9b-19 / trainer_state.json
furproxy's picture
Upload folder using huggingface_hub
75c3801 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 2619,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.002290950744558992,
"grad_norm": 0.9735844731330872,
"learning_rate": 7.633587786259542e-08,
"loss": 1.4360580444335938,
"step": 2
},
{
"epoch": 0.004581901489117984,
"grad_norm": 1.5835797786712646,
"learning_rate": 2.2900763358778629e-07,
"loss": 1.971843957901001,
"step": 4
},
{
"epoch": 0.006872852233676976,
"grad_norm": 0.8149327635765076,
"learning_rate": 3.8167938931297716e-07,
"loss": 1.8271703720092773,
"step": 6
},
{
"epoch": 0.009163802978235968,
"grad_norm": 2.2085700035095215,
"learning_rate": 5.34351145038168e-07,
"loss": 2.4666566848754883,
"step": 8
},
{
"epoch": 0.011454753722794959,
"grad_norm": 0.4170507788658142,
"learning_rate": 6.870229007633589e-07,
"loss": 1.2739520072937012,
"step": 10
},
{
"epoch": 0.013745704467353952,
"grad_norm": 0.33925074338912964,
"learning_rate": 8.396946564885497e-07,
"loss": 1.6276776790618896,
"step": 12
},
{
"epoch": 0.016036655211912942,
"grad_norm": 0.8046349883079529,
"learning_rate": 9.923664122137404e-07,
"loss": 1.938399076461792,
"step": 14
},
{
"epoch": 0.018327605956471937,
"grad_norm": 0.39774009585380554,
"learning_rate": 1.1450381679389313e-06,
"loss": 1.5784566402435303,
"step": 16
},
{
"epoch": 0.020618556701030927,
"grad_norm": 2.0479557514190674,
"learning_rate": 1.297709923664122e-06,
"loss": 2.207265615463257,
"step": 18
},
{
"epoch": 0.022909507445589918,
"grad_norm": 1.4070589542388916,
"learning_rate": 1.450381679389313e-06,
"loss": 2.08921480178833,
"step": 20
},
{
"epoch": 0.025200458190148912,
"grad_norm": 1.3663133382797241,
"learning_rate": 1.603053435114504e-06,
"loss": 1.5717980861663818,
"step": 22
},
{
"epoch": 0.027491408934707903,
"grad_norm": 0.5743522644042969,
"learning_rate": 1.7557251908396948e-06,
"loss": 1.9206372499465942,
"step": 24
},
{
"epoch": 0.029782359679266894,
"grad_norm": 0.5176920294761658,
"learning_rate": 1.908396946564886e-06,
"loss": 1.903019905090332,
"step": 26
},
{
"epoch": 0.032073310423825885,
"grad_norm": 0.5487824082374573,
"learning_rate": 2.0610687022900764e-06,
"loss": 1.724461555480957,
"step": 28
},
{
"epoch": 0.03436426116838488,
"grad_norm": 0.6357744932174683,
"learning_rate": 2.2137404580152674e-06,
"loss": 2.329512119293213,
"step": 30
},
{
"epoch": 0.03665521191294387,
"grad_norm": 0.5210757255554199,
"learning_rate": 2.3664122137404585e-06,
"loss": 1.7047933340072632,
"step": 32
},
{
"epoch": 0.038946162657502864,
"grad_norm": 0.45220088958740234,
"learning_rate": 2.5190839694656487e-06,
"loss": 1.8027178049087524,
"step": 34
},
{
"epoch": 0.041237113402061855,
"grad_norm": 0.6867555975914001,
"learning_rate": 2.67175572519084e-06,
"loss": 1.8687299489974976,
"step": 36
},
{
"epoch": 0.043528064146620846,
"grad_norm": 1.022306203842163,
"learning_rate": 2.824427480916031e-06,
"loss": 1.516510248184204,
"step": 38
},
{
"epoch": 0.045819014891179836,
"grad_norm": 0.4305284023284912,
"learning_rate": 2.9770992366412218e-06,
"loss": 1.3579959869384766,
"step": 40
},
{
"epoch": 0.048109965635738834,
"grad_norm": 0.530888020992279,
"learning_rate": 3.129770992366413e-06,
"loss": 1.86295747756958,
"step": 42
},
{
"epoch": 0.050400916380297825,
"grad_norm": 0.545220673084259,
"learning_rate": 3.2824427480916034e-06,
"loss": 1.8225860595703125,
"step": 44
},
{
