{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.17178441056474125,
  "eval_steps": 500,
  "global_step": 300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005726147018824708,
      "grad_norm": 7.761023998260498,
      "learning_rate": 0.0,
      "loss": 6.0592,
      "step": 1
    },
    {
      "epoch": 0.0011452294037649416,
      "grad_norm": 7.8541951179504395,
      "learning_rate": 5.714285714285715e-07,
      "loss": 6.0156,
      "step": 2
    },
    {
      "epoch": 0.0022904588075298832,
      "grad_norm": 7.347611904144287,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 6.0103,
      "step": 4
    },
    {
      "epoch": 0.003435688211294825,
      "grad_norm": 5.382428169250488,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 5.9221,
      "step": 6
    },
    {
      "epoch": 0.0045809176150597665,
      "grad_norm": 5.063406467437744,
      "learning_rate": 4.000000000000001e-06,
      "loss": 6.0365,
      "step": 8
    },
    {
      "epoch": 0.005726147018824708,
      "grad_norm": 9.779157638549805,
      "learning_rate": 5.142857142857143e-06,
      "loss": 6.0336,
      "step": 10
    },
    {
      "epoch": 0.00687137642258965,
      "grad_norm": 7.555446147918701,
      "learning_rate": 6.285714285714287e-06,
      "loss": 6.0328,
      "step": 12
    },
    {
      "epoch": 0.008016605826354592,
      "grad_norm": 6.790043354034424,
      "learning_rate": 7.428571428571429e-06,
      "loss": 5.7848,
      "step": 14
    },
    {
      "epoch": 0.009161835230119533,
      "grad_norm": 4.4132208824157715,
      "learning_rate": 8.571428571428573e-06,
      "loss": 5.8207,
      "step": 16
    },
    {
      "epoch": 0.010307064633884476,
      "grad_norm": 4.064995765686035,
      "learning_rate": 9.714285714285715e-06,
      "loss": 5.6497,
      "step": 18
    },
    {
      "epoch": 0.011452294037649417,
      "grad_norm": 3.357184410095215,
      "learning_rate": 1.0857142857142858e-05,
      "loss": 5.7758,
      "step": 20
    },
    {
      "epoch": 0.012597523441414358,
      "grad_norm": 2.742230176925659,
      "learning_rate": 1.2e-05,
      "loss": 5.6173,
      "step": 22
    },
    {
      "epoch": 0.0137427528451793,
      "grad_norm": 2.491459369659424,
      "learning_rate": 1.3142857142857143e-05,
      "loss": 5.6681,
      "step": 24
    },
    {
      "epoch": 0.014887982248944241,
      "grad_norm": 2.7569029331207275,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 5.6393,
      "step": 26
    },
    {
      "epoch": 0.016033211652709184,
      "grad_norm": 2.208378791809082,
      "learning_rate": 1.5428571428571428e-05,
      "loss": 5.5768,
      "step": 28
    },
    {
      "epoch": 0.017178441056474127,
      "grad_norm": 3.2770133018493652,
      "learning_rate": 1.657142857142857e-05,
      "loss": 5.484,
      "step": 30
    },
    {
      "epoch": 0.018323670460239066,
      "grad_norm": 3.177299976348877,
      "learning_rate": 1.7714285714285713e-05,
      "loss": 5.528,
      "step": 32
    },
    {
      "epoch": 0.01946889986400401,
      "grad_norm": 2.1981537342071533,
      "learning_rate": 1.885714285714286e-05,
      "loss": 5.6327,
      "step": 34
    },
    {
      "epoch": 0.02061412926776895,
      "grad_norm": 3.265881061553955,
      "learning_rate": 2e-05,
      "loss": 5.6288,
      "step": 36
    },
    {
      "epoch": 0.02175935867153389,
      "grad_norm": 3.6059298515319824,
      "learning_rate": 2.1142857142857144e-05,
      "loss": 5.4789,
      "step": 38
    },
    {
      "epoch": 0.022904588075298833,
      "grad_norm": 2.4080026149749756,
      "learning_rate": 2.2285714285714287e-05,
      "loss": 5.4046,
      "step": 40
    },
    {
      "epoch": 0.024049817479063776,
