{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 1004,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00796812749003984,
      "grad_norm": 1.0482105016708374,
      "learning_rate": 9.803921568627452e-08,
      "loss": 2.0931646823883057,
      "step": 2
    },
    {
      "epoch": 0.01593625498007968,
      "grad_norm": 0.7188280820846558,
      "learning_rate": 2.9411764705882356e-07,
      "loss": 2.010826587677002,
      "step": 4
    },
    {
      "epoch": 0.02390438247011952,
      "grad_norm": 5.83541202545166,
      "learning_rate": 4.901960784313725e-07,
      "loss": 2.198476791381836,
      "step": 6
    },
    {
      "epoch": 0.03187250996015936,
      "grad_norm": 0.6360687613487244,
      "learning_rate": 6.862745098039217e-07,
      "loss": 1.9164416790008545,
      "step": 8
    },
    {
      "epoch": 0.0398406374501992,
      "grad_norm": 1.5391534566879272,
      "learning_rate": 8.823529411764707e-07,
      "loss": 1.5691001415252686,
      "step": 10
    },
    {
      "epoch": 0.04780876494023904,
      "grad_norm": 0.8800476789474487,
      "learning_rate": 1.0784313725490197e-06,
      "loss": 1.8166179656982422,
      "step": 12
    },
    {
      "epoch": 0.055776892430278883,
      "grad_norm": 0.6787946224212646,
      "learning_rate": 1.2745098039215686e-06,
      "loss": 2.74003529548645,
      "step": 14
    },
    {
      "epoch": 0.06374501992031872,
      "grad_norm": 3.190701484680176,
      "learning_rate": 1.4705882352941177e-06,
      "loss": 4.131917476654053,
      "step": 16
    },
    {
      "epoch": 0.07171314741035857,
      "grad_norm": 1.3658446073532104,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.747936725616455,
      "step": 18
    },
    {
      "epoch": 0.0796812749003984,
      "grad_norm": 0.5726428031921387,
      "learning_rate": 1.8627450980392158e-06,
      "loss": 1.6613850593566895,
      "step": 20
    },
    {
      "epoch": 0.08764940239043825,
      "grad_norm": 0.9114245176315308,
      "learning_rate": 2.058823529411765e-06,
      "loss": 1.5366348028182983,
      "step": 22
    },
    {
      "epoch": 0.09561752988047809,
      "grad_norm": 1.8325222730636597,
      "learning_rate": 2.254901960784314e-06,
      "loss": 1.445319414138794,
      "step": 24
    },
    {
      "epoch": 0.10358565737051793,
      "grad_norm": 0.8449510335922241,
      "learning_rate": 2.450980392156863e-06,
      "loss": 1.5491961240768433,
      "step": 26
    },
    {
      "epoch": 0.11155378486055777,
      "grad_norm": 0.592979371547699,
      "learning_rate": 2.647058823529412e-06,
      "loss": 1.6504753828048706,
      "step": 28
    },
    {
      "epoch": 0.11952191235059761,
      "grad_norm": 0.8001125454902649,
      "learning_rate": 2.843137254901961e-06,
      "loss": 1.410933017730713,
      "step": 30
    },
    {
      "epoch": 0.12749003984063745,
      "grad_norm": 0.5599525570869446,
      "learning_rate": 3.03921568627451e-06,
      "loss": 1.22208571434021,
      "step": 32
    },
    {
      "epoch": 0.13545816733067728,
      "grad_norm": 1.3079359531402588,
      "learning_rate": 3.2352941176470594e-06,
      "loss": 1.3165050745010376,
      "step": 34
    },
    {
      "epoch": 0.14342629482071714,
      "grad_norm": 1.8055962324142456,
      "learning_rate": 3.431372549019608e-06,
      "loss": 1.469071626663208,
      "step": 36
    },
    {
      "epoch": 0.15139442231075698,
      "grad_norm": 0.6552554368972778,
      "learning_rate": 3.6274509803921573e-06,
      "loss": 1.2445398569107056,
      "step": 38
    },
    {
      "epoch": 0.1593625498007968,
      "grad_norm": 0.43180155754089355,
      "learning_rate": 3.8235294117647055e-06,
      "loss": 1.3755123615264893,
      "step": 40
    },
    {
      "epoch": 0.16733067729083664,
      "grad_norm": 1.9730018377304077,
      "learning_rate": 4.019607843137255e-06,
      "loss": 0.842776894569397,
      "step": 42
    },
    {
      "epoch": 0.1752988047808765,
      "grad_norm": 0.44326290488243103,
      "learning_rate": 4.215686274509805e-06,
      "loss": 1.0599851608276367,
      "step": 44
    },
    {
      "epoch": 0.18326693227091634,
      "grad_norm": 0.6639605164527893,
      "learning_rate": 4.411764705882353e-06,
      "loss": 0.9642063975334167,
      "step": 46
    },
    {
      "epoch": 0.19123505976095617,
      "grad_norm": 0.9954250454902649,
      "learning_rate": 4.607843137254902e-06,
      "loss": 1.216873049736023,
      "step": 48
    },
    {
      "epoch": 0.199203187250996,
      "grad_norm": 0.5103225111961365,
      "learning_rate": 4.803921568627452e-06,
      "loss": 1.373557686805725,
      "step": 50
    },
    {
      "epoch": 0.20717131474103587,
      "grad_norm": 0.48163822293281555,
      "learning_rate": 5e-06,
      "loss": 1.3197792768478394,
      "step": 52
    },
    {
      "epoch": 0.2151394422310757,
      "grad_norm": 3.0629022121429443,
      "learning_rate": 4.999951098195543e-06,
      "loss": 1.1307473182678223,
      "step": 54
    },
    {
      "epoch": 0.22310756972111553,
      "grad_norm": 0.58570796251297,
      "learning_rate": 4.9998043949078495e-06,
      "loss": 1.544344186782837,
      "step": 56
    },
    {
      "epoch": 0.23107569721115537,
      "grad_norm": 0.59855055809021,
      "learning_rate": 4.9995598965138565e-06,
      "loss": 1.370363473892212,
      "step": 58
    },
    {
      "epoch": 0.23904382470119523,
      "grad_norm": 0.4802455008029938,
      "learning_rate": 4.9992176136414875e-06,
      "loss": 0.7791692018508911,
      "step": 60
    },
    {
      "epoch": 0.24701195219123506,
      "grad_norm": 0.5747417211532593,
      "learning_rate": 4.998777561169186e-06,
      "loss": 1.3756986856460571,
      "step": 62
    },
    {
      "epoch": 0.2549800796812749,
      "grad_norm": 0.9288692474365234,
      "learning_rate": 4.998239758225272e-06,
      "loss": 1.1367663145065308,
      "step": 64
    },
    {
      "epoch": 0.26294820717131473,
      "grad_norm": 1.5288456678390503,
      "learning_rate": 4.99760422818711e-06,
      "loss": 1.2961167097091675,
      "step": 66
    },
    {
      "epoch": 0.27091633466135456,
      "grad_norm": 0.9747698903083801,
      "learning_rate": 4.9968709986800925e-06,
      "loss": 0.7980412244796753,
      "step": 68
    },
    {
      "epoch": 0.2788844621513944,
      "grad_norm": 0.9103356599807739,
      "learning_rate": 4.996040101576436e-06,
      "loss": 1.5506565570831299,
      "step": 70
    },
    {
      "epoch": 0.2868525896414343,
      "grad_norm": 0.8636741042137146,
      "learning_rate": 4.995111572993804e-06,
      "loss": 0.8155264854431152,
      "step": 72
    },
    {
      "epoch": 0.2948207171314741,
      "grad_norm": 5.1900410652160645,
      "learning_rate": 4.994085453293725e-06,
      "loss": 1.1488771438598633,
      "step": 74
    },
    {
      "epoch": 0.30278884462151395,
      "grad_norm": 0.8851644992828369,
      "learning_rate": 4.992961787079848e-06,
      "loss": 1.4605716466903687,
      "step": 76
    },
    {
      "epoch": 0.3107569721115538,
      "grad_norm": 3.386746883392334,
      "learning_rate": 4.991740623196001e-06,
      "loss": 0.9014193415641785,
      "step": 78
    },
    {
      "epoch": 0.3187250996015936,
      "grad_norm": 0.30997979640960693,
      "learning_rate": 4.990422014724068e-06,
      "loss": 0.8465147614479065,
      "step": 80
    },
    {
      "epoch": 0.32669322709163345,
      "grad_norm": 0.26163193583488464,
      "learning_rate": 4.989006018981675e-06,
      "loss": 1.3285325765609741,
      "step": 82
    },
    {
      "epoch": 0.3346613545816733,
      "grad_norm": 0.5161040425300598,
      "learning_rate": 4.987492697519712e-06,
      "loss": 0.87628573179245,
      "step": 84
    },
    {
      "epoch": 0.3426294820717131,
      "grad_norm": 0.4508543312549591,
      "learning_rate": 4.985882116119643e-06,
      "loss": 1.3086351156234741,
      "step": 86
    },
    {
      "epoch": 0.350597609561753,
      "grad_norm": 2.1310510635375977,
      "learning_rate": 4.9841743447906576e-06,
      "loss": 0.889850378036499,
      "step": 88
    },
    {
      "epoch": 0.35856573705179284,
      "grad_norm": 0.5565502643585205,
      "learning_rate": 4.982369457766621e-06,
      "loss": 1.3267958164215088,
      "step": 90
    },
    {
      "epoch": 0.3665338645418327,
      "grad_norm": 1.5401748418807983,
      "learning_rate": 4.980467533502853e-06,
      "loss": 1.1103841066360474,
      "step": 92
    },
    {
      "epoch": 0.3745019920318725,
      "grad_norm": 0.5681043863296509,
      "learning_rate": 4.978468654672709e-06,
      "loss": 1.244372010231018,
      "step": 94
    },
    {
      "epoch": 0.38247011952191234,
      "grad_norm": 1.1569910049438477,
      "learning_rate": 4.976372908163999e-06,
      "loss": 1.0917153358459473,
      "step": 96
    },
    {
      "epoch": 0.3904382470119522,
      "grad_norm": 0.3823840320110321,
      "learning_rate": 4.974180385075195e-06,
      "loss": 1.2654279470443726,
      "step": 98
    },
    {
      "epoch": 0.398406374501992,
      "grad_norm": 0.44632136821746826,
      "learning_rate": 4.971891180711488e-06,
      "loss": 0.8519091606140137,
      "step": 100
    },
    {
      "epoch": 0.4063745019920319,
      "grad_norm": 1.2224202156066895,
      "learning_rate": 4.9695053945806295e-06,
      "loss": 1.278132677078247,
      "step": 102
    },
    {
      "epoch": 0.41434262948207173,
      "grad_norm": 1.5874347686767578,
      "learning_rate": 4.9670231303886194e-06,
      "loss": 0.7744755148887634,
      "step": 104
    },
    {
      "epoch": 0.42231075697211157,
      "grad_norm": 3.0088553428649902,
      "learning_rate": 4.9644444960351865e-06,
      "loss": 0.718371570110321,
      "step": 106
    },
    {
      "epoch": 0.4302788844621514,
      "grad_norm": 0.4625864326953888,
      "learning_rate": 4.9617696036091095e-06,
      "loss": 0.9740097522735596,
      "step": 108
    },
    {
      "epoch": 0.43824701195219123,
      "grad_norm": 0.37480488419532776,
      "learning_rate": 4.958998569383335e-06,
      "loss": 1.2884842157363892,
      "step": 110
    },
    {
      "epoch": 0.44621513944223107,
      "grad_norm": 0.5690470933914185,
      "learning_rate": 4.956131513809928e-06,
      "loss": 0.8113313317298889,
      "step": 112
    },
    {
      "epoch": 0.4541832669322709,
      "grad_norm": 0.7082220911979675,
      "learning_rate": 4.953168561514838e-06,
      "loss": 1.127591848373413,
      "step": 114
    },
    {
      "epoch": 0.46215139442231074,
      "grad_norm": 0.6480415463447571,
      "learning_rate": 4.950109841292475e-06,
      "loss": 1.2901209592819214,
      "step": 116
    },
    {
      "epoch": 0.4701195219123506,
      "grad_norm": 0.6014508605003357,
      "learning_rate": 4.946955486100119e-06,
      "loss": 1.3495646715164185,
      "step": 118
    },
    {
      "epoch": 0.47808764940239046,
      "grad_norm": 0.8713721036911011,
      "learning_rate": 4.943705633052134e-06,
      "loss": 1.2338377237319946,
      "step": 120
    },
    {
      "epoch": 0.4860557768924303,
      "grad_norm": 0.6911714673042297,
      "learning_rate": 4.940360423414013e-06,
      "loss": 0.9574092626571655,
      "step": 122
    },
    {
      "epoch": 0.4940239043824701,
      "grad_norm": 3.459429979324341,
      "learning_rate": 4.936920002596233e-06,
      "loss": 1.020338535308838,
      "step": 124
    },
    {
      "epoch": 0.50199203187251,
      "grad_norm": 0.7369417548179626,
      "learning_rate": 4.9333845201479375e-06,
      "loss": 0.9255322813987732,