"epoch": 0.052691867124856816,
"grad_norm": 4.15400505065918,
"learning_rate": 3.4351145038167944e-06,
"loss": 2.479480743408203,
"step": 46
},
{
"epoch": 0.054982817869415807,
"grad_norm": 2.230881452560425,
"learning_rate": 3.587786259541985e-06,
"loss": 1.9663472175598145,
"step": 48
},
{
"epoch": 0.0572737686139748,
"grad_norm": 0.5587590932846069,
"learning_rate": 3.740458015267176e-06,
"loss": 1.7919893264770508,
"step": 50
},
{
"epoch": 0.05956471935853379,
"grad_norm": 0.2856103777885437,
"learning_rate": 3.893129770992366e-06,
"loss": 1.5717711448669434,
"step": 52
},
{
"epoch": 0.061855670103092786,
"grad_norm": 0.680905282497406,
"learning_rate": 4.045801526717557e-06,
"loss": 1.937774658203125,
"step": 54
},
{
"epoch": 0.06414662084765177,
"grad_norm": 1.862351894378662,
"learning_rate": 4.198473282442748e-06,
"loss": 1.5575186014175415,
"step": 56
},
{
"epoch": 0.06643757159221077,
"grad_norm": 0.7627670764923096,
"learning_rate": 4.351145038167939e-06,
"loss": 2.3777570724487305,
"step": 58
},
{
"epoch": 0.06872852233676977,
"grad_norm": 3.405487537384033,
"learning_rate": 4.5038167938931296e-06,
"loss": 2.064112663269043,
"step": 60
},
{
"epoch": 0.07101947308132875,
"grad_norm": 0.22335933148860931,
"learning_rate": 4.656488549618321e-06,
"loss": 1.5149509906768799,
"step": 62
},
{
"epoch": 0.07331042382588775,
"grad_norm": 1.8504610061645508,
"learning_rate": 4.8091603053435125e-06,
"loss": 2.0704846382141113,
"step": 64
},
{
"epoch": 0.07560137457044673,
"grad_norm": 0.48437029123306274,
"learning_rate": 4.961832061068703e-06,
"loss": 1.5357882976531982,
"step": 66
},
{
"epoch": 0.07789232531500573,
"grad_norm": 0.3509262800216675,
"learning_rate": 5.114503816793893e-06,
"loss": 1.7561030387878418,
"step": 68
},
{
"epoch": 0.08018327605956473,
"grad_norm": 1.3714377880096436,
"learning_rate": 5.267175572519084e-06,
"loss": 1.6338742971420288,
"step": 70
},
{
"epoch": 0.08247422680412371,
"grad_norm": 1.4160592555999756,
"learning_rate": 5.419847328244276e-06,
"loss": 1.5981454849243164,
"step": 72
},
{
"epoch": 0.08476517754868271,
"grad_norm": 0.396519273519516,
"learning_rate": 5.572519083969467e-06,
"loss": 1.8558708429336548,
"step": 74
},
{
"epoch": 0.08705612829324169,
"grad_norm": 0.9855345487594604,
"learning_rate": 5.725190839694656e-06,
"loss": 1.506034016609192,
"step": 76
},
{
"epoch": 0.08934707903780069,
"grad_norm": 1.6816610097885132,
"learning_rate": 5.877862595419848e-06,
"loss": 1.978309988975525,
"step": 78
},
{
"epoch": 0.09163802978235967,
"grad_norm": 0.49164819717407227,
"learning_rate": 6.030534351145039e-06,
"loss": 1.7104287147521973,
"step": 80
},
{
"epoch": 0.09392898052691867,
"grad_norm": 0.5583735704421997,
"learning_rate": 6.18320610687023e-06,
"loss": 1.4178941249847412,
"step": 82
},
{
"epoch": 0.09621993127147767,
"grad_norm": 0.5168147087097168,
"learning_rate": 6.335877862595419e-06,
"loss": 1.7817795276641846,
"step": 84
},
{
"epoch": 0.09851088201603665,
"grad_norm": 1.2413756847381592,
"learning_rate": 6.488549618320611e-06,
"loss": 1.3074209690093994,
"step": 86
},
{
"epoch": 0.10080183276059565,
"grad_norm": 0.6331862807273865,
"learning_rate": 6.641221374045802e-06,
"loss": 1.5984269380569458,
"step": 88
},
{
"epoch": 0.10309278350515463,
"grad_norm": 0.5664435029029846,
"learning_rate": 6.793893129770993e-06,
"loss": 1.7241010665893555,
"step": 90
},
{
"epoch": 0.10538373424971363,
"grad_norm": 0.32952845096588135,