      "grad_norm": 2.142902135848999,
      "learning_rate": 2.342857142857143e-05,
      "loss": 5.4738,
      "step": 42
    },
    {
      "epoch": 0.025195046882828715,
      "grad_norm": 2.4021224975585938,
      "learning_rate": 2.4571428571428572e-05,
      "loss": 5.4649,
      "step": 44
    },
    {
      "epoch": 0.026340276286593658,
      "grad_norm": 2.172009229660034,
      "learning_rate": 2.5714285714285714e-05,
      "loss": 5.4302,
      "step": 46
    },
    {
      "epoch": 0.0274855056903586,
      "grad_norm": 2.9737730026245117,
      "learning_rate": 2.6857142857142857e-05,
      "loss": 5.3045,
      "step": 48
    },
    {
      "epoch": 0.028630735094123543,
      "grad_norm": 3.0378615856170654,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 5.2185,
      "step": 50
    },
    {
      "epoch": 0.029775964497888482,
      "grad_norm": 3.4448676109313965,
      "learning_rate": 2.9142857142857146e-05,
      "loss": 5.1838,
      "step": 52
    },
    {
      "epoch": 0.030921193901653425,
      "grad_norm": 2.469245672225952,
      "learning_rate": 3.0285714285714288e-05,
      "loss": 5.1637,
      "step": 54
    },
    {
      "epoch": 0.03206642330541837,
      "grad_norm": 3.58486008644104,
      "learning_rate": 3.142857142857143e-05,
      "loss": 5.2063,
      "step": 56
    },
    {
      "epoch": 0.03321165270918331,
      "grad_norm": 3.0815446376800537,
      "learning_rate": 3.257142857142857e-05,
      "loss": 5.2317,
      "step": 58
    },
    {
      "epoch": 0.03435688211294825,
      "grad_norm": 3.6842119693756104,
      "learning_rate": 3.3714285714285716e-05,
      "loss": 5.2695,
      "step": 60
    },
    {
      "epoch": 0.03550211151671319,
      "grad_norm": 2.9440791606903076,
      "learning_rate": 3.485714285714286e-05,
      "loss": 5.2686,
      "step": 62
    },
    {
      "epoch": 0.03664734092047813,
      "grad_norm": 3.9632568359375,
      "learning_rate": 3.6e-05,
      "loss": 5.1262,
      "step": 64
    },
    {
      "epoch": 0.037792570324243074,
      "grad_norm": 4.045065402984619,
      "learning_rate": 3.7142857142857143e-05,
      "loss": 5.1546,
      "step": 66
    },
    {
      "epoch": 0.03893779972800802,
      "grad_norm": 3.5707085132598877,
      "learning_rate": 3.8285714285714286e-05,
      "loss": 5.0036,
      "step": 68
    },
    {
      "epoch": 0.04008302913177296,
      "grad_norm": 3.014404535293579,
      "learning_rate": 3.942857142857143e-05,
      "loss": 5.026,
      "step": 70
    },
    {
      "epoch": 0.0412282585355379,
      "grad_norm": 2.708796977996826,
      "learning_rate": 4.057142857142857e-05,
      "loss": 4.9442,
      "step": 72
    },
    {
      "epoch": 0.04237348793930284,
      "grad_norm": 2.5384011268615723,
      "learning_rate": 4.1714285714285714e-05,
      "loss": 5.0223,
      "step": 74
    },
    {
      "epoch": 0.04351871734306778,
      "grad_norm": 3.006281852722168,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 4.9827,
      "step": 76
    },
    {
      "epoch": 0.044663946746832724,
      "grad_norm": 2.5772130489349365,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 4.9675,
      "step": 78
    },
    {
      "epoch": 0.045809176150597666,
      "grad_norm": 3.456017255783081,
      "learning_rate": 4.514285714285714e-05,
      "loss": 5.0341,
      "step": 80
    },
    {
      "epoch": 0.04695440555436261,
      "grad_norm": 3.3163113594055176,
      "learning_rate": 4.628571428571429e-05,
      "loss": 4.9867,
      "step": 82
    },
    {
      "epoch": 0.04809963495812755,
      "grad_norm": 3.7568469047546387,
      "learning_rate": 4.742857142857143e-05,
      "loss": 4.8652,
      "step": 84
    },
    {
      "epoch": 0.049244864361892494,