      "step": 126
    },
    {
      "epoch": 0.5099601593625498,
      "grad_norm": 0.4953453540802002,
      "learning_rate": 4.9297541297504345e-06,
      "loss": 1.2648040056228638,
      "step": 128
    },
    {
      "epoch": 0.5179282868525896,
      "grad_norm": 0.34135428071022034,
      "learning_rate": 4.9260289892105164e-06,
      "loss": 0.9346351623535156,
      "step": 130
    },
    {
      "epoch": 0.5258964143426295,
      "grad_norm": 0.5552108883857727,
      "learning_rate": 4.922209260453599e-06,
      "loss": 0.9497959017753601,
      "step": 132
    },
    {
      "epoch": 0.5338645418326693,
      "grad_norm": 0.49469247460365295,
      "learning_rate": 4.9182951095166875e-06,
      "loss": 0.9156337976455688,
      "step": 134
    },
    {
      "epoch": 0.5418326693227091,
      "grad_norm": 1.0928794145584106,
      "learning_rate": 4.914286706541153e-06,
      "loss": 1.2673850059509277,
      "step": 136
    },
    {
      "epoch": 0.549800796812749,
      "grad_norm": 0.5963903665542603,
      "learning_rate": 4.910184225765341e-06,
      "loss": 0.9366213083267212,
      "step": 138
    },
    {
      "epoch": 0.5577689243027888,
      "grad_norm": 1.3514964580535889,
      "learning_rate": 4.905987845516996e-06,
      "loss": 1.0772197246551514,
      "step": 140
    },
    {
      "epoch": 0.5657370517928287,
      "grad_norm": 0.5856824517250061,
      "learning_rate": 4.90169774820551e-06,
      "loss": 1.0076651573181152,
      "step": 142
    },
    {
      "epoch": 0.5737051792828686,
      "grad_norm": 0.5355371236801147,
      "learning_rate": 4.897314120313996e-06,
      "loss": 1.2183135747909546,
      "step": 144
    },
    {
      "epoch": 0.5816733067729084,
      "grad_norm": 0.6621857285499573,
      "learning_rate": 4.892837152391174e-06,
      "loss": 1.3255361318588257,
      "step": 146
    },
    {
      "epoch": 0.5896414342629482,
      "grad_norm": 0.4817178547382355,
      "learning_rate": 4.888267039043101e-06,
      "loss": 0.9633849859237671,
      "step": 148
    },
    {
      "epoch": 0.5976095617529881,
      "grad_norm": 0.9291821122169495,
      "learning_rate": 4.883603978924698e-06,
      "loss": 0.9330665469169617,
      "step": 150
    },
    {
      "epoch": 0.6055776892430279,
      "grad_norm": 0.924976646900177,
      "learning_rate": 4.878848174731126e-06,
      "loss": 0.9294073581695557,
      "step": 152
    },
    {
      "epoch": 0.6135458167330677,
      "grad_norm": 0.6977740526199341,
      "learning_rate": 4.873999833188967e-06,
      "loss": 1.102484941482544,
      "step": 154
    },
    {
      "epoch": 0.6215139442231076,
      "grad_norm": 4.427737712860107,
      "learning_rate": 4.8690591650472455e-06,
      "loss": 0.8629850149154663,
      "step": 156
    },
    {
      "epoch": 0.6294820717131474,
      "grad_norm": 0.6539608240127563,
      "learning_rate": 4.864026385068259e-06,
      "loss": 1.1252446174621582,
      "step": 158
    },
    {
      "epoch": 0.6374501992031872,
      "grad_norm": 0.7317726612091064,
      "learning_rate": 4.858901712018251e-06,
      "loss": 1.0031373500823975,
      "step": 160
    },
    {
      "epoch": 0.6454183266932271,
      "grad_norm": 0.2943965792655945,
      "learning_rate": 4.8536853686578965e-06,
      "loss": 0.7457835078239441,
      "step": 162
    },
    {
      "epoch": 0.6533864541832669,
      "grad_norm": 0.45527637004852295,
      "learning_rate": 4.84837758173262e-06,
      "loss": 1.12839937210083,
      "step": 164
    },
    {
      "epoch": 0.6613545816733067,
      "grad_norm": 0.7738525867462158,
      "learning_rate": 4.8429785819627395e-06,
      "loss": 0.7770780324935913,
      "step": 166
    },
    {
      "epoch": 0.6693227091633466,
      "grad_norm": 0.7740633487701416,
      "learning_rate": 4.83748860403344e-06,
      "loss": 1.2263681888580322,
      "step": 168
    },
    {
      "epoch": 0.6772908366533864,
      "grad_norm": 6.864331245422363,
      "learning_rate": 4.831907886584566e-06,
      "loss": 0.8660371899604797,
      "step": 170
    },
    {
      "epoch": 0.6852589641434262,
      "grad_norm": 0.5765089392662048,
      "learning_rate": 4.8262366722002535e-06,
      "loss": 0.8344528675079346,
      "step": 172
    },
    {
      "epoch": 0.6932270916334662,
      "grad_norm": 2.0259857177734375,
      "learning_rate": 4.820475207398386e-06,
      "loss": 1.2226604223251343,
      "step": 174
    },
    {
      "epoch": 0.701195219123506,
      "grad_norm": 1.5704021453857422,
      "learning_rate": 4.8146237426198735e-06,
      "loss": 0.7324874401092529,
      "step": 176
    },
    {
      "epoch": 0.7091633466135459,
      "grad_norm": 1.0825178623199463,
      "learning_rate": 4.808682532217772e-06,
      "loss": 0.9091277718544006,
      "step": 178
    },
    {
      "epoch": 0.7171314741035857,
      "grad_norm": 0.8339952230453491,
      "learning_rate": 4.802651834446221e-06,
      "loss": 1.0656994581222534,
      "step": 180
    },
    {
      "epoch": 0.7250996015936255,
      "grad_norm": 0.6570742726325989,
      "learning_rate": 4.796531911449226e-06,
      "loss": 0.5032971501350403,
      "step": 182
    },
    {
      "epoch": 0.7330677290836654,
      "grad_norm": 0.4309996962547302,
      "learning_rate": 4.790323029249252e-06,
      "loss": 1.268394112586975,
      "step": 184
    },
    {
      "epoch": 0.7410358565737052,
      "grad_norm": 0.4892251789569855,
      "learning_rate": 4.784025457735675e-06,
      "loss": 1.2508758306503296,
      "step": 186
    },
    {
      "epoch": 0.749003984063745,
      "grad_norm": 0.785757839679718,
      "learning_rate": 4.777639470653035e-06,
      "loss": 1.0874583721160889,
      "step": 188
    },
    {
      "epoch": 0.7569721115537849,
      "grad_norm": 0.7063523530960083,
      "learning_rate": 4.771165345589148e-06,
      "loss": 0.862820565700531,
      "step": 190
    },
    {
      "epoch": 0.7649402390438247,
      "grad_norm": 1.1867594718933105,
      "learning_rate": 4.7646033639630355e-06,
      "loss": 1.189739465713501,
      "step": 192
    },
    {
      "epoch": 0.7729083665338645,
      "grad_norm": 0.5214909911155701,
      "learning_rate": 4.757953811012688e-06,
      "loss": 0.8712761998176575,
      "step": 194
    },
    {
      "epoch": 0.7808764940239044,
      "grad_norm": 2.2575485706329346,
      "learning_rate": 4.7512169757826735e-06,
      "loss": 0.572406530380249,
      "step": 196
    },
    {
      "epoch": 0.7888446215139442,
      "grad_norm": 1.4402085542678833,
      "learning_rate": 4.74439315111157e-06,
      "loss": 0.7620638608932495,
      "step": 198
    },
    {
      "epoch": 0.796812749003984,
      "grad_norm": 2.033158779144287,
      "learning_rate": 4.737482633619234e-06,
      "loss": 1.0241475105285645,
      "step": 200
    },
    {
      "epoch": 0.8047808764940239,
      "grad_norm": 0.2704363167285919,
      "learning_rate": 4.730485723693912e-06,
      "loss": 0.9573251008987427,
      "step": 202
    },
    {
      "epoch": 0.8127490039840638,
      "grad_norm": 0.818199098110199,
      "learning_rate": 4.723402725479177e-06,
      "loss": 1.5041148662567139,
      "step": 204
    },
    {
      "epoch": 0.8207171314741036,
      "grad_norm": 0.45879414677619934,
      "learning_rate": 4.71623394686071e-06,
      "loss": 1.215051293373108,
      "step": 206
    },
    {
      "epoch": 0.8286852589641435,
      "grad_norm": 0.469187468290329,
      "learning_rate": 4.708979699452923e-06,
      "loss": 0.5916974544525146,
      "step": 208
    },
    {
      "epoch": 0.8366533864541833,
      "grad_norm": 0.4233838617801666,
      "learning_rate": 4.701640298585405e-06,
      "loss": 1.2567811012268066,
      "step": 210
    },
    {
      "epoch": 0.8446215139442231,
      "grad_norm": 0.3439527451992035,
      "learning_rate": 4.694216063289219e-06,
      "loss": 1.2087786197662354,
      "step": 212
    },
    {
      "epoch": 0.852589641434263,
      "grad_norm": 0.4059322476387024,
      "learning_rate": 4.686707316283033e-06,
      "loss": 0.937766432762146,
      "step": 214
    },
    {
      "epoch": 0.8605577689243028,
      "grad_norm": 0.9048867225646973,
      "learning_rate": 4.679114383959095e-06,
      "loss": 1.4285495281219482,
      "step": 216
    },
    {
      "epoch": 0.8685258964143426,
      "grad_norm": 1.5159313678741455,
      "learning_rate": 4.6714375963690415e-06,
      "loss": 0.8288885354995728,
      "step": 218
    },
    {
      "epoch": 0.8764940239043825,
      "grad_norm": 0.4237569272518158,
      "learning_rate": 4.6636772872095545e-06,
      "loss": 1.2278707027435303,
      "step": 220
    },
    {
      "epoch": 0.8844621513944223,
      "grad_norm": 0.44335445761680603,
      "learning_rate": 4.6558337938078505e-06,
      "loss": 1.4773732423782349,
      "step": 222
    },
    {
      "epoch": 0.8924302788844621,
      "grad_norm": 1.114700198173523,
      "learning_rate": 4.647907457107025e-06,
      "loss": 0.6804800629615784,
      "step": 224
    },
    {
      "epoch": 0.900398406374502,
      "grad_norm": 0.41417407989501953,
      "learning_rate": 4.639898621651226e-06,
      "loss": 0.9733927845954895,
      "step": 226
    },
    {
      "epoch": 0.9083665338645418,
      "grad_norm": 0.6831799745559692,
      "learning_rate": 4.631807635570681e-06,
      "loss": 1.0711555480957031,
      "step": 228
    },
    {
      "epoch": 0.9163346613545816,
      "grad_norm": 2.650090456008911,
      "learning_rate": 4.623634850566563e-06,
      "loss": 1.1747108697891235,
      "step": 230
    },
    {
      "epoch": 0.9243027888446215,
      "grad_norm": 0.671407163143158,
      "learning_rate": 4.615380621895701e-06,
      "loss": 1.3159785270690918,
      "step": 232
    },
    {
      "epoch": 0.9322709163346613,
      "grad_norm": 0.38622164726257324,
      "learning_rate": 4.607045308355141e-06,
      "loss": 0.794230580329895,
      "step": 234
    },
    {
      "epoch": 0.9402390438247012,
      "grad_norm": 0.45070669054985046,
      "learning_rate": 4.598629272266549e-06,
      "loss": 0.808007001876831,
      "step": 236
    },
    {
      "epoch": 0.9482071713147411,
      "grad_norm": 1.7445424795150757,
      "learning_rate": 4.590132879460458e-06,
      "loss": 0.4287770688533783,
      "step": 238
    },
    {
      "epoch": 0.9561752988047809,
      "grad_norm": 1.3033678531646729,
      "learning_rate": 4.5815564992603705e-06,
      "loss": 0.834378719329834,
      "step": 240
    },
    {
      "epoch": 0.9641434262948207,
      "grad_norm": 0.6997353434562683,
      "learning_rate": 4.572900504466702e-06,
      "loss": 1.119818091392517,
      "step": 242
    },
    {
      "epoch": 0.9721115537848606,
      "grad_norm": 0.4263991415500641,
      "learning_rate": 4.564165271340577e-06,
      "loss": 1.2070400714874268,
      "step": 244
    },
    {
      "epoch": 0.9800796812749004,
      "grad_norm": 1.0852404832839966,
      "learning_rate": 4.555351179587472e-06,
      "loss": 0.9451199173927307,
      "step": 246
    },
    {
      "epoch": 0.9880478087649402,
      "grad_norm": 0.4493652880191803,
      "learning_rate": 4.546458612340714e-06,
      "loss": 0.9811272621154785,
      "step": 248
    },
    {
      "epoch": 0.9960159362549801,
      "grad_norm": 0.47842052578926086,
      "learning_rate": 4.537487956144821e-06,
      "loss": 0.970288097858429,
      "step": 250
    },
    {
      "epoch": 1.00398406374502,
      "grad_norm": 0.9224259853363037,
      "learning_rate": 4.528439600938706e-06,
      "loss": 0.902433454990387,
      "step": 252
    },
    {