"learning_rate": 6.946564885496184e-06,
"loss": 1.5395644903182983,
"step": 92
},
{
"epoch": 0.10767468499427263,
"grad_norm": 0.41334185004234314,
"learning_rate": 7.0992366412213746e-06,
"loss": 1.7171587944030762,
"step": 94
},
{
"epoch": 0.10996563573883161,
"grad_norm": 0.3259168863296509,
"learning_rate": 7.251908396946566e-06,
"loss": 1.3641917705535889,
"step": 96
},
{
"epoch": 0.11225658648339061,
"grad_norm": 0.8971285223960876,
"learning_rate": 7.404580152671757e-06,
"loss": 1.6456998586654663,
"step": 98
},
{
"epoch": 0.1145475372279496,
"grad_norm": 2.403928279876709,
"learning_rate": 7.557251908396948e-06,
"loss": 1.6644506454467773,
"step": 100
},
{
"epoch": 0.11683848797250859,
"grad_norm": 0.5824722647666931,
"learning_rate": 7.709923664122137e-06,
"loss": 1.7684416770935059,
"step": 102
},
{
"epoch": 0.11912943871706758,
"grad_norm": 0.2737939655780792,
"learning_rate": 7.862595419847328e-06,
"loss": 1.457775354385376,
"step": 104
},
{
"epoch": 0.12142038946162657,
"grad_norm": 0.20071668922901154,
"learning_rate": 8.015267175572519e-06,
"loss": 1.3262975215911865,
"step": 106
},
{
"epoch": 0.12371134020618557,
"grad_norm": 0.43577471375465393,
"learning_rate": 8.16793893129771e-06,
"loss": 1.5433118343353271,
"step": 108
},
{
"epoch": 0.12600229095074456,
"grad_norm": 0.43102404475212097,
"learning_rate": 8.320610687022901e-06,
"loss": 1.4821274280548096,
"step": 110
},
{
"epoch": 0.12829324169530354,
"grad_norm": 0.450750470161438,
"learning_rate": 8.473282442748092e-06,
"loss": 1.7386858463287354,
"step": 112
},
{
"epoch": 0.13058419243986255,
"grad_norm": 0.49831417202949524,
"learning_rate": 8.625954198473283e-06,
"loss": 1.5783125162124634,
"step": 114
},
{
"epoch": 0.13287514318442153,
"grad_norm": 0.5326170921325684,
"learning_rate": 8.778625954198474e-06,
"loss": 1.6655735969543457,
"step": 116
},
{
"epoch": 0.13516609392898052,
"grad_norm": 0.4478776752948761,
"learning_rate": 8.931297709923665e-06,
"loss": 1.5613455772399902,
"step": 118
},
{
"epoch": 0.13745704467353953,
"grad_norm": 1.2814429998397827,
"learning_rate": 9.083969465648855e-06,
"loss": 1.3779879808425903,
"step": 120
},
{
"epoch": 0.13974799541809851,
"grad_norm": 0.3124754726886749,
"learning_rate": 9.236641221374046e-06,
"loss": 1.2922519445419312,
"step": 122
},
{
"epoch": 0.1420389461626575,
"grad_norm": 0.5107002258300781,
"learning_rate": 9.389312977099237e-06,
"loss": 1.204099416732788,
"step": 124
},
{
"epoch": 0.14432989690721648,
"grad_norm": 0.42890945076942444,
"learning_rate": 9.54198473282443e-06,
"loss": 1.5295710563659668,
"step": 126
},
{
"epoch": 0.1466208476517755,
"grad_norm": 0.3888114094734192,
"learning_rate": 9.694656488549619e-06,
"loss": 1.446217656135559,
"step": 128
},
{
"epoch": 0.14891179839633448,
"grad_norm": 1.3135894536972046,
"learning_rate": 9.84732824427481e-06,
"loss": 1.696859359741211,
"step": 130
},
{
"epoch": 0.15120274914089346,
"grad_norm": 0.5034719705581665,
"learning_rate": 1e-05,
"loss": 1.4995859861373901,
"step": 132
},
{
"epoch": 0.15349369988545247,
"grad_norm": 0.22584287822246552,
"learning_rate": 9.999985650351204e-06,
"loss": 1.1352298259735107,
"step": 134
},
{
"epoch": 0.15578465063001146,
"grad_norm": 0.3132134974002838,
"learning_rate": 9.999942601496331e-06,
"loss": 1.208528995513916,
"step": 136
},
{
"epoch": 0.15807560137457044,
"grad_norm": 0.41238704323768616,
"learning_rate": 9.999870853709929e-06,
"loss": 1.4549105167388916,