      "grad_norm": 4.19318151473999,
      "learning_rate": 4.8571428571428576e-05,
      "loss": 5.0602,
      "step": 86
    },
    {
      "epoch": 0.05039009376565743,
      "grad_norm": 5.1034064292907715,
      "learning_rate": 4.971428571428572e-05,
      "loss": 4.9757,
      "step": 88
    },
    {
      "epoch": 0.05153532316942237,
      "grad_norm": 4.0827484130859375,
      "learning_rate": 5.085714285714286e-05,
      "loss": 4.8486,
      "step": 90
    },
    {
      "epoch": 0.052680552573187316,
      "grad_norm": 4.6189446449279785,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 4.9595,
      "step": 92
    },
    {
      "epoch": 0.05382578197695226,
      "grad_norm": 3.988513469696045,
      "learning_rate": 5.314285714285715e-05,
      "loss": 4.9035,
      "step": 94
    },
    {
      "epoch": 0.0549710113807172,
      "grad_norm": 3.857276678085327,
      "learning_rate": 5.428571428571428e-05,
      "loss": 4.8277,
      "step": 96
    },
    {
      "epoch": 0.056116240784482144,
      "grad_norm": 3.5372354984283447,
      "learning_rate": 5.542857142857143e-05,
      "loss": 4.7718,
      "step": 98
    },
    {
      "epoch": 0.057261470188247086,
      "grad_norm": 3.3853676319122314,
      "learning_rate": 5.6571428571428574e-05,
      "loss": 4.8098,
      "step": 100
    },
    {
      "epoch": 0.05840669959201202,
      "grad_norm": 2.1142077445983887,
      "learning_rate": 5.771428571428572e-05,
      "loss": 4.7975,
      "step": 102
    },
    {
      "epoch": 0.059551928995776965,
      "grad_norm": 3.2275538444519043,
      "learning_rate": 5.885714285714285e-05,
      "loss": 4.8509,
      "step": 104
    },
    {
      "epoch": 0.06069715839954191,
      "grad_norm": 3.5413126945495605,
      "learning_rate": 6e-05,
      "loss": 4.6069,
      "step": 106
    },
    {
      "epoch": 0.06184238780330685,
      "grad_norm": 2.755648374557495,
      "learning_rate": 6.114285714285714e-05,
      "loss": 4.6951,
      "step": 108
    },
    {
      "epoch": 0.06298761720707179,
      "grad_norm": 2.980039596557617,
      "learning_rate": 6.22857142857143e-05,
      "loss": 4.7012,
      "step": 110
    },
    {
      "epoch": 0.06413284661083674,
      "grad_norm": 4.890020370483398,
      "learning_rate": 6.342857142857143e-05,
      "loss": 4.8008,
      "step": 112
    },
    {
      "epoch": 0.06527807601460167,
      "grad_norm": 4.35846471786499,
      "learning_rate": 6.457142857142856e-05,
      "loss": 4.8587,
      "step": 114
    },
    {
      "epoch": 0.06642330541836662,
      "grad_norm": 3.6171813011169434,
      "learning_rate": 6.571428571428571e-05,
      "loss": 4.7473,
      "step": 116
    },
    {
      "epoch": 0.06756853482213156,
      "grad_norm": 2.4927010536193848,
      "learning_rate": 6.685714285714286e-05,
      "loss": 4.7113,
      "step": 118
    },
    {
      "epoch": 0.0687137642258965,
      "grad_norm": 3.3327009677886963,
      "learning_rate": 6.800000000000001e-05,
      "loss": 4.6105,
      "step": 120
    },
    {
      "epoch": 0.06985899362966144,
      "grad_norm": 3.1123206615448,
      "learning_rate": 6.914285714285715e-05,
      "loss": 4.5968,
      "step": 122
    },
    {
      "epoch": 0.07100422303342638,
      "grad_norm": 2.6985421180725098,
      "learning_rate": 7.028571428571428e-05,
      "loss": 4.6323,
      "step": 124
    },
    {
      "epoch": 0.07214945243719133,
      "grad_norm": 2.058084011077881,
      "learning_rate": 7.142857142857143e-05,
      "loss": 4.5721,
      "step": 126
    },
    {
      "epoch": 0.07329468184095626,
      "grad_norm": 2.144658327102661,
      "learning_rate": 7.257142857142858e-05,
      "loss": 4.6125,
      "step": 128
    },
    {