      "epoch": 1.0119521912350598,
      "grad_norm": 0.4232642352581024,
      "learning_rate": 4.519313940038721e-06,
      "loss": 0.7095803022384644,
      "step": 254
    },
    {
      "epoch": 1.0199203187250996,
      "grad_norm": 0.5890194177627563,
      "learning_rate": 4.510111370121563e-06,
      "loss": 1.1579766273498535,
      "step": 256
    },
    {
      "epoch": 1.0278884462151394,
      "grad_norm": 0.5157012343406677,
      "learning_rate": 4.500832291207033e-06,
      "loss": 1.1368385553359985,
      "step": 258
    },
    {
      "epoch": 1.0358565737051793,
      "grad_norm": 0.45233067870140076,
      "learning_rate": 4.491477106640646e-06,
      "loss": 1.0531448125839233,
      "step": 260
    },
    {
      "epoch": 1.043824701195219,
      "grad_norm": 0.6613919138908386,
      "learning_rate": 4.482046223076092e-06,
      "loss": 1.021222472190857,
      "step": 262
    },
    {
      "epoch": 1.051792828685259,
      "grad_norm": 0.48852109909057617,
      "learning_rate": 4.4725400504575725e-06,
      "loss": 1.0678744316101074,
      "step": 264
    },
    {
      "epoch": 1.0597609561752988,
      "grad_norm": 3.3076560497283936,
      "learning_rate": 4.462959002001971e-06,
      "loss": 0.627478837966919,
      "step": 266
    },
    {
      "epoch": 1.0677290836653386,
      "grad_norm": 0.2713685631752014,
      "learning_rate": 4.453303494180895e-06,
      "loss": 0.9125673770904541,
      "step": 268
    },
    {
      "epoch": 1.0756972111553784,
      "grad_norm": 0.35026949644088745,
      "learning_rate": 4.443573946702573e-06,
      "loss": 0.550538957118988,
      "step": 270
    },
    {
      "epoch": 1.0836653386454183,
      "grad_norm": 0.6925679445266724,
      "learning_rate": 4.433770782493606e-06,
      "loss": 1.2292189598083496,
      "step": 272
    },
    {
      "epoch": 1.091633466135458,
      "grad_norm": 0.9226264953613281,
      "learning_rate": 4.42389442768059e-06,
      "loss": 0.5920213460922241,
      "step": 274
    },
    {
      "epoch": 1.099601593625498,
      "grad_norm": 0.6933616995811462,
      "learning_rate": 4.413945311571586e-06,
      "loss": 0.9114224314689636,
      "step": 276
    },
    {
      "epoch": 1.1075697211155378,
      "grad_norm": 0.6991507411003113,
      "learning_rate": 4.40392386663747e-06,
      "loss": 0.8171969056129456,
      "step": 278
    },
    {
      "epoch": 1.1155378486055776,
      "grad_norm": 0.9814272522926331,
      "learning_rate": 4.393830528493118e-06,
      "loss": 1.0298011302947998,
      "step": 280
    },
    {
      "epoch": 1.1235059760956174,
      "grad_norm": 0.6469027996063232,
      "learning_rate": 4.3836657358784875e-06,
      "loss": 1.1517784595489502,
      "step": 282
    },
    {
      "epoch": 1.1314741035856573,
      "grad_norm": 0.5742753744125366,
      "learning_rate": 4.373429930639534e-06,
      "loss": 1.1802786588668823,
      "step": 284
    },
    {
      "epoch": 1.139442231075697,
      "grad_norm": 0.24241602420806885,
      "learning_rate": 4.3631235577090085e-06,
      "loss": 0.31692224740982056,
      "step": 286
    },
    {
      "epoch": 1.1474103585657371,
      "grad_norm": 0.29871371388435364,
      "learning_rate": 4.35274706508712e-06,
      "loss": 0.7072598338127136,
      "step": 288
    },
    {
      "epoch": 1.155378486055777,
      "grad_norm": 0.7520034909248352,
      "learning_rate": 4.342300903822057e-06,
      "loss": 0.9134758710861206,
      "step": 290
    },
    {
      "epoch": 1.1633466135458168,
      "grad_norm": 0.5191479921340942,
      "learning_rate": 4.331785527990384e-06,
      "loss": 0.5000742077827454,
      "step": 292
    },
    {
      "epoch": 1.1713147410358566,
      "grad_norm": 0.6116405129432678,
      "learning_rate": 4.3212013946773045e-06,
      "loss": 1.1868194341659546,
      "step": 294
    },
    {
      "epoch": 1.1792828685258965,
      "grad_norm": 0.47647368907928467,
      "learning_rate": 4.31054896395679e-06,
      "loss": 1.2218159437179565,
      "step": 296
    },
    {
      "epoch": 1.1872509960159363,
      "grad_norm": 0.6232501864433289,
      "learning_rate": 4.299828698871581e-06,
      "loss": 1.3418570756912231,
      "step": 298
    },
    {
      "epoch": 1.1952191235059761,
      "grad_norm": 0.8073158860206604,
      "learning_rate": 4.289041065413062e-06,
      "loss": 0.5976051092147827,
      "step": 300
    },
    {
      "epoch": 1.203187250996016,
      "grad_norm": 0.8489817976951599,
      "learning_rate": 4.278186532501003e-06,
      "loss": 0.9240081310272217,
      "step": 302
    },
    {
      "epoch": 1.2111553784860558,
      "grad_norm": 2.4210026264190674,
      "learning_rate": 4.267265571963179e-06,
      "loss": 0.5354931950569153,
      "step": 304
    },
    {
      "epoch": 1.2191235059760956,
      "grad_norm": 0.4991712272167206,
      "learning_rate": 4.256278658514858e-06,
      "loss": 1.1688907146453857,
      "step": 306
    },
    {
      "epoch": 1.2270916334661355,
      "grad_norm": 2.856494665145874,
      "learning_rate": 4.245226269738165e-06,
      "loss": 1.0135537385940552,
      "step": 308
    },
    {
      "epoch": 1.2350597609561753,
      "grad_norm": 0.5454363226890564,
      "learning_rate": 4.23410888606133e-06,
      "loss": 1.1315621137619019,
      "step": 310
    },
    {
      "epoch": 1.2430278884462151,
      "grad_norm": 0.8048802018165588,
      "learning_rate": 4.222926990737793e-06,
      "loss": 1.0523592233657837,
      "step": 312
    },
    {
      "epoch": 1.250996015936255,
      "grad_norm": 0.9349963665008545,
      "learning_rate": 4.211681069825207e-06,
      "loss": 0.8602126836776733,
      "step": 314
    },
    {
      "epoch": 1.2589641434262948,
      "grad_norm": 0.46230587363243103,
      "learning_rate": 4.200371612164305e-06,
      "loss": 0.7222084999084473,
      "step": 316
    },
    {
      "epoch": 1.2669322709163346,
      "grad_norm": 1.1461910009384155,
      "learning_rate": 4.1889991093576555e-06,
      "loss": 0.6046090722084045,
      "step": 318
    },
    {
      "epoch": 1.2749003984063745,
      "grad_norm": 3.1880223751068115,
      "learning_rate": 4.177564055748284e-06,
      "loss": 0.5082796812057495,
      "step": 320
    },
    {
      "epoch": 1.2828685258964143,
      "grad_norm": 1.1039406061172485,
      "learning_rate": 4.166066948398199e-06,
      "loss": 0.976899266242981,
      "step": 322
    },
    {
      "epoch": 1.2908366533864541,
      "grad_norm": 0.9321113228797913,
      "learning_rate": 4.154508287066772e-06,
      "loss": 0.9157530069351196,
      "step": 324
    },
    {
      "epoch": 1.298804780876494,
      "grad_norm": 0.7307058572769165,
      "learning_rate": 4.142888574189025e-06,
      "loss": 1.318347692489624,
      "step": 326
    },
    {
      "epoch": 1.3067729083665338,
      "grad_norm": 0.481071799993515,
      "learning_rate": 4.131208314853778e-06,
      "loss": 0.6828722357749939,
      "step": 328
    },
    {
      "epoch": 1.3147410358565736,
      "grad_norm": 0.5134776830673218,
      "learning_rate": 4.119468016781707e-06,
      "loss": 0.8917183876037598,
      "step": 330
    },
    {
      "epoch": 1.3227091633466135,
      "grad_norm": 5.735156059265137,
      "learning_rate": 4.107668190303266e-06,
      "loss": 0.9180288910865784,
      "step": 332
    },
    {
      "epoch": 1.3306772908366533,
      "grad_norm": 2.1222012042999268,
      "learning_rate": 4.095809348336506e-06,
      "loss": 0.6715309023857117,
      "step": 334
    },
    {
      "epoch": 1.3386454183266931,
      "grad_norm": 1.1036757230758667,
      "learning_rate": 4.083892006364777e-06,
      "loss": 0.8704354166984558,
      "step": 336
    },
    {
      "epoch": 1.3466135458167332,
      "grad_norm": 0.7863856554031372,
      "learning_rate": 4.071916682414327e-06,
      "loss": 1.1333597898483276,
      "step": 338
    },
    {
      "epoch": 1.354581673306773,
      "grad_norm": 0.35504376888275146,
      "learning_rate": 4.059883897031777e-06,
      "loss": 1.128110647201538,
      "step": 340
    },
    {
      "epoch": 1.3625498007968129,
      "grad_norm": 0.743506908416748,
      "learning_rate": 4.0477941732615e-06,
      "loss": 1.1713272333145142,
      "step": 342
    },
    {
      "epoch": 1.3705179282868527,
      "grad_norm": 2.113736152648926,
      "learning_rate": 4.035648036622879e-06,
      "loss": 1.1382900476455688,
      "step": 344
    },
    {
      "epoch": 1.3784860557768925,
      "grad_norm": 0.6295567750930786,
      "learning_rate": 4.023446015087469e-06,
      "loss": 0.7032626271247864,
      "step": 346
    },
    {
      "epoch": 1.3864541832669324,
      "grad_norm": 0.42242372035980225,
      "learning_rate": 4.0111886390560446e-06,
      "loss": 1.1623692512512207,
      "step": 348
    },
    {
      "epoch": 1.3944223107569722,
      "grad_norm": 1.0018078088760376,
      "learning_rate": 3.998876441335544e-06,
      "loss": 0.6742269992828369,
      "step": 350
    },
    {
      "epoch": 1.402390438247012,
      "grad_norm": 0.3838563561439514,
      "learning_rate": 3.98650995711591e-06,
      "loss": 1.1845507621765137,
      "step": 352
    },
    {
      "epoch": 1.4103585657370519,
      "grad_norm": 0.4354332983493805,
      "learning_rate": 3.974089723946826e-06,
      "loss": 1.1132577657699585,
      "step": 354
    },
    {
      "epoch": 1.4183266932270917,
      "grad_norm": 0.7023118734359741,
      "learning_rate": 3.9616162817143465e-06,
      "loss": 0.9047975540161133,
      "step": 356
    },
    {
      "epoch": 1.4262948207171315,
      "grad_norm": 0.6091473698616028,
      "learning_rate": 3.949090172617437e-06,
      "loss": 1.1451220512390137,
      "step": 358
    },
    {
      "epoch": 1.4342629482071714,
      "grad_norm": 0.6251921653747559,
      "learning_rate": 3.936511941144397e-06,
      "loss": 0.5954256653785706,
      "step": 360
    },
    {
      "epoch": 1.4422310756972112,
      "grad_norm": 0.4494275450706482,
      "learning_rate": 3.9238821340491936e-06,
      "loss": 0.7277942895889282,
      "step": 362
    },
    {
      "epoch": 1.450199203187251,
      "grad_norm": 1.3889179229736328,
      "learning_rate": 3.911201300327703e-06,
      "loss": 1.067896842956543,
      "step": 364
    },
    {
      "epoch": 1.4581673306772909,
      "grad_norm": 0.4396206736564636,
      "learning_rate": 3.898469991193835e-06,
      "loss": 1.1362249851226807,
      "step": 366
    },
    {
      "epoch": 1.4661354581673307,
      "grad_norm": 0.4632607400417328,
      "learning_rate": 3.885688760055581e-06,
      "loss": 0.6078506112098694,
      "step": 368
    },
    {
      "epoch": 1.4741035856573705,
      "grad_norm": 0.5464065074920654,
      "learning_rate": 3.872858162490956e-06,
      "loss": 0.6817663908004761,
      "step": 370
    },
    {
      "epoch": 1.4820717131474104,
      "grad_norm": 0.9941257238388062,
      "learning_rate": 3.859978756223845e-06,
      "loss": 1.2282507419586182,
      "step": 372
    },
    {
      "epoch": 1.4900398406374502,
      "grad_norm": 0.46521931886672974,
      "learning_rate": 3.847051101099768e-06,
      "loss": 1.1383668184280396,
      "step": 374
    },
    {
      "epoch": 1.49800796812749,
      "grad_norm": 0.5094945430755615,
      "learning_rate": 3.834075759061535e-06,
      "loss": 1.1322904825210571,
      "step": 376
    },
    {
      "epoch": 1.5059760956175299,
      "grad_norm": 0.7781340479850769,
      "learning_rate": 3.8210532941248255e-06,
      "loss": 0.910107433795929,
      "step": 378
    },
    {
      "epoch": 1.5139442231075697,