"step": 138
},
{
"epoch": 0.16036655211912945,
"grad_norm": 0.18063485622406006,
"learning_rate": 9.999770407449582e-06,
"loss": 1.1948139667510986,
"step": 140
},
{
"epoch": 0.16265750286368844,
"grad_norm": 0.7935987710952759,
"learning_rate": 9.999641263355893e-06,
"loss": 1.3297820091247559,
"step": 142
},
{
"epoch": 0.16494845360824742,
"grad_norm": 0.3728874921798706,
"learning_rate": 9.999483422252499e-06,
"loss": 1.4431211948394775,
"step": 144
},
{
"epoch": 0.1672394043528064,
"grad_norm": 0.32941386103630066,
"learning_rate": 9.999296885146047e-06,
"loss": 1.2147037982940674,
"step": 146
},
{
"epoch": 0.16953035509736541,
"grad_norm": 0.4308471381664276,
"learning_rate": 9.999081653226205e-06,
"loss": 1.3048083782196045,
"step": 148
},
{
"epoch": 0.1718213058419244,
"grad_norm": 0.7290281057357788,
"learning_rate": 9.998837727865636e-06,
"loss": 1.1929588317871094,
"step": 150
},
{
"epoch": 0.17411225658648338,
"grad_norm": 1.055001139640808,
"learning_rate": 9.998565110620006e-06,
"loss": 1.0773563385009766,
"step": 152
},
{
"epoch": 0.1764032073310424,
"grad_norm": 0.3784625828266144,
"learning_rate": 9.998263803227965e-06,
"loss": 1.4228711128234863,
"step": 154
},
{
"epoch": 0.17869415807560138,
"grad_norm": 0.9958500862121582,
"learning_rate": 9.997933807611133e-06,
"loss": 1.798612356185913,
"step": 156
},
{
"epoch": 0.18098510882016036,
"grad_norm": 0.370061457157135,
"learning_rate": 9.997575125874104e-06,
"loss": 1.4642460346221924,
"step": 158
},
{
"epoch": 0.18327605956471935,
"grad_norm": 0.7844455242156982,
"learning_rate": 9.997187760304411e-06,
"loss": 1.0867466926574707,
"step": 160
},
{
"epoch": 0.18556701030927836,
"grad_norm": 0.5508542060852051,
"learning_rate": 9.996771713372525e-06,
"loss": 1.295719861984253,
"step": 162
},
{
"epoch": 0.18785796105383734,
"grad_norm": 0.393774151802063,
"learning_rate": 9.996326987731836e-06,
"loss": 1.0488743782043457,
"step": 164
},
{
"epoch": 0.19014891179839633,
"grad_norm": 0.3985389173030853,
"learning_rate": 9.995853586218636e-06,
"loss": 1.5313889980316162,
"step": 166
},
{
"epoch": 0.19243986254295534,
"grad_norm": 1.3936595916748047,
"learning_rate": 9.995351511852102e-06,
"loss": 0.6738256812095642,
"step": 168
},
{
"epoch": 0.19473081328751432,
"grad_norm": 0.3983599543571472,
"learning_rate": 9.994820767834273e-06,
"loss": 1.4490211009979248,
"step": 170
},
{
"epoch": 0.1970217640320733,
"grad_norm": 0.8028084635734558,
"learning_rate": 9.994261357550034e-06,
"loss": 1.5482444763183594,
"step": 172
},
{
"epoch": 0.19931271477663232,
"grad_norm": 0.5651741623878479,
"learning_rate": 9.993673284567092e-06,
"loss": 1.6904449462890625,
"step": 174
},
{
"epoch": 0.2016036655211913,
"grad_norm": 0.6857726573944092,
"learning_rate": 9.993056552635954e-06,
"loss": 1.4769976139068604,
"step": 176
},
{
"epoch": 0.20389461626575028,
"grad_norm": 1.0751330852508545,
"learning_rate": 9.992411165689902e-06,
"loss": 1.4480886459350586,
"step": 178
},
{
"epoch": 0.20618556701030927,
"grad_norm": 0.9354397654533386,
"learning_rate": 9.99173712784497e-06,
"loss": 1.581247329711914,
"step": 180
},
{
"epoch": 0.20847651775486828,
"grad_norm": 0.49207422137260437,
"learning_rate": 9.99103444339992e-06,
"loss": 1.327151894569397,
"step": 182
},
{
"epoch": 0.21076746849942726,
"grad_norm": 0.7072939872741699,
"learning_rate": 9.990303116836204e-06,
"loss": 1.3334730863571167,
"step": 184
},
{
"epoch": 0.21305841924398625,