      "epoch": 0.07443991124472121,
      "grad_norm": 2.477219820022583,
      "learning_rate": 7.371428571428572e-05,
      "loss": 4.4727,
      "step": 130
    },
    {
      "epoch": 0.07558514064848615,
      "grad_norm": 3.8517298698425293,
      "learning_rate": 7.485714285714285e-05,
      "loss": 4.5696,
      "step": 132
    },
    {
      "epoch": 0.0767303700522511,
      "grad_norm": 3.0253565311431885,
      "learning_rate": 7.6e-05,
      "loss": 4.4838,
      "step": 134
    },
    {
      "epoch": 0.07787559945601603,
      "grad_norm": 3.397569179534912,
      "learning_rate": 7.714285714285715e-05,
      "loss": 4.6431,
      "step": 136
    },
    {
      "epoch": 0.07902082885978097,
      "grad_norm": 2.435197114944458,
      "learning_rate": 7.828571428571429e-05,
      "loss": 4.4681,
      "step": 138
    },
    {
      "epoch": 0.08016605826354592,
      "grad_norm": 2.6476476192474365,
      "learning_rate": 7.942857142857143e-05,
      "loss": 4.4462,
      "step": 140
    },
    {
      "epoch": 0.08131128766731086,
      "grad_norm": 2.1929690837860107,
      "learning_rate": 8.057142857142857e-05,
      "loss": 4.5136,
      "step": 142
    },
    {
      "epoch": 0.0824565170710758,
      "grad_norm": 2.4533395767211914,
      "learning_rate": 8.171428571428572e-05,
      "loss": 4.5572,
      "step": 144
    },
    {
      "epoch": 0.08360174647484074,
      "grad_norm": 2.601806879043579,
      "learning_rate": 8.285714285714287e-05,
      "loss": 4.4121,
      "step": 146
    },
    {
      "epoch": 0.08474697587860568,
      "grad_norm": 3.233973741531372,
      "learning_rate": 8.4e-05,
      "loss": 4.4599,
      "step": 148
    },
    {
      "epoch": 0.08589220528237063,
      "grad_norm": 2.6353538036346436,
      "learning_rate": 8.514285714285714e-05,
      "loss": 4.4533,
      "step": 150
    },
    {
      "epoch": 0.08703743468613556,
      "grad_norm": 2.8465511798858643,
      "learning_rate": 8.62857142857143e-05,
      "loss": 4.5246,
      "step": 152
    },
    {
      "epoch": 0.08818266408990051,
      "grad_norm": 2.8642711639404297,
      "learning_rate": 8.742857142857144e-05,
      "loss": 4.4659,
      "step": 154
    },
    {
      "epoch": 0.08932789349366545,
      "grad_norm": 2.793112277984619,
      "learning_rate": 8.857142857142857e-05,
      "loss": 4.5107,
      "step": 156
    },
    {
      "epoch": 0.0904731228974304,
      "grad_norm": 3.43472957611084,
      "learning_rate": 8.971428571428571e-05,
      "loss": 4.4079,
      "step": 158
    },
    {
      "epoch": 0.09161835230119533,
      "grad_norm": 2.9260294437408447,
      "learning_rate": 9.085714285714286e-05,
      "loss": 4.4047,
      "step": 160
    },
    {
      "epoch": 0.09276358170496027,
      "grad_norm": 2.6336724758148193,
      "learning_rate": 9.200000000000001e-05,
      "loss": 4.4777,
      "step": 162
    },
    {
      "epoch": 0.09390881110872522,
      "grad_norm": 2.8348231315612793,
      "learning_rate": 9.314285714285715e-05,
      "loss": 4.3445,
      "step": 164
    },
    {
      "epoch": 0.09505404051249015,
      "grad_norm": 4.271595478057861,
      "learning_rate": 9.428571428571429e-05,
      "loss": 4.4234,
      "step": 166
    },
    {
      "epoch": 0.0961992699162551,
      "grad_norm": 3.4789109230041504,
      "learning_rate": 9.542857142857143e-05,
      "loss": 4.2872,
      "step": 168
    },
    {
      "epoch": 0.09734449932002004,
      "grad_norm": 2.57273530960083,
      "learning_rate": 9.657142857142858e-05,
      "loss": 4.4177,
      "step": 170
    },
    {
      "epoch": 0.09848972872378499,
      "grad_norm": 2.185086250305176,
      "learning_rate": 9.771428571428572e-05,
      "loss": 4.3568,
      "step": 172
    },
    {