      "grad_norm": 0.412298321723938,
      "learning_rate": 3.8079842723536688e-06,
      "loss": 1.1197013854980469,
      "step": 380
    },
    {
      "epoch": 1.5219123505976095,
      "grad_norm": 0.46745553612709045,
      "learning_rate": 3.7948692618358424e-06,
      "loss": 0.7607988715171814,
      "step": 382
    },
    {
      "epoch": 1.5298804780876494,
      "grad_norm": 0.4306337833404541,
      "learning_rate": 3.781708832658169e-06,
      "loss": 0.7653713822364807,
      "step": 384
    },
    {
      "epoch": 1.5378486055776892,
      "grad_norm": 0.5786705017089844,
      "learning_rate": 3.7685035568817492e-06,
      "loss": 1.014581561088562,
      "step": 386
    },
    {
      "epoch": 1.545816733067729,
      "grad_norm": 0.9059370160102844,
      "learning_rate": 3.7552540085170826e-06,
      "loss": 1.1434062719345093,
      "step": 388
    },
    {
      "epoch": 1.5537848605577689,
      "grad_norm": 0.8374323844909668,
      "learning_rate": 3.7419607634991235e-06,
      "loss": 0.5667718052864075,
      "step": 390
    },
    {
      "epoch": 1.5617529880478087,
      "grad_norm": 0.9594655632972717,
      "learning_rate": 3.7286243996622432e-06,
      "loss": 0.588732898235321,
      "step": 392
    },
    {
      "epoch": 1.5697211155378485,
      "grad_norm": 1.5200785398483276,
      "learning_rate": 3.715245496715116e-06,
      "loss": 1.1121203899383545,
      "step": 394
    },
    {
      "epoch": 1.5776892430278884,
      "grad_norm": 0.9333744645118713,
      "learning_rate": 3.7018246362155134e-06,
      "loss": 1.040864109992981,
      "step": 396
    },
    {
      "epoch": 1.5856573705179282,
      "grad_norm": 0.5568398833274841,
      "learning_rate": 3.6883624015450344e-06,
      "loss": 1.1321748495101929,
      "step": 398
    },
    {
      "epoch": 1.593625498007968,
      "grad_norm": 0.4377819001674652,
      "learning_rate": 3.6748593778837367e-06,
      "loss": 0.5998024940490723,
      "step": 400
    },
    {
      "epoch": 1.6015936254980079,
      "grad_norm": 0.5778295397758484,
      "learning_rate": 3.661316152184708e-06,
      "loss": 1.1140027046203613,
      "step": 402
    },
    {
      "epoch": 1.6095617529880477,
      "grad_norm": 0.5599844455718994,
      "learning_rate": 3.6477333131485483e-06,
      "loss": 1.2616959810256958,
      "step": 404
    },
    {
      "epoch": 1.6175298804780875,
      "grad_norm": 3.061119318008423,
      "learning_rate": 3.634111451197781e-06,
      "loss": 0.729964554309845,
      "step": 406
    },
    {
      "epoch": 1.6254980079681274,
      "grad_norm": 0.29869139194488525,
      "learning_rate": 3.6204511584511865e-06,
      "loss": 0.5784701704978943,
      "step": 408
    },
    {
      "epoch": 1.6334661354581672,
      "grad_norm": 0.6870218515396118,
      "learning_rate": 3.606753028698067e-06,
      "loss": 0.7673273682594299,
      "step": 410
    },
    {
      "epoch": 1.641434262948207,
      "grad_norm": 1.2924615144729614,
      "learning_rate": 3.593017657372436e-06,
      "loss": 0.46346667408943176,
      "step": 412
    },
    {
      "epoch": 1.6494023904382469,
      "grad_norm": 0.6762600541114807,
      "learning_rate": 3.5792456415271288e-06,
      "loss": 0.8462737798690796,
      "step": 414
    },
    {
      "epoch": 1.6573705179282867,
      "grad_norm": 0.6838541030883789,
      "learning_rate": 3.565437579807858e-06,
      "loss": 0.7577676773071289,
      "step": 416
    },
    {
      "epoch": 1.6653386454183265,
      "grad_norm": 1.115029215812683,
      "learning_rate": 3.551594072427188e-06,
      "loss": 0.979465663433075,
      "step": 418
    },
    {
      "epoch": 1.6733067729083664,
      "grad_norm": 0.4735439717769623,
      "learning_rate": 3.537715721138444e-06,
      "loss": 1.1334731578826904,
      "step": 420
    },
    {
      "epoch": 1.6812749003984062,
      "grad_norm": 0.7888314723968506,
      "learning_rate": 3.523803129209554e-06,
      "loss": 1.0223737955093384,
      "step": 422
    },
    {
      "epoch": 1.6892430278884463,
      "grad_norm": 0.6206667423248291,
      "learning_rate": 3.5098569013968314e-06,
      "loss": 0.6118888854980469,
      "step": 424
    },
    {
      "epoch": 1.697211155378486,
      "grad_norm": 0.38839226961135864,
      "learning_rate": 3.49587764391868e-06,
      "loss": 1.1986186504364014,
      "step": 426
    },
    {
      "epoch": 1.705179282868526,
      "grad_norm": 0.4477618932723999,
      "learning_rate": 3.4818659644292473e-06,
      "loss": 1.1396212577819824,
      "step": 428
    },
    {
      "epoch": 1.7131474103585658,
      "grad_norm": 0.7599515914916992,
      "learning_rate": 3.4678224719920085e-06,
      "loss": 0.7236371636390686,
      "step": 430
    },
    {
      "epoch": 1.7211155378486056,
      "grad_norm": 0.7597646713256836,
      "learning_rate": 3.4537477770532933e-06,
      "loss": 0.9114875197410583,
      "step": 432
    },
    {
      "epoch": 1.7290836653386454,
      "grad_norm": 1.0384747982025146,
      "learning_rate": 3.4396424914157527e-06,
      "loss": 0.6820817589759827,
      "step": 434
    },
    {
      "epoch": 1.7370517928286853,
      "grad_norm": 2.8560690879821777,
      "learning_rate": 3.4255072282117574e-06,
      "loss": 0.3695479929447174,
      "step": 436
    },
    {
      "epoch": 1.745019920318725,
      "grad_norm": 0.4039926528930664,
      "learning_rate": 3.411342601876754e-06,
      "loss": 0.7630254626274109,
      "step": 438
    },
    {
      "epoch": 1.752988047808765,
      "grad_norm": 19.535093307495117,
      "learning_rate": 3.3971492281225554e-06,
      "loss": 0.6585661768913269,
      "step": 440
    },
    {
      "epoch": 1.7609561752988048,
      "grad_norm": 0.9871184825897217,
      "learning_rate": 3.382927723910572e-06,
      "loss": 1.4820557832717896,
      "step": 442
    },
    {
      "epoch": 1.7689243027888446,
      "grad_norm": 0.6368772983551025,
      "learning_rate": 3.3686787074249966e-06,
      "loss": 0.7925757169723511,
      "step": 444
    },
    {
      "epoch": 1.7768924302788844,
      "grad_norm": 0.43703126907348633,
      "learning_rate": 3.354402798045935e-06,
      "loss": 1.0465331077575684,
      "step": 446
    },
    {
      "epoch": 1.7848605577689243,
      "grad_norm": 0.498674601316452,
      "learning_rate": 3.340100616322479e-06,
      "loss": 1.0953220129013062,
      "step": 448
    },
    {
      "epoch": 1.792828685258964,
      "grad_norm": 0.38230448961257935,
      "learning_rate": 3.3257727839457356e-06,
      "loss": 0.8427063226699829,
      "step": 450
    },
    {
      "epoch": 1.800796812749004,
      "grad_norm": 0.5142718553543091,
      "learning_rate": 3.311419923721798e-06,
      "loss": 0.5237646102905273,
      "step": 452
    },
    {
      "epoch": 1.8087649402390438,
      "grad_norm": 0.9807317852973938,
      "learning_rate": 3.29704265954468e-06,
      "loss": 0.9383044242858887,
      "step": 454
    },
    {
      "epoch": 1.8167330677290838,
      "grad_norm": 1.4895225763320923,
      "learning_rate": 3.282641616369193e-06,
      "loss": 0.7038347125053406,
      "step": 456
    },
    {
      "epoch": 1.8247011952191237,
      "grad_norm": 0.6303024291992188,
      "learning_rate": 3.2682174201837773e-06,
      "loss": 1.041494607925415,
      "step": 458
    },
    {
      "epoch": 1.8326693227091635,
      "grad_norm": 0.7198854088783264,
      "learning_rate": 3.2537706979832955e-06,
      "loss": 0.7393121123313904,
      "step": 460
    },
    {
      "epoch": 1.8406374501992033,
      "grad_norm": 0.5643909573554993,
      "learning_rate": 3.23930207774178e-06,
      "loss": 0.7915079593658447,
      "step": 462
    },
    {
      "epoch": 1.8486055776892432,
      "grad_norm": 0.3143250346183777,
      "learning_rate": 3.2248121883851296e-06,
      "loss": 0.11594223231077194,
      "step": 464
    },
    {
      "epoch": 1.856573705179283,
      "grad_norm": 0.47754329442977905,
      "learning_rate": 3.210301659763777e-06,
      "loss": 1.152090072631836,
      "step": 466
    },
    {
      "epoch": 1.8645418326693228,
      "grad_norm": 2.1267971992492676,
      "learning_rate": 3.1957711226253073e-06,
      "loss": 0.5501399040222168,
      "step": 468
    },
    {
      "epoch": 1.8725099601593627,
      "grad_norm": 0.8010175228118896,
      "learning_rate": 3.181221208587041e-06,
      "loss": 1.139042615890503,
      "step": 470
    },
    {
      "epoch": 1.8804780876494025,
      "grad_norm": 0.7370595335960388,
      "learning_rate": 3.16665255010858e-06,
      "loss": 1.020094633102417,
      "step": 472
    },
    {
      "epoch": 1.8884462151394423,
      "grad_norm": 0.8562900424003601,
      "learning_rate": 3.152065780464311e-06,
      "loss": 1.2348934412002563,
      "step": 474
    },
    {
      "epoch": 1.8964143426294822,
      "grad_norm": 0.7433300018310547,
      "learning_rate": 3.1374615337158874e-06,
      "loss": 0.9736502766609192,
      "step": 476
    },
    {
      "epoch": 1.904382470119522,
      "grad_norm": 0.3962269127368927,
      "learning_rate": 3.1228404446846577e-06,
      "loss": 0.8500668406486511,
      "step": 478
    },
    {
      "epoch": 1.9123505976095618,
      "grad_norm": 2.1908648014068604,
      "learning_rate": 3.1082031489240775e-06,
      "loss": 0.6073490977287292,
      "step": 480
    },
    {
      "epoch": 1.9203187250996017,
      "grad_norm": 1.400114893913269,
      "learning_rate": 3.093550282692081e-06,
      "loss": 0.7533238530158997,
      "step": 482
    },
    {
      "epoch": 1.9282868525896415,
      "grad_norm": 0.5276299118995667,
      "learning_rate": 3.0788824829234206e-06,
      "loss": 1.111577033996582,
      "step": 484
    },
    {
      "epoch": 1.9362549800796813,
      "grad_norm": 0.9245796799659729,
      "learning_rate": 3.064200387201987e-06,
      "loss": 1.1432287693023682,
      "step": 486
    },
    {
      "epoch": 1.9442231075697212,
      "grad_norm": 0.3796207010746002,
      "learning_rate": 3.0495046337330917e-06,
      "loss": 1.1035107374191284,
      "step": 488
    },
    {
      "epoch": 1.952191235059761,
      "grad_norm": 0.2961997389793396,
      "learning_rate": 3.034795861315723e-06,
      "loss": 1.116678237915039,
      "step": 490
    },
    {
      "epoch": 1.9601593625498008,
      "grad_norm": 0.5441898703575134,
      "learning_rate": 3.020074709314781e-06,
      "loss": 0.7877160310745239,
      "step": 492
    },
    {
      "epoch": 1.9681274900398407,
      "grad_norm": 4.0410237312316895,
      "learning_rate": 3.0053418176332844e-06,
      "loss": 0.6664410829544067,
      "step": 494
    },
    {
      "epoch": 1.9760956175298805,
      "grad_norm": 0.38009560108184814,
      "learning_rate": 2.9905978266845565e-06,
      "loss": 1.0719088315963745,
      "step": 496
    },
    {
      "epoch": 1.9840637450199203,
      "grad_norm": 1.2558846473693848,
      "learning_rate": 2.975843377364385e-06,
      "loss": 0.8662068843841553,
      "step": 498
    },
    {
      "epoch": 1.9920318725099602,
      "grad_norm": 0.42220982909202576,
      "learning_rate": 2.961079111023167e-06,
      "loss": 1.1645177602767944,
      "step": 500
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.3077664375305176,
      "learning_rate": 2.9463056694380263e-06,
      "loss": 0.5905117392539978,
      "step": 502
    },
    {
      "epoch": 2.00796812749004,
      "grad_norm": 0.42836296558380127,
      "learning_rate": 2.9315236947849213e-06,
      "loss": 0.6424471735954285,
      "step": 504
    },
    {
      "epoch": 2.0159362549800797,