"grad_norm": 0.9365259408950806,
"learning_rate": 9.989543152817945e-06,
"loss": 1.1465511322021484,
"step": 186
},
{
"epoch": 0.21534936998854526,
"grad_norm": 0.4507530629634857,
"learning_rate": 9.98875455619191e-06,
"loss": 1.3069469928741455,
"step": 188
},
{
"epoch": 0.21764032073310424,
"grad_norm": 0.2077871412038803,
"learning_rate": 9.987937331987466e-06,
"loss": 1.2803224325180054,
"step": 190
},
{
"epoch": 0.21993127147766323,
"grad_norm": 0.2749679386615753,
"learning_rate": 9.987091485416564e-06,
"loss": 1.1666932106018066,
"step": 192
},
{
"epoch": 0.2222222222222222,
"grad_norm": 0.5907368659973145,
"learning_rate": 9.986217021873688e-06,
"loss": 1.28110933303833,
"step": 194
},
{
"epoch": 0.22451317296678122,
"grad_norm": 0.34716930985450745,
"learning_rate": 9.985313946935841e-06,
"loss": 1.1845691204071045,
"step": 196
},
{
"epoch": 0.2268041237113402,
"grad_norm": 0.5304384231567383,
"learning_rate": 9.98438226636249e-06,
"loss": 1.4581682682037354,
"step": 198
},
{
"epoch": 0.2290950744558992,
"grad_norm": 0.5311458110809326,
"learning_rate": 9.983421986095543e-06,
"loss": 1.5132863521575928,
"step": 200
},
{
"epoch": 0.2313860252004582,
"grad_norm": 0.9410415291786194,
"learning_rate": 9.982433112259304e-06,
"loss": 1.147586464881897,
"step": 202
},
{
"epoch": 0.23367697594501718,
"grad_norm": 0.5174320936203003,
"learning_rate": 9.981415651160434e-06,
"loss": 1.3612395524978638,
"step": 204
},
{
"epoch": 0.23596792668957617,
"grad_norm": 0.3282371461391449,
"learning_rate": 9.980369609287918e-06,
"loss": 1.2773010730743408,
"step": 206
},
{
"epoch": 0.23825887743413515,
"grad_norm": 1.3350720405578613,
"learning_rate": 9.979294993313013e-06,
"loss": 1.3632477521896362,
"step": 208
},
{
"epoch": 0.24054982817869416,
"grad_norm": 0.31105804443359375,
"learning_rate": 9.978191810089213e-06,
"loss": 1.2480723857879639,
"step": 210
},
{
"epoch": 0.24284077892325315,
"grad_norm": 0.3278603255748749,
"learning_rate": 9.977060066652208e-06,
"loss": 1.1874191761016846,
"step": 212
},
{
"epoch": 0.24513172966781213,
"grad_norm": 0.21448446810245514,
"learning_rate": 9.975899770219823e-06,
"loss": 1.24106764793396,
"step": 214
},
{
"epoch": 0.24742268041237114,
"grad_norm": 0.8085087537765503,
"learning_rate": 9.974710928191994e-06,
"loss": 1.4966447353363037,
"step": 216
},
{
"epoch": 0.24971363115693013,
"grad_norm": 0.4925968647003174,
"learning_rate": 9.973493548150705e-06,
"loss": 1.246765375137329,
"step": 218
},
{
"epoch": 0.2520045819014891,
"grad_norm": 0.5599343180656433,
"learning_rate": 9.972247637859942e-06,
"loss": 1.5764102935791016,
"step": 220
},
{
"epoch": 0.2542955326460481,
"grad_norm": 1.497615933418274,
"learning_rate": 9.970973205265654e-06,
"loss": 1.1964993476867676,
"step": 222
},
{
"epoch": 0.2565864833906071,
"grad_norm": 0.284583181142807,
"learning_rate": 9.969670258495689e-06,
"loss": 1.4045443534851074,
"step": 224
},
{
"epoch": 0.2588774341351661,
"grad_norm": 0.37673652172088623,
"learning_rate": 9.968338805859746e-06,
"loss": 1.1107515096664429,
"step": 226
},
{
"epoch": 0.2611683848797251,
"grad_norm": 0.1895330548286438,
"learning_rate": 9.966978855849328e-06,
"loss": 1.2244822978973389,
"step": 228
},
{
"epoch": 0.2634593356242841,
"grad_norm": 0.6602473258972168,
"learning_rate": 9.965590417137683e-06,
"loss": 1.4389466047286987,
"step": 230
},
{
"epoch": 0.26575028636884307,
"grad_norm": 0.7053688764572144,