      "epoch": 0.09963495812754992,
      "grad_norm": 2.771744966506958,
      "learning_rate": 9.885714285714286e-05,
      "loss": 4.3392,
      "step": 174
    },
    {
      "epoch": 0.10078018753131486,
      "grad_norm": 1.950353741645813,
      "learning_rate": 0.0001,
      "loss": 4.1931,
      "step": 176
    },
    {
      "epoch": 0.10192541693507981,
      "grad_norm": 2.4709694385528564,
      "learning_rate": 9.999991040472416e-05,
      "loss": 4.2936,
      "step": 178
    },
    {
      "epoch": 0.10307064633884475,
      "grad_norm": 2.140997886657715,
      "learning_rate": 9.999964161921776e-05,
      "loss": 4.1653,
      "step": 180
    },
    {
      "epoch": 0.1042158757426097,
      "grad_norm": 2.491321563720703,
      "learning_rate": 9.999919364444403e-05,
      "loss": 4.3202,
      "step": 182
    },
    {
      "epoch": 0.10536110514637463,
      "grad_norm": 2.5410189628601074,
      "learning_rate": 9.999856648200845e-05,
      "loss": 4.2657,
      "step": 184
    },
    {
      "epoch": 0.10650633455013958,
      "grad_norm": 2.1820590496063232,
      "learning_rate": 9.999776013415866e-05,
      "loss": 4.2282,
      "step": 186
    },
    {
      "epoch": 0.10765156395390452,
      "grad_norm": 1.7251808643341064,
      "learning_rate": 9.999677460378444e-05,
      "loss": 4.3421,
      "step": 188
    },
    {
      "epoch": 0.10879679335766945,
      "grad_norm": 2.002145290374756,
      "learning_rate": 9.999560989441779e-05,
      "loss": 4.1361,
      "step": 190
    },
    {
      "epoch": 0.1099420227614344,
      "grad_norm": 1.9663431644439697,
      "learning_rate": 9.999426601023274e-05,
      "loss": 4.201,
      "step": 192
    },
    {
      "epoch": 0.11108725216519934,
      "grad_norm": 2.1406776905059814,
      "learning_rate": 9.999274295604558e-05,
      "loss": 4.1086,
      "step": 194
    },
    {
      "epoch": 0.11223248156896429,
      "grad_norm": 3.3888607025146484,
      "learning_rate": 9.999104073731458e-05,
      "loss": 4.2723,
      "step": 196
    },
    {
      "epoch": 0.11337771097272922,
      "grad_norm": 2.371840715408325,
      "learning_rate": 9.998915936014024e-05,
      "loss": 4.1893,
      "step": 198
    },
    {
      "epoch": 0.11452294037649417,
      "grad_norm": 2.0502302646636963,
      "learning_rate": 9.998709883126502e-05,
      "loss": 4.1395,
      "step": 200
    },
    {
      "epoch": 0.11566816978025911,
      "grad_norm": 1.6674678325653076,
      "learning_rate": 9.998485915807347e-05,
      "loss": 4.071,
      "step": 202
    },
    {
      "epoch": 0.11681339918402404,
      "grad_norm": 1.7829004526138306,
      "learning_rate": 9.998244034859219e-05,
      "loss": 4.1107,
      "step": 204
    },
    {
      "epoch": 0.117958628587789,
      "grad_norm": 1.763493299484253,
      "learning_rate": 9.997984241148967e-05,
      "loss": 4.1142,
      "step": 206
    },
    {
      "epoch": 0.11910385799155393,
      "grad_norm": 2.069258213043213,
      "learning_rate": 9.997706535607649e-05,
      "loss": 4.047,
      "step": 208
    },
    {
      "epoch": 0.12024908739531888,
      "grad_norm": 2.4262139797210693,
      "learning_rate": 9.997410919230505e-05,
      "loss": 4.0396,
      "step": 210
    },
    {
      "epoch": 0.12139431679908382,
      "grad_norm": 1.820494532585144,
      "learning_rate": 9.997097393076971e-05,
      "loss": 4.1548,
      "step": 212
    },
    {
      "epoch": 0.12253954620284876,
      "grad_norm": 2.1332643032073975,
      "learning_rate": 9.996765958270664e-05,
      "loss": 4.1384,
      "step": 214
    },
    {
      "epoch": 0.1236847756066137,
      "grad_norm": 2.1329920291900635,
      "learning_rate": 9.996416615999384e-05,
      "loss": 4.0315,
