": 8.150919773946165e-08,
      "loss": 0.5871,
      "step": 293
    },
    {
      "epoch": 2.8289156626506022,
      "grad_norm": 0.05712141832748347,
      "learning_rate": 7.166259347382854e-08,
      "loss": 0.5526,
      "step": 294
    },
    {
      "epoch": 2.83855421686747,
      "grad_norm": 0.06542740373895323,
      "learning_rate": 6.244535910258697e-08,
      "loss": 0.5502,
      "step": 295
    },
    {
      "epoch": 2.8481927710843373,
      "grad_norm": 0.053860816649099204,
      "learning_rate": 5.3858671706230605e-08,
      "loss": 0.5654,
      "step": 296
    },
    {
      "epoch": 2.8578313253012047,
      "grad_norm": 0.057862897174585454,
      "learning_rate": 4.590362784169022e-08,
      "loss": 0.5248,
      "step": 297
    },
    {
      "epoch": 2.8674698795180724,
      "grad_norm": 0.04998420980952799,
      "learning_rate": 3.858124340229863e-08,
      "loss": 0.5558,
      "step": 298
    },
    {
      "epoch": 2.8771084337349397,
      "grad_norm": 0.07024515673340713,
      "learning_rate": 3.1892453488058803e-08,
      "loss": 0.5335,
      "step": 299
    },
    {
      "epoch": 2.886746987951807,
      "grad_norm": 0.05961311900445455,
      "learning_rate": 2.5838112286226123e-08,
      "loss": 0.5456,
      "step": 300
    },
    {
      "epoch": 2.896385542168675,
      "grad_norm": 0.056751540967121886,
      "learning_rate": 2.0418992962224495e-08,
      "loss": 0.5273,
      "step": 301
    },
    {
      "epoch": 2.906024096385542,
      "grad_norm": 0.05998655290019423,
      "learning_rate": 1.563578756091144e-08,
      "loss": 0.542,
      "step": 302
    },
    {
      "epoch": 2.9156626506024095,
      "grad_norm": 0.05626179494647873,
      "learning_rate": 1.1489106918200487e-08,
      "loss": 0.5507,
      "step": 303
    },
    {
      "epoch": 2.9253012048192772,
      "grad_norm": 0.0588808013463836,
      "learning_rate": 7.979480583052423e-09,
      "loss": 0.5572,
      "step": 304
    },
    {
      "epoch": 2.9349397590361446,
      "grad_norm": 0.06020531165656935,
      "learning_rate": 5.107356749853298e-09,
      "loss": 0.5345,
      "step": 305
    },
    {
      "epoch": 2.944578313253012,
      "grad_norm": 0.06194492848591381,
      "learning_rate": 2.8731022011757593e-09,
      "loss": 0.5595,
      "step": 306
    },
    {
      "epoch": 2.9542168674698797,
      "grad_norm": 0.06174931334891141,
      "learning_rate": 1.2770022609409628e-09,
      "loss": 0.5209,
      "step": 307
    },
    {
      "epoch": 2.963855421686747,
      "grad_norm": 0.0627769874173011,
      "learning_rate": 3.1926075797827914e-10,
      "loss": 0.5404,
      "step": 308
    },
    {
      "epoch": 2.9734939759036143,
      "grad_norm": 0.056693084454889416,
      "learning_rate": 0.0,
      "loss": 0.5376,
      "step": 309
    },
    {
      "epoch": 2.9734939759036143,
      "step": 309,
      "total_flos": 1.0191470808845517e+17,
      "train_loss": 0.6618778399857889,
      "train_runtime": 3030.6803,
      "train_samples_per_second": 4.928,
      "train_steps_per_second": 0.102
    }
  ],
  "logging_steps": 1,
  "max_steps": 309,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0191470808845517e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}