      "grad_norm": 0.8567079901695251,
      "learning_rate": 2.916733829610726e-06,
      "loss": 0.6191027760505676,
      "step": 506
    },
    {
      "epoch": 2.0239043824701195,
      "grad_norm": 0.08191274106502533,
      "learning_rate": 2.901936716805302e-06,
      "loss": 0.5529719591140747,
      "step": 508
    },
    {
      "epoch": 2.0318725099601593,
      "grad_norm": 0.6572995781898499,
      "learning_rate": 2.887132999573554e-06,
      "loss": 0.8550249338150024,
      "step": 510
    },
    {
      "epoch": 2.039840637450199,
      "grad_norm": 2.07453989982605,
      "learning_rate": 2.8723233214074675e-06,
      "loss": 0.8121019005775452,
      "step": 512
    },
    {
      "epoch": 2.047808764940239,
      "grad_norm": 0.3679105341434479,
      "learning_rate": 2.857508326058142e-06,
      "loss": 0.27556997537612915,
      "step": 514
    },
    {
      "epoch": 2.055776892430279,
      "grad_norm": 0.4557670056819916,
      "learning_rate": 2.842688657507805e-06,
      "loss": 0.9122077822685242,
      "step": 516
    },
    {
      "epoch": 2.0637450199203187,
      "grad_norm": 0.8464804291725159,
      "learning_rate": 2.8278649599418194e-06,
      "loss": 0.562198281288147,
      "step": 518
    },
    {
      "epoch": 2.0717131474103585,
      "grad_norm": 0.7567138075828552,
      "learning_rate": 2.813037877720683e-06,
      "loss": 0.8457916378974915,
      "step": 520
    },
    {
      "epoch": 2.0796812749003983,
      "grad_norm": 0.45435866713523865,
      "learning_rate": 2.798208055352017e-06,
      "loss": 1.0143765211105347,
      "step": 522
    },
    {
      "epoch": 2.087649402390438,
      "grad_norm": 0.9655630588531494,
      "learning_rate": 2.7833761374625546e-06,
      "loss": 0.9420502781867981,
      "step": 524
    },
    {
      "epoch": 2.095617529880478,
      "grad_norm": 0.8306463956832886,
      "learning_rate": 2.7685427687701172e-06,
      "loss": 0.42675721645355225,
      "step": 526
    },
    {
      "epoch": 2.103585657370518,
      "grad_norm": 0.4111825227737427,
      "learning_rate": 2.753708594055588e-06,
      "loss": 0.7460598349571228,
      "step": 528
    },
    {
      "epoch": 2.1115537848605577,
      "grad_norm": 0.28190386295318604,
      "learning_rate": 2.7388742581348893e-06,
      "loss": 0.6903210282325745,
      "step": 530
    },
    {
      "epoch": 2.1195219123505975,
      "grad_norm": 0.7143102884292603,
      "learning_rate": 2.7240404058309487e-06,
      "loss": 0.21371248364448547,
      "step": 532
    },
    {
      "epoch": 2.1274900398406373,
      "grad_norm": 0.6563390493392944,
      "learning_rate": 2.709207681945673e-06,
      "loss": 0.7819314002990723,
      "step": 534
    },
    {
      "epoch": 2.135458167330677,
      "grad_norm": 0.8471841216087341,
      "learning_rate": 2.6943767312319167e-06,
      "loss": 0.6592696905136108,
      "step": 536
    },
    {
      "epoch": 2.143426294820717,
      "grad_norm": 2.0856447219848633,
      "learning_rate": 2.679548198365462e-06,
      "loss": 0.7053009271621704,
      "step": 538
    },
    {
      "epoch": 2.151394422310757,
      "grad_norm": 0.5350284576416016,
      "learning_rate": 2.6647227279169873e-06,
      "loss": 0.9408332705497742,
      "step": 540
    },
    {
      "epoch": 2.1593625498007967,
      "grad_norm": 0.9136908054351807,
      "learning_rate": 2.6499009643240527e-06,
      "loss": 0.9364646673202515,
      "step": 542
    },
    {
      "epoch": 2.1673306772908365,
      "grad_norm": 0.6921311616897583,
      "learning_rate": 2.6350835518630907e-06,
      "loss": 0.8164045214653015,
      "step": 544
    },
    {
      "epoch": 2.1752988047808763,
      "grad_norm": 0.5685932040214539,
      "learning_rate": 2.620271134621397e-06,
      "loss": 0.6746962070465088,
      "step": 546
    },
    {
      "epoch": 2.183266932270916,
      "grad_norm": 0.7439118027687073,
      "learning_rate": 2.605464356469131e-06,
      "loss": 0.9891138076782227,
      "step": 548
    },
    {
      "epoch": 2.191235059760956,
      "grad_norm": 0.40500935912132263,
      "learning_rate": 2.5906638610313336e-06,
      "loss": 1.0538139343261719,
      "step": 550
    },
    {
      "epoch": 2.199203187250996,
      "grad_norm": 1.385208010673523,
      "learning_rate": 2.575870291659946e-06,
      "loss": 0.7194516062736511,
      "step": 552
    },
    {
      "epoch": 2.2071713147410357,
      "grad_norm": 1.0170742273330688,
      "learning_rate": 2.5610842914058446e-06,
      "loss": 0.4520931839942932,
      "step": 554
    },
    {
      "epoch": 2.2151394422310755,
      "grad_norm": 0.668999969959259,
      "learning_rate": 2.5463065029908906e-06,
      "loss": 0.6776040196418762,
      "step": 556
    },
    {
      "epoch": 2.2231075697211153,
      "grad_norm": 0.8320958018302917,
      "learning_rate": 2.5315375687799896e-06,
      "loss": 0.8635607957839966,
      "step": 558
    },
    {
      "epoch": 2.231075697211155,
      "grad_norm": 0.5701048374176025,
      "learning_rate": 2.5167781307531714e-06,
      "loss": 0.9859389662742615,
      "step": 560
    },
    {
      "epoch": 2.239043824701195,
      "grad_norm": 0.7431889176368713,
      "learning_rate": 2.5020288304776822e-06,
      "loss": 1.0431369543075562,
      "step": 562
    },
    {
      "epoch": 2.247011952191235,
      "grad_norm": 0.7676903605461121,
      "learning_rate": 2.487290309080097e-06,
      "loss": 0.8322744965553284,
      "step": 564
    },
    {
      "epoch": 2.2549800796812747,
      "grad_norm": 0.5741667151451111,
      "learning_rate": 2.472563207218454e-06,
      "loss": 0.6705302596092224,
      "step": 566
    },
    {
      "epoch": 2.2629482071713145,
      "grad_norm": 0.47801673412323,
      "learning_rate": 2.457848165054402e-06,
      "loss": 0.7613603472709656,
      "step": 568
    },
    {
      "epoch": 2.2709163346613543,
      "grad_norm": 0.3870161771774292,
      "learning_rate": 2.4431458222253755e-06,
      "loss": 1.0242830514907837,
      "step": 570
    },
    {
      "epoch": 2.278884462151394,
      "grad_norm": 0.8235284686088562,
      "learning_rate": 2.4284568178167923e-06,
      "loss": 0.6202554106712341,
      "step": 572
    },
    {
      "epoch": 2.2868525896414345,
      "grad_norm": 0.787039041519165,
      "learning_rate": 2.413781790334272e-06,
      "loss": 0.7180704474449158,
      "step": 574
    },
    {
      "epoch": 2.2948207171314743,
      "grad_norm": 0.29202213883399963,
      "learning_rate": 2.3991213776758776e-06,
      "loss": 0.25315162539482117,
      "step": 576
    },
    {
      "epoch": 2.302788844621514,
      "grad_norm": 0.5496858358383179,
      "learning_rate": 2.3844762171043975e-06,
      "loss": 0.6483262181282043,
      "step": 578
    },
    {
      "epoch": 2.310756972111554,
      "grad_norm": 0.4383964240550995,
      "learning_rate": 2.3698469452196328e-06,
      "loss": 0.6292935609817505,
      "step": 580
    },
    {
      "epoch": 2.318725099601594,
      "grad_norm": 0.9571503400802612,
      "learning_rate": 2.355234197930735e-06,
      "loss": 0.6716011166572571,
      "step": 582
    },
    {
      "epoch": 2.3266932270916336,
      "grad_norm": 0.4092956483364105,
      "learning_rate": 2.340638610428554e-06,
      "loss": 0.8172141909599304,
      "step": 584
    },
    {
      "epoch": 2.3346613545816735,
      "grad_norm": 1.26423978805542,
      "learning_rate": 2.3260608171580412e-06,
      "loss": 0.6475146412849426,
      "step": 586
    },
    {
      "epoch": 2.3426294820717133,
      "grad_norm": 0.7899402976036072,
      "learning_rate": 2.311501451790658e-06,
      "loss": 0.9060088992118835,
      "step": 588
    },
    {
      "epoch": 2.350597609561753,
      "grad_norm": 0.4412243366241455,
      "learning_rate": 2.2969611471968386e-06,
      "loss": 1.0190945863723755,
      "step": 590
    },
    {
      "epoch": 2.358565737051793,
      "grad_norm": 0.46188387274742126,
      "learning_rate": 2.2824405354184776e-06,
      "loss": 1.051806926727295,
      "step": 592
    },
    {
      "epoch": 2.366533864541833,
      "grad_norm": 0.4580332636833191,
      "learning_rate": 2.267940247641458e-06,
      "loss": 0.7564969062805176,
      "step": 594
    },
    {
      "epoch": 2.3745019920318726,
      "grad_norm": 0.4431792199611664,
      "learning_rate": 2.2534609141682135e-06,
      "loss": 0.9973644614219666,
      "step": 596
    },
    {
      "epoch": 2.3824701195219125,
      "grad_norm": 0.4597433805465698,
      "learning_rate": 2.2390031643903286e-06,
      "loss": 0.7255712151527405,
      "step": 598
    },
    {
      "epoch": 2.3904382470119523,
      "grad_norm": 1.4946357011795044,
      "learning_rate": 2.224567626761185e-06,
      "loss": 0.9277011752128601,
      "step": 600
    },
    {
      "epoch": 2.398406374501992,
      "grad_norm": 0.4691190719604492,
      "learning_rate": 2.2101549287686384e-06,
      "loss": 1.0322167873382568,
      "step": 602
    },
    {
      "epoch": 2.406374501992032,
      "grad_norm": 0.20021410286426544,
      "learning_rate": 2.195765696907745e-06,
      "loss": 0.5854684710502625,
      "step": 604
    },
    {
      "epoch": 2.414342629482072,
      "grad_norm": 1.7279207706451416,
      "learning_rate": 2.1814005566535303e-06,
      "loss": 0.24664445221424103,
      "step": 606
    },
    {
      "epoch": 2.4223107569721116,
      "grad_norm": 0.14960403740406036,
      "learning_rate": 2.167060132433798e-06,
      "loss": 0.5261163115501404,
      "step": 608
    },
    {
      "epoch": 2.4302788844621515,
      "grad_norm": 0.20952092111110687,
      "learning_rate": 2.1527450476019895e-06,
      "loss": 0.9457234740257263,
      "step": 610
    },
    {
      "epoch": 2.4382470119521913,
      "grad_norm": 0.7471716403961182,
      "learning_rate": 2.138455924410085e-06,
      "loss": 0.7355945706367493,
      "step": 612
    },
    {
      "epoch": 2.446215139442231,
      "grad_norm": 0.639157772064209,
      "learning_rate": 2.1241933839815606e-06,
      "loss": 0.6025002598762512,
      "step": 614
    },
    {
      "epoch": 2.454183266932271,
      "grad_norm": 0.6259835362434387,
      "learning_rate": 2.109958046284382e-06,
      "loss": 0.9215922951698303,
      "step": 616
    },
    {
      "epoch": 2.462151394422311,
      "grad_norm": 0.8553691506385803,
      "learning_rate": 2.0957505301040604e-06,
      "loss": 0.7287384271621704,
      "step": 618
    },
    {
      "epoch": 2.4701195219123506,
      "grad_norm": 0.8121323585510254,
      "learning_rate": 2.081571453016756e-06,
      "loss": 0.8274771571159363,
      "step": 620
    },
    {
      "epoch": 2.4780876494023905,
      "grad_norm": 0.76878422498703,
      "learning_rate": 2.067421431362427e-06,
      "loss": 0.8371984958648682,
      "step": 622
    },
    {
      "epoch": 2.4860557768924303,
      "grad_norm": 1.1341300010681152,
      "learning_rate": 2.053301080218044e-06,
      "loss": 0.8069764375686646,
      "step": 624
    },
    {
      "epoch": 2.49402390438247,
      "grad_norm": 0.6942153573036194,
      "learning_rate": 2.039211013370853e-06,
      "loss": 0.6599969863891602,
      "step": 626
    },
    {
      "epoch": 2.50199203187251,
      "grad_norm": 2.4954700469970703,
      "learning_rate": 2.025151843291693e-06,
      "loss": 1.010385274887085,
      "step": 628
    },
    {
      "epoch": 2.50996015936255,
      "grad_norm": 0.49280789494514465,
      "learning_rate": 2.0111241811083733e-06,
      "loss": 0.5018163919448853,
      "step": 630
    },
    {
      "epoch": 2.5179282868525896,
      "grad_norm": 0.8095399141311646,