      "step": 216
    },
    {
      "epoch": 0.12483000501037864,
      "grad_norm": 2.29955792427063,
      "learning_rate": 9.996049367515108e-05,
      "loss": 4.0963,
      "step": 218
    },
    {
      "epoch": 0.12597523441414357,
      "grad_norm": 2.225827693939209,
      "learning_rate": 9.995664214133983e-05,
      "loss": 4.1247,
      "step": 220
    },
    {
      "epoch": 0.12712046381790854,
      "grad_norm": 1.794838786125183,
      "learning_rate": 9.99526115723633e-05,
      "loss": 4.0449,
      "step": 222
    },
    {
      "epoch": 0.12826569322167347,
      "grad_norm": 1.7548491954803467,
      "learning_rate": 9.994840198266626e-05,
      "loss": 3.927,
      "step": 224
    },
    {
      "epoch": 0.1294109226254384,
      "grad_norm": 1.487001895904541,
      "learning_rate": 9.994401338733508e-05,
      "loss": 3.9714,
      "step": 226
    },
    {
      "epoch": 0.13055615202920334,
      "grad_norm": 1.9811242818832397,
      "learning_rate": 9.993944580209768e-05,
      "loss": 4.0094,
      "step": 228
    },
    {
      "epoch": 0.13170138143296828,
      "grad_norm": 1.4257248640060425,
      "learning_rate": 9.99346992433234e-05,
      "loss": 4.0213,
      "step": 230
    },
    {
      "epoch": 0.13284661083673324,
      "grad_norm": 1.545812726020813,
      "learning_rate": 9.992977372802302e-05,
      "loss": 4.0076,
      "step": 232
    },
    {
      "epoch": 0.13399184024049818,
      "grad_norm": 1.8193179368972778,
      "learning_rate": 9.992466927384865e-05,
      "loss": 4.0536,
      "step": 234
    },
    {
      "epoch": 0.1351370696442631,
      "grad_norm": 2.329951763153076,
      "learning_rate": 9.991938589909369e-05,
      "loss": 3.9284,
      "step": 236
    },
    {
      "epoch": 0.13628229904802805,
      "grad_norm": 1.928336501121521,
      "learning_rate": 9.991392362269276e-05,
      "loss": 3.9462,
      "step": 238
    },
    {
      "epoch": 0.137427528451793,
      "grad_norm": 1.4073456525802612,
      "learning_rate": 9.990828246422164e-05,
      "loss": 3.9525,
      "step": 240
    },
    {
      "epoch": 0.13857275785555795,
      "grad_norm": 1.6663973331451416,
      "learning_rate": 9.990246244389713e-05,
      "loss": 3.9685,
      "step": 242
    },
    {
      "epoch": 0.13971798725932288,
      "grad_norm": 1.8091737031936646,
      "learning_rate": 9.989646358257715e-05,
      "loss": 3.9284,
      "step": 244
    },
    {
      "epoch": 0.14086321666308782,
      "grad_norm": 1.5511283874511719,
      "learning_rate": 9.989028590176044e-05,
      "loss": 3.9289,
      "step": 246
    },
    {
      "epoch": 0.14200844606685276,
      "grad_norm": 1.5394625663757324,
      "learning_rate": 9.988392942358664e-05,
      "loss": 3.9849,
      "step": 248
    },
    {
      "epoch": 0.14315367547061772,
      "grad_norm": 1.680882453918457,
      "learning_rate": 9.98773941708362e-05,
      "loss": 3.9452,
      "step": 250
    },
    {
      "epoch": 0.14429890487438266,
      "grad_norm": 1.6341670751571655,
      "learning_rate": 9.98706801669302e-05,
      "loss": 3.8317,
      "step": 252
    },
    {
      "epoch": 0.1454441342781476,
      "grad_norm": 1.9933757781982422,
      "learning_rate": 9.986378743593036e-05,
      "loss": 3.9665,
      "step": 254
    },
    {
      "epoch": 0.14658936368191253,
      "grad_norm": 2.2253994941711426,
      "learning_rate": 9.985671600253894e-05,
      "loss": 3.9239,
      "step": 256
    },
    {
      "epoch": 0.14773459308567746,
      "grad_norm": 2.2543365955352783,
      "learning_rate": 9.984946589209862e-05,
      "loss": 3.8639,
      "step": 258
    },
    {
      "epoch": 0.14887982248944243,
      "grad_norm": 1.8106629848480225,
      "learning_rate": 9.984203713059241e-05,