"learning_rate": 9.964173498579744e-06,
"loss": 1.3052401542663574,
"step": 232
},
{
"epoch": 0.26804123711340205,
"grad_norm": 0.46937069296836853,
"learning_rate": 9.962728109212087e-06,
"loss": 1.6006953716278076,
"step": 234
},
{
"epoch": 0.27033218785796104,
"grad_norm": 0.6499706506729126,
"learning_rate": 9.961254258252853e-06,
"loss": 1.7075961828231812,
"step": 236
},
{
"epoch": 0.27262313860252,
"grad_norm": 0.28833746910095215,
"learning_rate": 9.95975195510171e-06,
"loss": 1.2591079473495483,
"step": 238
},
{
"epoch": 0.27491408934707906,
"grad_norm": 0.33017802238464355,
"learning_rate": 9.958221209339776e-06,
"loss": 1.3728235960006714,
"step": 240
},
{
"epoch": 0.27720504009163804,
"grad_norm": 0.42698171734809875,
"learning_rate": 9.956662030729571e-06,
"loss": 1.4098225831985474,
"step": 242
},
{
"epoch": 0.27949599083619703,
"grad_norm": 1.0393098592758179,
"learning_rate": 9.955074429214945e-06,
"loss": 1.2753400802612305,
"step": 244
},
{
"epoch": 0.281786941580756,
"grad_norm": 0.3506874144077301,
"learning_rate": 9.95345841492102e-06,
"loss": 1.3643712997436523,
"step": 246
},
{
"epoch": 0.284077892325315,
"grad_norm": 0.36692970991134644,
"learning_rate": 9.951813998154122e-06,
"loss": 1.1439402103424072,
"step": 248
},
{
"epoch": 0.286368843069874,
"grad_norm": 1.3749401569366455,
"learning_rate": 9.950141189401722e-06,
"loss": 1.4724159240722656,
"step": 250
},
{
"epoch": 0.28865979381443296,
"grad_norm": 0.8339785933494568,
"learning_rate": 9.948439999332362e-06,
"loss": 1.4834284782409668,
"step": 252
},
{
"epoch": 0.290950744558992,
"grad_norm": 0.7085672616958618,
"learning_rate": 9.946710438795586e-06,
"loss": 1.4935444593429565,
"step": 254
},
{
"epoch": 0.293241695303551,
"grad_norm": 0.39228612184524536,
"learning_rate": 9.944952518821877e-06,
"loss": 0.9836521148681641,
"step": 256
},
{
"epoch": 0.29553264604810997,
"grad_norm": 0.8522858023643494,
"learning_rate": 9.943166250622585e-06,
"loss": 1.2665026187896729,
"step": 258
},
{
"epoch": 0.29782359679266895,
"grad_norm": 1.1095147132873535,
"learning_rate": 9.941351645589853e-06,
"loss": 0.9271557331085205,
"step": 260
},
{
"epoch": 0.30011454753722794,
"grad_norm": 0.5586176514625549,
"learning_rate": 9.939508715296543e-06,
"loss": 1.329547643661499,
"step": 262
},
{
"epoch": 0.3024054982817869,
"grad_norm": 0.662735641002655,
"learning_rate": 9.93763747149617e-06,
"loss": 1.247317910194397,
"step": 264
},
{
"epoch": 0.30469644902634596,
"grad_norm": 0.8601267337799072,
"learning_rate": 9.935737926122816e-06,
"loss": 1.2623186111450195,
"step": 266
},
{
"epoch": 0.30698739977090495,
"grad_norm": 0.20243214070796967,
"learning_rate": 9.933810091291065e-06,
"loss": 1.289118766784668,
"step": 268
},
{
"epoch": 0.30927835051546393,
"grad_norm": 0.28063586354255676,
"learning_rate": 9.93185397929592e-06,
"loss": 1.1982347965240479,
"step": 270
},
{
"epoch": 0.3115693012600229,
"grad_norm": 0.42941388487815857,
"learning_rate": 9.929869602612718e-06,
"loss": 1.137695074081421,
"step": 272
},
{
"epoch": 0.3138602520045819,
"grad_norm": 0.45877066254615784,
"learning_rate": 9.927856973897068e-06,
"loss": 0.5235239267349243,
"step": 274
},
{
"epoch": 0.3161512027491409,
"grad_norm": 0.6997596025466919,
"learning_rate": 9.925816105984751e-06,
"loss": 1.3950223922729492,
"step": 276
},
{
"epoch": 0.31844215349369986,
"grad_norm": 1.7510356903076172,
"learning_rate": 9.923747011891653e-06,
"loss": 0.6258569955825806,