      "learning_rate": 1.9971286365791092e-06,
      "loss": 0.4755622446537018,
      "step": 632
    },
    {
      "epoch": 2.5258964143426295,
      "grad_norm": 0.6170195937156677,
      "learning_rate": 1.9831658180660197e-06,
      "loss": 1.02364981174469,
      "step": 634
    },
    {
      "epoch": 2.5338645418326693,
      "grad_norm": 0.7571488618850708,
      "learning_rate": 1.9692363325086774e-06,
      "loss": 0.9616248607635498,
      "step": 636
    },
    {
      "epoch": 2.541832669322709,
      "grad_norm": 4.147839546203613,
      "learning_rate": 1.95534078539773e-06,
      "loss": 0.990222692489624,
      "step": 638
    },
    {
      "epoch": 2.549800796812749,
      "grad_norm": 0.18889543414115906,
      "learning_rate": 1.9414797807485796e-06,
      "loss": 0.7146550416946411,
      "step": 640
    },
    {
      "epoch": 2.557768924302789,
      "grad_norm": 0.6064460277557373,
      "learning_rate": 1.9276539210751295e-06,
      "loss": 1.0073672533035278,
      "step": 642
    },
    {
      "epoch": 2.5657370517928286,
      "grad_norm": 0.39991745352745056,
      "learning_rate": 1.9138638073635886e-06,
      "loss": 1.099884271621704,
      "step": 644
    },
    {
      "epoch": 2.5737051792828685,
      "grad_norm": 0.4935781955718994,
      "learning_rate": 1.90011003904635e-06,
      "loss": 0.566713273525238,
      "step": 646
    },
    {
      "epoch": 2.5816733067729083,
      "grad_norm": 0.7958781123161316,
      "learning_rate": 1.886393213975939e-06,
      "loss": 1.0329523086547852,
      "step": 648
    },
    {
      "epoch": 2.589641434262948,
      "grad_norm": 0.6274623274803162,
      "learning_rate": 1.8727139283990203e-06,
      "loss": 0.8009387254714966,
      "step": 650
    },
    {
      "epoch": 2.597609561752988,
      "grad_norm": 0.8008977174758911,
      "learning_rate": 1.8590727769304789e-06,
      "loss": 0.8522580862045288,
      "step": 652
    },
    {
      "epoch": 2.605577689243028,
      "grad_norm": 1.0197432041168213,
      "learning_rate": 1.8454703525275814e-06,
      "loss": 0.6617482304573059,
      "step": 654
    },
    {
      "epoch": 2.6135458167330676,
      "grad_norm": 0.4137818217277527,
      "learning_rate": 1.8319072464641912e-06,
      "loss": 0.3659578263759613,
      "step": 656
    },
    {
      "epoch": 2.6215139442231075,
      "grad_norm": 0.7393137812614441,
      "learning_rate": 1.8183840483050741e-06,
      "loss": 0.5374277234077454,
      "step": 658
    },
    {
      "epoch": 2.6294820717131473,
      "grad_norm": 0.4037087857723236,
      "learning_rate": 1.8049013458802667e-06,
      "loss": 0.9020153284072876,
      "step": 660
    },
    {
      "epoch": 2.637450199203187,
      "grad_norm": 1.1154837608337402,
      "learning_rate": 1.7914597252595267e-06,
      "loss": 1.0864851474761963,
      "step": 662
    },
    {
      "epoch": 2.645418326693227,
      "grad_norm": 0.5857207179069519,
      "learning_rate": 1.7780597707268582e-06,
      "loss": 0.6059953570365906,
      "step": 664
    },
    {
      "epoch": 2.653386454183267,
      "grad_norm": 1.0118664503097534,
      "learning_rate": 1.7647020647551105e-06,
      "loss": 0.9303960800170898,
      "step": 666
    },
    {
      "epoch": 2.6613545816733066,
      "grad_norm": 0.17573826014995575,
      "learning_rate": 1.7513871879806605e-06,
      "loss": 0.5772197246551514,
      "step": 668
    },
    {
      "epoch": 2.6693227091633465,
      "grad_norm": 0.6035557985305786,
      "learning_rate": 1.7381157191781766e-06,
      "loss": 0.7535842061042786,
      "step": 670
    },
    {
      "epoch": 2.6772908366533863,
      "grad_norm": 0.19330205023288727,
      "learning_rate": 1.7248882352354571e-06,
      "loss": 0.3687368929386139,
      "step": 672
    },
    {
      "epoch": 2.685258964143426,
      "grad_norm": 1.1029802560806274,
      "learning_rate": 1.7117053111283528e-06,
      "loss": 1.0314401388168335,
      "step": 674
    },
    {
      "epoch": 2.6932270916334664,
      "grad_norm": 1.0480084419250488,
      "learning_rate": 1.6985675198957752e-06,
      "loss": 0.6864706873893738,
      "step": 676
    },
    {
      "epoch": 2.7011952191235062,
      "grad_norm": 0.9156056642532349,
      "learning_rate": 1.6854754326147916e-06,
      "loss": 0.6051532030105591,
      "step": 678
    },
    {
      "epoch": 2.709163346613546,
      "grad_norm": 0.4044283330440521,
      "learning_rate": 1.6724296183757925e-06,
      "loss": 1.0206609964370728,
      "step": 680
    },
    {
      "epoch": 2.717131474103586,
      "grad_norm": 0.7035194039344788,
      "learning_rate": 1.6594306442577629e-06,
      "loss": 1.050747275352478,
      "step": 682
    },
    {
      "epoch": 2.7250996015936257,
      "grad_norm": 0.4720914959907532,
      "learning_rate": 1.6464790753036298e-06,
      "loss": 0.7009389400482178,
      "step": 684
    },
    {
      "epoch": 2.7330677290836656,
      "grad_norm": 1.1745597124099731,
      "learning_rate": 1.6335754744956935e-06,
      "loss": 0.4856918752193451,
      "step": 686
    },
    {
      "epoch": 2.7410358565737054,
      "grad_norm": 0.5046957731246948,
      "learning_rate": 1.6207204027311685e-06,
      "loss": 1.0343010425567627,
      "step": 688
    },
    {
      "epoch": 2.7490039840637452,
      "grad_norm": 0.45445072650909424,
      "learning_rate": 1.6079144187977925e-06,
      "loss": 0.5356013774871826,
      "step": 690
    },
    {
      "epoch": 2.756972111553785,
      "grad_norm": 0.4808848798274994,
      "learning_rate": 1.595158079349542e-06,
      "loss": 1.1045536994934082,
      "step": 692
    },
    {
      "epoch": 2.764940239043825,
      "grad_norm": 0.6008504629135132,
      "learning_rate": 1.5824519388824311e-06,
      "loss": 0.8576187491416931,
      "step": 694
    },
    {
      "epoch": 2.7729083665338647,
      "grad_norm": 0.7451703548431396,
      "learning_rate": 1.5697965497104117e-06,
      "loss": 0.6828925609588623,
      "step": 696
    },
    {
      "epoch": 2.7808764940239046,
      "grad_norm": 0.4513993561267853,
      "learning_rate": 1.557192461941367e-06,
      "loss": 0.8965566158294678,
      "step": 698
    },
    {
      "epoch": 2.7888446215139444,
      "grad_norm": 0.3922881484031677,
      "learning_rate": 1.5446402234531934e-06,
      "loss": 1.051708459854126,
      "step": 700
    },
    {
      "epoch": 2.7968127490039842,
      "grad_norm": 0.2590172588825226,
      "learning_rate": 1.5321403798699924e-06,
      "loss": 0.6456558704376221,
      "step": 702
    },
    {
      "epoch": 2.804780876494024,
      "grad_norm": 0.7220768332481384,
      "learning_rate": 1.5196934745383456e-06,
      "loss": 1.2301793098449707,
      "step": 704
    },
    {
      "epoch": 2.812749003984064,
      "grad_norm": 0.3174591064453125,
      "learning_rate": 1.5073000485037044e-06,
      "loss": 1.2239747047424316,
      "step": 706
    },
    {
      "epoch": 2.8207171314741037,
      "grad_norm": 0.9559182524681091,
      "learning_rate": 1.4949606404868638e-06,
      "loss": 1.1182080507278442,
      "step": 708
    },
    {
      "epoch": 2.8286852589641436,
      "grad_norm": 0.1768503040075302,
      "learning_rate": 1.4826757868605519e-06,
      "loss": 0.5365989208221436,
      "step": 710
    },
    {
      "epoch": 2.8366533864541834,
      "grad_norm": 0.44148382544517517,
      "learning_rate": 1.4704460216261125e-06,
      "loss": 1.064496397972107,
      "step": 712
    },
    {
      "epoch": 2.8446215139442232,
      "grad_norm": 1.0274395942687988,
      "learning_rate": 1.458271876390287e-06,
      "loss": 0.6138657927513123,
      "step": 714
    },
    {
      "epoch": 2.852589641434263,
      "grad_norm": 0.4575527608394623,
      "learning_rate": 1.446153880342117e-06,
      "loss": 1.032732605934143,
      "step": 716
    },
    {
      "epoch": 2.860557768924303,
      "grad_norm": 0.9148539304733276,
      "learning_rate": 1.4340925602299337e-06,
      "loss": 0.6516750454902649,
      "step": 718
    },
    {
      "epoch": 2.8685258964143427,
      "grad_norm": 0.5192955732345581,
      "learning_rate": 1.4220884403384656e-06,
      "loss": 0.5977855324745178,
      "step": 720
    },
    {
      "epoch": 2.8764940239043826,
      "grad_norm": 0.9483743906021118,
      "learning_rate": 1.4101420424660433e-06,
      "loss": 0.7096575498580933,
      "step": 722
    },
    {
      "epoch": 2.8844621513944224,
      "grad_norm": 0.4027494192123413,
      "learning_rate": 1.3982538859019219e-06,
      "loss": 0.6428722739219666,
      "step": 724
    },
    {
      "epoch": 2.8924302788844622,
      "grad_norm": 0.4757281541824341,
      "learning_rate": 1.3864244874037096e-06,
      "loss": 1.0020397901535034,
      "step": 726
    },
    {
      "epoch": 2.900398406374502,
      "grad_norm": 0.5693173408508301,
      "learning_rate": 1.3746543611749007e-06,
      "loss": 0.5779163241386414,
      "step": 728
    },
    {
      "epoch": 2.908366533864542,
      "grad_norm": 0.41845542192459106,
      "learning_rate": 1.3629440188425285e-06,
      "loss": 0.9897056818008423,
      "step": 730
    },
    {
      "epoch": 2.9163346613545817,
      "grad_norm": 0.7383747696876526,
      "learning_rate": 1.351293969434921e-06,
      "loss": 0.3598988652229309,
      "step": 732
    },
    {
      "epoch": 2.9243027888446216,
      "grad_norm": 0.7787171602249146,
      "learning_rate": 1.3397047193595809e-06,
      "loss": 1.0502912998199463,
      "step": 734
    },
    {
      "epoch": 2.9322709163346614,
      "grad_norm": 1.4530410766601562,
      "learning_rate": 1.3281767723811651e-06,
      "loss": 0.42645499110221863,
      "step": 736
    },
    {
      "epoch": 2.9402390438247012,
      "grad_norm": 1.4574962854385376,
      "learning_rate": 1.3167106295995923e-06,
      "loss": 0.9167231321334839,
      "step": 738
    },
    {
      "epoch": 2.948207171314741,
      "grad_norm": 0.41042253375053406,
      "learning_rate": 1.3053067894282624e-06,
      "loss": 1.158491849899292,
      "step": 740
    },
    {
      "epoch": 2.956175298804781,
      "grad_norm": 2.207376718521118,
      "learning_rate": 1.2939657475723844e-06,
      "loss": 0.4148605763912201,
      "step": 742
    },
    {
      "epoch": 2.9641434262948207,
      "grad_norm": 0.4657469093799591,
      "learning_rate": 1.2826879970074346e-06,
      "loss": 0.8907448649406433,
      "step": 744
    },
    {
      "epoch": 2.9721115537848606,
      "grad_norm": 0.6856690645217896,
      "learning_rate": 1.271474027957727e-06,
      "loss": 0.8996275663375854,
      "step": 746
    },
    {
      "epoch": 2.9800796812749004,
      "grad_norm": 0.85634845495224,
      "learning_rate": 1.2603243278751043e-06,
      "loss": 0.5467954277992249,
      "step": 748
    },
    {
      "epoch": 2.9880478087649402,
      "grad_norm": 0.8076357841491699,
      "learning_rate": 1.249239381417746e-06,
      "loss": 0.14352411031723022,
      "step": 750
    },
    {
      "epoch": 2.99601593625498,
      "grad_norm": 0.46906089782714844,
      "learning_rate": 1.238219670429104e-06,
      "loss": 1.016222357749939,
      "step": 752
    },
    {
      "epoch": 3.00398406374502,
      "grad_norm": 0.42473721504211426,
      "learning_rate": 1.2272656739169589e-06,
      "loss": 0.5547373294830322,
      "step": 754
    },
    {
      "epoch": 3.0119521912350598,
      "grad_norm": 1.4837422370910645,
      "learning_rate": 1.2163778680325948e-06,
      "loss": 0.6226297616958618,
      "step": 756
    },
    {
      "epoch": 3.0199203187250996,