      "loss": 3.9178,
      "step": 260
    },
    {
      "epoch": 0.15002505189320736,
      "grad_norm": 1.638542652130127,
      "learning_rate": 9.983442974464362e-05,
      "loss": 3.9169,
      "step": 262
    },
    {
      "epoch": 0.1511702812969723,
      "grad_norm": 1.3521384000778198,
      "learning_rate": 9.982664376151564e-05,
      "loss": 3.8682,
      "step": 264
    },
    {
      "epoch": 0.15231551070073723,
      "grad_norm": 1.6458699703216553,
      "learning_rate": 9.981867920911201e-05,
      "loss": 3.9566,
      "step": 266
    },
    {
      "epoch": 0.1534607401045022,
      "grad_norm": 1.7851066589355469,
      "learning_rate": 9.981053611597615e-05,
      "loss": 3.9085,
      "step": 268
    },
    {
      "epoch": 0.15460596950826713,
      "grad_norm": 1.6740517616271973,
      "learning_rate": 9.980221451129137e-05,
      "loss": 3.8899,
      "step": 270
    },
    {
      "epoch": 0.15575119891203207,
      "grad_norm": 1.117129921913147,
      "learning_rate": 9.979371442488073e-05,
      "loss": 3.7544,
      "step": 272
    },
    {
      "epoch": 0.156896428315797,
      "grad_norm": 1.5676058530807495,
      "learning_rate": 9.978503588720694e-05,
      "loss": 3.7753,
      "step": 274
    },
    {
      "epoch": 0.15804165771956194,
      "grad_norm": 1.6609163284301758,
      "learning_rate": 9.977617892937223e-05,
      "loss": 3.8463,
      "step": 276
    },
    {
      "epoch": 0.1591868871233269,
      "grad_norm": 1.7229987382888794,
      "learning_rate": 9.976714358311828e-05,
      "loss": 3.8446,
      "step": 278
    },
    {
      "epoch": 0.16033211652709184,
      "grad_norm": 1.6770962476730347,
      "learning_rate": 9.975792988082603e-05,
      "loss": 3.8684,
      "step": 280
    },
    {
      "epoch": 0.16147734593085677,
      "grad_norm": 1.215281367301941,
      "learning_rate": 9.974853785551568e-05,
      "loss": 3.7788,
      "step": 282
    },
    {
      "epoch": 0.1626225753346217,
      "grad_norm": 1.208257794380188,
      "learning_rate": 9.973896754084646e-05,
      "loss": 3.8338,
      "step": 284
    },
    {
      "epoch": 0.16376780473838665,
      "grad_norm": 1.4068255424499512,
      "learning_rate": 9.972921897111658e-05,
      "loss": 3.8583,
      "step": 286
    },
    {
      "epoch": 0.1649130341421516,
      "grad_norm": 1.4898021221160889,
      "learning_rate": 9.971929218126306e-05,
      "loss": 3.8051,
      "step": 288
    },
    {
      "epoch": 0.16605826354591655,
      "grad_norm": 1.6303211450576782,
      "learning_rate": 9.970918720686164e-05,
      "loss": 3.8598,
      "step": 290
    },
    {
      "epoch": 0.16720349294968148,
      "grad_norm": 1.6599496603012085,
      "learning_rate": 9.969890408412665e-05,
      "loss": 3.7214,
      "step": 292
    },
    {
      "epoch": 0.16834872235344642,
      "grad_norm": 1.1958950757980347,
      "learning_rate": 9.968844284991086e-05,
      "loss": 3.7042,
      "step": 294
    },
    {
      "epoch": 0.16949395175721135,
      "grad_norm": 1.3099420070648193,
      "learning_rate": 9.967780354170533e-05,
      "loss": 3.7405,
      "step": 296
    },
    {
      "epoch": 0.17063918116097632,
      "grad_norm": 1.5054072141647339,
      "learning_rate": 9.966698619763936e-05,
      "loss": 3.7827,
      "step": 298
    },
    {
      "epoch": 0.17178441056474125,
      "grad_norm": 1.444757103919983,
      "learning_rate": 9.965599085648025e-05,
      "loss": 3.7361,
      "step": 300
    }
  ],
  "logging_steps": 2,
  "max_steps": 3494,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 300,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.377550336196608e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}