"step": 278
},
{
"epoch": 0.3207331042382589,
"grad_norm": 0.5782098770141602,
"learning_rate": 9.92164970481367e-06,
"loss": 1.348238229751587,
"step": 280
},
{
"epoch": 0.3230240549828179,
"grad_norm": 0.8980709910392761,
"learning_rate": 9.919524198126637e-06,
"loss": 1.1449611186981201,
"step": 282
},
{
"epoch": 0.32531500572737687,
"grad_norm": 0.8491981029510498,
"learning_rate": 9.91737050538623e-06,
"loss": 1.620896816253662,
"step": 284
},
{
"epoch": 0.32760595647193586,
"grad_norm": 0.6017739772796631,
"learning_rate": 9.915188640327887e-06,
"loss": 0.7882466316223145,
"step": 286
},
{
"epoch": 0.32989690721649484,
"grad_norm": 0.41330039501190186,
"learning_rate": 9.912978616866716e-06,
"loss": 1.4667167663574219,
"step": 288
},
{
"epoch": 0.3321878579610538,
"grad_norm": 1.4287890195846558,
"learning_rate": 9.910740449097412e-06,
"loss": 1.4368730783462524,
"step": 290
},
{
"epoch": 0.3344788087056128,
"grad_norm": 0.7372951507568359,
"learning_rate": 9.908474151294161e-06,
"loss": 1.0971758365631104,
"step": 292
},
{
"epoch": 0.33676975945017185,
"grad_norm": 1.4568077325820923,
"learning_rate": 9.906179737910554e-06,
"loss": 1.174417495727539,
"step": 294
},
{
"epoch": 0.33906071019473083,
"grad_norm": 0.5502243638038635,
"learning_rate": 9.903857223579496e-06,
"loss": 0.9996079206466675,
"step": 296
},
{
"epoch": 0.3413516609392898,
"grad_norm": 0.2131074070930481,
"learning_rate": 9.901506623113098e-06,
"loss": 0.9655183553695679,
"step": 298
},
{
"epoch": 0.3436426116838488,
"grad_norm": 0.45683225989341736,
"learning_rate": 9.899127951502601e-06,
"loss": 1.239179253578186,
"step": 300
},
{
"epoch": 0.3459335624284078,
"grad_norm": 0.7904418110847473,
"learning_rate": 9.896721223918276e-06,
"loss": 1.5110833644866943,
"step": 302
},
{
"epoch": 0.34822451317296677,
"grad_norm": 0.19148200750350952,
"learning_rate": 9.89428645570932e-06,
"loss": 1.169365406036377,
"step": 304
},
{
"epoch": 0.35051546391752575,
"grad_norm": 0.9735309481620789,
"learning_rate": 9.891823662403763e-06,
"loss": 1.0482746362686157,
"step": 306
},
{
"epoch": 0.3528064146620848,
"grad_norm": 0.5635564923286438,
"learning_rate": 9.88933285970837e-06,
"loss": 1.3521157503128052,
"step": 308
},
{
"epoch": 0.3550973654066438,
"grad_norm": 0.41833338141441345,
"learning_rate": 9.886814063508536e-06,
"loss": 1.1932569742202759,
"step": 310
},
{
"epoch": 0.35738831615120276,
"grad_norm": 0.5040867328643799,
"learning_rate": 9.884267289868194e-06,
"loss": 1.2639671564102173,
"step": 312
},
{
"epoch": 0.35967926689576174,
"grad_norm": 0.3760809004306793,
"learning_rate": 9.8816925550297e-06,
"loss": 1.36051607131958,
"step": 314
},
{
"epoch": 0.3619702176403207,
"grad_norm": 1.2754006385803223,
"learning_rate": 9.879089875413736e-06,
"loss": 1.253896713256836,
"step": 316
},
{
"epoch": 0.3642611683848797,
"grad_norm": 0.43398040533065796,
"learning_rate": 9.876459267619215e-06,
"loss": 1.2675503492355347,
"step": 318
},
{
"epoch": 0.3665521191294387,
"grad_norm": 1.7002805471420288,
"learning_rate": 9.873800748423152e-06,
"loss": 1.3285064697265625,
"step": 320
},
{
"epoch": 0.36884306987399773,
"grad_norm": 1.125661849975586,
"learning_rate": 9.871114334780583e-06,
"loss": 1.3152642250061035,
"step": 322
},
{
"epoch": 0.3711340206185567,
"grad_norm": 0.5501522421836853,
"learning_rate": 9.868400043824431e-06,
"loss": 1.3212919235229492,
"step": 324
},
{
"epoch": 0.3734249713631157,