      "grad_norm": 0.544873058795929,
      "learning_rate": 1.2055567260501055e-06,
      "loss": 0.33402544260025024,
      "step": 758
    },
    {
      "epoch": 3.0278884462151394,
      "grad_norm": 0.6402859687805176,
      "learning_rate": 1.1948027183458182e-06,
      "loss": 0.6753387451171875,
      "step": 760
    },
    {
      "epoch": 3.0358565737051793,
      "grad_norm": 0.5948603749275208,
      "learning_rate": 1.18411631237785e-06,
      "loss": 0.48431453108787537,
      "step": 762
    },
    {
      "epoch": 3.043824701195219,
      "grad_norm": 0.6400764584541321,
      "learning_rate": 1.173497972665788e-06,
      "loss": 0.9738553166389465,
      "step": 764
    },
    {
      "epoch": 3.051792828685259,
      "grad_norm": 0.6048662662506104,
      "learning_rate": 1.1629481607704964e-06,
      "loss": 0.16977912187576294,
      "step": 766
    },
    {
      "epoch": 3.0597609561752988,
      "grad_norm": 0.4570543169975281,
      "learning_rate": 1.1524673352740543e-06,
      "loss": 0.61164790391922,
      "step": 768
    },
    {
      "epoch": 3.0677290836653386,
      "grad_norm": 0.4360712170600891,
      "learning_rate": 1.1420559517598206e-06,
      "loss": 0.820690929889679,
      "step": 770
    },
    {
      "epoch": 3.0756972111553784,
      "grad_norm": 0.5304065942764282,
      "learning_rate": 1.13171446279263e-06,
      "loss": 0.9779512882232666,
      "step": 772
    },
    {
      "epoch": 3.0836653386454183,
      "grad_norm": 1.6234406232833862,
      "learning_rate": 1.1214433178991258e-06,
      "loss": 0.9019178152084351,
      "step": 774
    },
    {
      "epoch": 3.091633466135458,
      "grad_norm": 0.10643471032381058,
      "learning_rate": 1.111242963548212e-06,
      "loss": 0.5657032132148743,
      "step": 776
    },
    {
      "epoch": 3.099601593625498,
      "grad_norm": 0.6068316698074341,
      "learning_rate": 1.1011138431316538e-06,
      "loss": 0.6061086058616638,
      "step": 778
    },
    {
      "epoch": 3.1075697211155378,
      "grad_norm": 0.45254701375961304,
      "learning_rate": 1.0910563969447969e-06,
      "loss": 0.9793358445167542,
      "step": 780
    },
    {
      "epoch": 3.1155378486055776,
      "grad_norm": 0.18679897487163544,
      "learning_rate": 1.0810710621674355e-06,
      "loss": 0.3254126012325287,
      "step": 782
    },
    {
      "epoch": 3.1235059760956174,
      "grad_norm": 0.4458552896976471,
      "learning_rate": 1.0711582728448034e-06,
      "loss": 1.0158764123916626,
      "step": 784
    },
    {
      "epoch": 3.1314741035856573,
      "grad_norm": 0.43274039030075073,
      "learning_rate": 1.0613184598687088e-06,
      "loss": 0.7879319787025452,
      "step": 786
    },
    {
      "epoch": 3.139442231075697,
      "grad_norm": 1.74847412109375,
      "learning_rate": 1.0515520509588076e-06,
      "loss": 0.6197874546051025,
      "step": 788
    },
    {
      "epoch": 3.147410358565737,
      "grad_norm": 0.28513556718826294,
      "learning_rate": 1.0418594706440022e-06,
      "loss": 0.5337375998497009,
      "step": 790
    },
    {
      "epoch": 3.1553784860557768,
      "grad_norm": 1.270583987236023,
      "learning_rate": 1.0322411402439982e-06,
      "loss": 0.742657482624054,
      "step": 792
    },
    {
      "epoch": 3.1633466135458166,
      "grad_norm": 0.0665946677327156,
      "learning_rate": 1.0226974778509846e-06,
      "loss": 0.22772325575351715,
      "step": 794
    },
    {
      "epoch": 3.1713147410358564,
      "grad_norm": 0.7945287823677063,
      "learning_rate": 1.0132288983114611e-06,
      "loss": 0.6697213649749756,
      "step": 796
    },
    {
      "epoch": 3.1792828685258963,
      "grad_norm": 0.41244545578956604,
      "learning_rate": 1.003835813208206e-06,
      "loss": 0.5557263493537903,
      "step": 798
    },
    {
      "epoch": 3.187250996015936,
      "grad_norm": 1.4478341341018677,
      "learning_rate": 9.945186308423832e-07,
      "loss": 0.8352447748184204,
      "step": 800
    },
    {
      "epoch": 3.195219123505976,
      "grad_norm": 0.42246878147125244,
      "learning_rate": 9.852777562157983e-07,
      "loss": 0.7306307554244995,
      "step": 802
    },
    {
      "epoch": 3.2031872509960158,
      "grad_norm": 1.6348930597305298,
      "learning_rate": 9.7611359101329e-07,
      "loss": 0.6521217823028564,
      "step": 804
    },
    {
      "epoch": 3.2111553784860556,
      "grad_norm": 1.2540473937988281,
      "learning_rate": 9.670265335852725e-07,
      "loss": 1.1119095087051392,
      "step": 806
    },
    {
      "epoch": 3.2191235059760954,
      "grad_norm": 0.9749636054039001,
      "learning_rate": 9.580169789304152e-07,
      "loss": 0.4577837884426117,
      "step": 808
    },
    {
      "epoch": 3.2270916334661353,
      "grad_norm": 0.43074101209640503,
      "learning_rate": 9.490853186784801e-07,
      "loss": 0.6960765719413757,
      "step": 810
    },
    {
      "epoch": 3.235059760956175,
      "grad_norm": 0.9670475721359253,
      "learning_rate": 9.402319410732911e-07,
      "loss": 0.3812284767627716,
      "step": 812
    },
    {
      "epoch": 3.243027888446215,
      "grad_norm": 0.6557716131210327,
      "learning_rate": 9.314572309558617e-07,
      "loss": 0.8401023745536804,
      "step": 814
    },
    {
      "epoch": 3.2509960159362548,
      "grad_norm": 1.8736748695373535,
      "learning_rate": 9.227615697476674e-07,
      "loss": 0.3678154945373535,
      "step": 816
    },
    {
      "epoch": 3.2589641434262946,
      "grad_norm": 0.4513351321220398,
      "learning_rate": 9.141453354340619e-07,
      "loss": 0.9667196869850159,
      "step": 818
    },
    {
      "epoch": 3.2669322709163344,
      "grad_norm": 1.6300173997879028,
      "learning_rate": 9.056089025478495e-07,
      "loss": 0.49229466915130615,
      "step": 820
    },
    {
      "epoch": 3.2749003984063743,
      "grad_norm": 2.0868749618530273,
      "learning_rate": 8.971526421530067e-07,
      "loss": 0.7760501503944397,
      "step": 822
    },
    {
      "epoch": 3.2828685258964145,
      "grad_norm": 0.8033971190452576,
      "learning_rate": 8.887769218285469e-07,
      "loss": 0.5642234086990356,
      "step": 824
    },
    {
      "epoch": 3.2908366533864544,
      "grad_norm": 0.4739183187484741,
      "learning_rate": 8.804821056525498e-07,
      "loss": 0.9728425145149231,
      "step": 826
    },
    {
      "epoch": 3.298804780876494,
      "grad_norm": 0.5138052701950073,
      "learning_rate": 8.722685541863284e-07,
      "loss": 0.9591096043586731,
      "step": 828
    },
    {
      "epoch": 3.306772908366534,
      "grad_norm": 0.6322988867759705,
      "learning_rate": 8.641366244587612e-07,
      "loss": 0.8918768167495728,
      "step": 830
    },
    {
      "epoch": 3.314741035856574,
      "grad_norm": 1.0239696502685547,
      "learning_rate": 8.560866699507703e-07,
      "loss": 0.48340940475463867,
      "step": 832
    },
    {
      "epoch": 3.3227091633466137,
      "grad_norm": 0.9379016757011414,
      "learning_rate": 8.481190405799569e-07,
      "loss": 0.3693496882915497,
      "step": 834
    },
    {
      "epoch": 3.3306772908366535,
      "grad_norm": 1.4198274612426758,
      "learning_rate": 8.402340826853908e-07,
      "loss": 0.6240131855010986,
      "step": 836
    },
    {
      "epoch": 3.3386454183266934,
      "grad_norm": 0.9080134630203247,
      "learning_rate": 8.324321390125548e-07,
      "loss": 0.7740859389305115,
      "step": 838
    },
    {
      "epoch": 3.346613545816733,
      "grad_norm": 1.1049258708953857,
      "learning_rate": 8.247135486984487e-07,
      "loss": 0.24195581674575806,
      "step": 840
    },
    {
      "epoch": 3.354581673306773,
      "grad_norm": 0.8039106130599976,
      "learning_rate": 8.170786472568451e-07,
      "loss": 0.7066781520843506,
      "step": 842
    },
    {
      "epoch": 3.362549800796813,
      "grad_norm": 0.4682895839214325,
      "learning_rate": 8.095277665637061e-07,
      "loss": 0.679907500743866,
      "step": 844
    },
    {
      "epoch": 3.3705179282868527,
      "grad_norm": 0.5532957315444946,
      "learning_rate": 8.020612348427572e-07,
      "loss": 0.9976988434791565,
      "step": 846
    },
    {
      "epoch": 3.3784860557768925,
      "grad_norm": 0.46046438813209534,
      "learning_rate": 7.946793766512195e-07,
      "loss": 0.9377111792564392,
      "step": 848
    },
    {
      "epoch": 3.3864541832669324,
      "grad_norm": 2.080869197845459,
      "learning_rate": 7.873825128657034e-07,
      "loss": 1.0559757947921753,
      "step": 850
    },
    {
      "epoch": 3.394422310756972,
      "grad_norm": 0.4267537593841553,
      "learning_rate": 7.801709606682573e-07,
      "loss": 0.9420338273048401,
      "step": 852
    },
    {
      "epoch": 3.402390438247012,
      "grad_norm": 1.281989574432373,
      "learning_rate": 7.730450335325852e-07,
      "loss": 0.5549944639205933,
      "step": 854
    },
    {
      "epoch": 3.410358565737052,
      "grad_norm": 0.7911232113838196,
      "learning_rate": 7.660050412104153e-07,
      "loss": 0.9087567329406738,
      "step": 856
    },
    {
      "epoch": 3.4183266932270917,
      "grad_norm": 0.3308275640010834,
      "learning_rate": 7.5905128971804e-07,
      "loss": 1.0682170391082764,
      "step": 858
    },
    {
      "epoch": 3.4262948207171315,
      "grad_norm": 0.691639244556427,
      "learning_rate": 7.521840813230098e-07,
      "loss": 1.0168159008026123,
      "step": 860
    },
    {
      "epoch": 3.4342629482071714,
      "grad_norm": 0.3761932849884033,
      "learning_rate": 7.454037145309986e-07,
      "loss": 0.6987581253051758,
      "step": 862
    },
    {
      "epoch": 3.442231075697211,
      "grad_norm": 2.1681807041168213,
      "learning_rate": 7.38710484072826e-07,
      "loss": 0.550249457359314,
      "step": 864
    },
    {
      "epoch": 3.450199203187251,
      "grad_norm": 0.44106608629226685,
      "learning_rate": 7.321046808916427e-07,
      "loss": 0.9029756784439087,
      "step": 866
    },
    {
      "epoch": 3.458167330677291,
      "grad_norm": 0.5243621468544006,
      "learning_rate": 7.255865921302902e-07,
      "loss": 0.7247702479362488,
      "step": 868
    },
    {
      "epoch": 3.4661354581673307,
      "grad_norm": 0.3829691410064697,
      "learning_rate": 7.191565011188149e-07,
      "loss": 0.5426479578018188,
      "step": 870
    },
    {
      "epoch": 3.4741035856573705,
      "grad_norm": 0.12485034018754959,
      "learning_rate": 7.128146873621533e-07,
      "loss": 0.5784074068069458,
      "step": 872
    },
    {
      "epoch": 3.4820717131474104,
      "grad_norm": 0.2735823392868042,
      "learning_rate": 7.065614265279819e-07,
      "loss": 0.7004212141036987,
      "step": 874
    },
    {
      "epoch": 3.49003984063745,
      "grad_norm": 0.6446283459663391,
      "learning_rate": 7.00396990434734e-07,
      "loss": 0.5554342269897461,
      "step": 876
    },
    {
      "epoch": 3.49800796812749,
      "grad_norm": 0.7106437683105469,
      "learning_rate": 6.943216470397872e-07,
      "loss": 0.543057382106781,
      "step": 878
    },
    {
      "epoch": 3.50597609561753,
      "grad_norm": 0.31151682138442993,
      "learning_rate": 6.883356604278116e-07,
      "loss": 0.41106748580932617,
      "step": 880
    },
    {
      "epoch": 3.5139442231075697,
      "grad_norm": 0.5321460366249084,
      "learning_rate": 6.824392907992938e-07,
      "loss": 0.665117084980011,
      "step": 882
    },
    {
      "epoch": 3.5219123505976095,
      "grad_norm": 0.9792880415916443,
      "learning_rate": 6.76632794459224e-07,
      "loss": 0.2560465335845947,
      "step": 884
    },
    {
      "epoch": 3.5298804780876494,
      "grad_norm": 0.8835018873214722,
      "learning_rate": 6.709164238059568e-07,
      "loss": 0.7860435247421265,
      "step": 886
    },
    {
      "epoch": 3.537848605577689,
      "grad_norm": 0.8839598894119263,
      "learning_rate": 6.652904273202385e-07,
      "loss": 0.9369090795516968,
      "step": 888
    },
    {
      "epoch": 3.545816733067729,
      "grad_norm": 0.5607488751411438,
      "learning_rate": 6.597550495544071e-07,
      "loss": 0.49607735872268677,
      "step": 890
    },
    {
      "epoch": 3.553784860557769,
      "grad_norm": 1.187895655632019,
      "learning_rate": 6.54310531121762e-07,
      "loss": 0.9376204013824463,
      "step": 892
    },
    {
      "epoch": 3.5617529880478087,
      "grad_norm": 0.5547093749046326,
      "learning_rate": 6.489571086861027e-07,
      "loss": 1.0408369302749634,
      "step": 894
    },
    {
      "epoch": 3.5697211155378485,
      "grad_norm": 0.2572667598724365,
      "learning_rate": 6.436950149514448e-07,
      "loss": 0.23081542551517487,
      "step": 896
    },
    {
      "epoch": 3.5776892430278884,
      "grad_norm": 0.39407864212989807,
      "learning_rate": 6.385244786519036e-07,
      "loss": 0.6313282251358032,
      "step": 898
    },
    {
      "epoch": 3.585657370517928,
      "grad_norm": 2.497715950012207,
      "learning_rate": 6.334457245417491e-07,
      "loss": 0.23165220022201538,
      "step": 900
    },
    {
      "epoch": 3.593625498007968,
      "grad_norm": 0.5764774084091187,
      "learning_rate": 6.284589733856403e-07,
      "loss": 0.11190037429332733,
      "step": 902
    },
    {
      "epoch": 3.601593625498008,
      "grad_norm": 0.6453854441642761,
      "learning_rate": 6.235644419490253e-07,
      "loss": 0.7609789371490479,
      "step": 904
    },
    {
      "epoch": 3.6095617529880477,
      "grad_norm": 0.8670569062232971,
      "learning_rate": 6.18762342988722e-07,
      "loss": 0.6526831388473511,
      "step": 906
    },
    {
      "epoch": 3.6175298804780875,
      "grad_norm": 0.5277704000473022,
      "learning_rate": 6.140528852436666e-07,
      "loss": 0.9082297086715698,
      "step": 908
    },
    {
      "epoch": 3.6254980079681274,
      "grad_norm": 0.8381524682044983,
      "learning_rate": 6.094362734258438e-07,
      "loss": 0.8484107851982117,
      "step": 910
    },
    {
      "epoch": 3.633466135458167,
      "grad_norm": 0.2789282202720642,
      "learning_rate": 6.049127082113853e-07,
      "loss": 0.04587188735604286,
      "step": 912
    },
    {
      "epoch": 3.641434262948207,
      "grad_norm": 0.7530278563499451,
      "learning_rate": 6.004823862318475e-07,
      "loss": 0.7911416888237,
      "step": 914
    },
    {
      "epoch": 3.649402390438247,
      "grad_norm": 2.177725315093994,
      "learning_rate": 5.961455000656656e-07,
      "loss": 0.8250769376754761,
      "step": 916
    },
    {
      "epoch": 3.6573705179282867,
      "grad_norm": 4.118772029876709,
      "learning_rate": 5.919022382297808e-07,
      "loss": 0.19510747492313385,
      "step": 918
    },
    {
      "epoch": 3.6653386454183265,
      "grad_norm": 0.9756109714508057,
      "learning_rate": 5.877527851714479e-07,
      "loss": 0.9553953409194946,
      "step": 920
    },
    {
      "epoch": 3.6733067729083664,
      "grad_norm": 0.4340531826019287,
      "learning_rate": 5.836973212602144e-07,
      "loss": 0.17423762381076813,
      "step": 922
    },
    {
      "epoch": 3.681274900398406,
      "grad_norm": 1.1748831272125244,
      "learning_rate": 5.797360227800832e-07,
      "loss": 0.9892681837081909,
      "step": 924
    },
    {
      "epoch": 3.6892430278884465,
      "grad_norm": 5.325054168701172,
      "learning_rate": 5.758690619218493e-07,
      "loss": 0.31605854630470276,
      "step": 926
    },
    {
      "epoch": 3.6972111553784863,
      "grad_norm": 0.5281583070755005,
      "learning_rate": 5.720966067756131e-07,
      "loss": 0.5505653619766235,
      "step": 928
    },
    {
      "epoch": 3.705179282868526,
      "grad_norm": 0.3937387764453888,
      "learning_rate": 5.684188213234763e-07,
      "loss": 0.8612691164016724,
      "step": 930
    },
    {
      "epoch": 3.713147410358566,
      "grad_norm": 0.5615203976631165,
      "learning_rate": 5.648358654324119e-07,
      "loss": 0.9553990364074707,
      "step": 932
    },
    {
      "epoch": 3.721115537848606,
      "grad_norm": 0.739287793636322,
      "learning_rate": 5.613478948473166e-07,
      "loss": 0.5119466185569763,
      "step": 934
    },
    {
      "epoch": 3.7290836653386457,
      "grad_norm": 0.4925720989704132,
      "learning_rate": 5.579550611842395e-07,
      "loss": 0.8857870697975159,
      "step": 936
    },
    {
      "epoch": 3.7370517928286855,
      "grad_norm": 1.493027687072754,
      "learning_rate": 5.546575119237923e-07,
      "loss": 0.6420094966888428,
      "step": 938
    },
    {
      "epoch": 3.7450199203187253,
      "grad_norm": 0.40828871726989746,
      "learning_rate": 5.514553904047394e-07,
      "loss": 0.8009927272796631,
      "step": 940
    },
    {
      "epoch": 3.752988047808765,
      "grad_norm": 1.660335659980774,
      "learning_rate": 5.483488358177647e-07,
      "loss": 0.7522887587547302,
      "step": 942
    },
    {
      "epoch": 3.760956175298805,
      "grad_norm": 0.3406866490840912,
      "learning_rate": 5.453379831994238e-07,
      "loss": 0.5255100131034851,
      "step": 944
    },
    {
      "epoch": 3.768924302788845,
      "grad_norm": 0.7313100099563599,
      "learning_rate": 5.424229634262735e-07,
      "loss": 0.9878066182136536,
      "step": 946
    },
    {
      "epoch": 3.7768924302788847,
      "grad_norm": 0.7185702919960022,
      "learning_rate": 5.396039032091826e-07,
      "loss": 0.5519812703132629,
      "step": 948
    },
    {
      "epoch": 3.7848605577689245,
      "grad_norm": 0.3306468427181244,
      "learning_rate": 5.368809250878228e-07,
      "loss": 0.9590651392936707,
      "step": 950
    },
    {
      "epoch": 3.7928286852589643,
      "grad_norm": 0.4941675662994385,
      "learning_rate": 5.342541474253438e-07,
      "loss": 0.9663379192352295,
      "step": 952
    },
    {
      "epoch": 3.800796812749004,
      "grad_norm": 0.5179746150970459,
      "learning_rate": 5.317236844032285e-07,
      "loss": 0.6215814352035522,
      "step": 954
    },
    {
      "epoch": 3.808764940239044,
      "grad_norm": 0.9556832909584045,
      "learning_rate": 5.292896460163282e-07,
      "loss": 0.7942197322845459,
      "step": 956
    },
    {
      "epoch": 3.816733067729084,
      "grad_norm": 3.1209332942962646,
      "learning_rate": 5.269521380680823e-07,
      "loss": 0.5784275531768799,
      "step": 958
    },
    {
      "epoch": 3.8247011952191237,
      "grad_norm": 0.8426687121391296,
      "learning_rate": 5.247112621659186e-07,
      "loss": 0.7202901840209961,
      "step": 960
    },
    {
      "epoch": 3.8326693227091635,
      "grad_norm": 0.24685674905776978,
      "learning_rate": 5.225671157168378e-07,
      "loss": 0.5644170641899109,
      "step": 962
    },
    {
      "epoch": 3.8406374501992033,
      "grad_norm": 0.6406257152557373,
      "learning_rate": 5.205197919231773e-07,
      "loss": 0.9743782877922058,
      "step": 964
    },
    {
      "epoch": 3.848605577689243,
      "grad_norm": 0.848483145236969,
      "learning_rate": 5.185693797785624e-07,
      "loss": 0.3893330693244934,
      "step": 966
    },
    {
      "epoch": 3.856573705179283,
      "grad_norm": 0.6470706462860107,
      "learning_rate": 5.167159640640358e-07,
      "loss": 0.567273736000061,
      "step": 968
    },
    {
      "epoch": 3.864541832669323,
      "grad_norm": 1.1594817638397217,
      "learning_rate": 5.149596253443733e-07,
      "loss": 0.7379060983657837,
      "step": 970
    },
    {
      "epoch": 3.8725099601593627,
      "grad_norm": 0.20746168494224548,
      "learning_rate": 5.133004399645818e-07,
      "loss": 0.5668442249298096,
      "step": 972
    },
    {
      "epoch": 3.8804780876494025,
      "grad_norm": 0.29715651273727417,
      "learning_rate": 5.117384800465806e-07,
      "loss": 0.45394814014434814,
      "step": 974
    },
    {
      "epoch": 3.8884462151394423,
      "grad_norm": 0.9812098145484924,
      "learning_rate": 5.102738134860656e-07,
      "loss": 0.9694859981536865,
      "step": 976
    },
    {
      "epoch": 3.896414342629482,
      "grad_norm": 0.7796050906181335,
      "learning_rate": 5.089065039495598e-07,
      "loss": 0.5591210722923279,
      "step": 978
    },
    {
      "epoch": 3.904382470119522,
      "grad_norm": 0.22947020828723907,
      "learning_rate": 5.076366108716438e-07,
      "loss": 0.4146263599395752,
      "step": 980
    },
    {
      "epoch": 3.912350597609562,
      "grad_norm": 0.7211007475852966,
      "learning_rate": 5.064641894523737e-07,
      "loss": 0.48491913080215454,
      "step": 982
    },
    {
      "epoch": 3.9203187250996017,
      "grad_norm": 0.4875924289226532,
      "learning_rate": 5.053892906548811e-07,
      "loss": 0.9094846844673157,
      "step": 984
    },
    {
      "epoch": 3.9282868525896415,
      "grad_norm": 0.44494158029556274,
      "learning_rate": 5.044119612031576e-07,
      "loss": 0.14815448224544525,
      "step": 986
    },
    {
      "epoch": 3.9362549800796813,
      "grad_norm": 2.167361259460449,
      "learning_rate": 5.035322435800248e-07,
      "loss": 0.6271823048591614,
      "step": 988
    },
    {
      "epoch": 3.944223107569721,
      "grad_norm": 0.16184689104557037,
      "learning_rate": 5.027501760252857e-07,
      "loss": 0.5810338854789734,
      "step": 990
    },
    {
      "epoch": 3.952191235059761,
      "grad_norm": 1.3082406520843506,
      "learning_rate": 5.02065792534065e-07,
      "loss": 1.0052835941314697,
      "step": 992
    },
    {
      "epoch": 3.960159362549801,
      "grad_norm": 0.9668223261833191,
      "learning_rate": 5.014791228553294e-07,
      "loss": 0.24346432089805603,
      "step": 994
    },
    {
      "epoch": 3.9681274900398407,
      "grad_norm": 1.2428168058395386,
      "learning_rate": 5.00990192490595e-07,
      "loss": 0.8840433359146118,
      "step": 996
    },
    {
      "epoch": 3.9760956175298805,
      "grad_norm": 0.28113725781440735,
      "learning_rate": 5.005990226928196e-07,
      "loss": 0.5683450102806091,
      "step": 998
    },
    {
      "epoch": 3.9840637450199203,
      "grad_norm": 0.6847888231277466,
      "learning_rate": 5.003056304654774e-07,
      "loss": 0.4292658269405365,
      "step": 1000
    },
    {
      "epoch": 3.99203187250996,
      "grad_norm": 0.511762261390686,
      "learning_rate": 5.001100285618214e-07,
      "loss": 0.971055269241333,
      "step": 1002
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.2060747742652893,
      "learning_rate": 5.000122254843282e-07,
      "loss": 0.31422263383865356,
      "step": 1004
    },
    {
      "epoch": 4.0,
      "step": 1004,
      "total_flos": 3.519329211045118e+18,
      "train_loss": 0.8785166950206239,
      "train_runtime": 8976.4943,
      "train_samples_per_second": 6.711,
      "train_steps_per_second": 0.112
    }
  ],
  "logging_steps": 2,
  "max_steps": 1004,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 99999,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.519329211045118e+18,
  "train_batch_size": 3,
  "trial_name": null,
  "trial_params": null
}