"grad_norm": 3.444338083267212,
"learning_rate": 9.86565789286542e-06,
"loss": 1.4326891899108887,
"step": 326
},
{
"epoch": 0.3757159221076747,
"grad_norm": 0.23539075255393982,
"learning_rate": 9.862887899391953e-06,
"loss": 1.2380142211914062,
"step": 328
},
{
"epoch": 0.37800687285223367,
"grad_norm": 0.7268555164337158,
"learning_rate": 9.860090081069998e-06,
"loss": 1.1204144954681396,
"step": 330
},
{
"epoch": 0.38029782359679265,
"grad_norm": 0.8944157958030701,
"learning_rate": 9.857264455742983e-06,
"loss": 1.2395751476287842,
"step": 332
},
{
"epoch": 0.38258877434135163,
"grad_norm": 0.3225472867488861,
"learning_rate": 9.854411041431678e-06,
"loss": 1.1439430713653564,
"step": 334
},
{
"epoch": 0.3848797250859107,
"grad_norm": 0.7560716867446899,
"learning_rate": 9.851529856334079e-06,
"loss": 1.2860313653945923,
"step": 336
},
{
"epoch": 0.38717067583046966,
"grad_norm": 0.2908443212509155,
"learning_rate": 9.848620918825294e-06,
"loss": 1.2428202629089355,
"step": 338
},
{
"epoch": 0.38946162657502864,
"grad_norm": 0.6176069974899292,
"learning_rate": 9.845684247457425e-06,
"loss": 1.371431827545166,
"step": 340
},
{
"epoch": 0.3917525773195876,
"grad_norm": 0.308256059885025,
"learning_rate": 9.842719860959455e-06,
"loss": 1.2362736463546753,
"step": 342
},
{
"epoch": 0.3940435280641466,
"grad_norm": 0.2752687931060791,
"learning_rate": 9.839727778237116e-06,
"loss": 1.3125758171081543,
"step": 344
},
{
"epoch": 0.3963344788087056,
"grad_norm": 1.0439542531967163,
"learning_rate": 9.836708018372782e-06,
"loss": 1.3099775314331055,
"step": 346
},
{
"epoch": 0.39862542955326463,
"grad_norm": 0.514342188835144,
"learning_rate": 9.833660600625338e-06,
"loss": 1.212874412536621,
"step": 348
},
{
"epoch": 0.4009163802978236,
"grad_norm": 0.5850813388824463,
"learning_rate": 9.83058554443006e-06,
"loss": 1.2086424827575684,
"step": 350
},
{
"epoch": 0.4032073310423826,
"grad_norm": 1.0894215106964111,
"learning_rate": 9.827482869398496e-06,
"loss": 1.3154696226119995,
"step": 352
},
{
"epoch": 0.4054982817869416,
"grad_norm": 3.362022876739502,
"learning_rate": 9.82435259531833e-06,
"loss": 1.058045506477356,
"step": 354
},
{
"epoch": 0.40778923253150057,
"grad_norm": 0.2783519923686981,
"learning_rate": 9.82119474215327e-06,
"loss": 0.8597267866134644,
"step": 356
},
{
"epoch": 0.41008018327605955,
"grad_norm": 0.3932453691959381,
"learning_rate": 9.818009330042906e-06,
"loss": 1.4340499639511108,
"step": 358
},
{
"epoch": 0.41237113402061853,
"grad_norm": 0.5507175922393799,
"learning_rate": 9.814796379302592e-06,
"loss": 1.2930049896240234,
"step": 360
},
{
"epoch": 0.4146620847651776,
"grad_norm": 0.40081146359443665,
"learning_rate": 9.811555910423312e-06,
"loss": 1.1826398372650146,
"step": 362
},
{
"epoch": 0.41695303550973656,
"grad_norm": 0.4745732545852661,
"learning_rate": 9.808287944071552e-06,
"loss": 1.3191637992858887,
"step": 364
},
{
"epoch": 0.41924398625429554,
"grad_norm": 0.5023146867752075,
"learning_rate": 9.804992501089164e-06,
"loss": 1.4524482488632202,
"step": 366
},
{
"epoch": 0.4215349369988545,
"grad_norm": 0.6656132340431213,
"learning_rate": 9.801669602493236e-06,
"loss": 1.2732700109481812,
"step": 368
},
{
"epoch": 0.4238258877434135,
"grad_norm": 2.312171220779419,
"learning_rate": 9.798319269475959e-06,
"loss": 1.1589066982269287,
"step": 370
},
{
"epoch": 0.4261168384879725,
"grad_norm": 2.0123043060302734,
"learning_rate": 9.794941523404491e-06,
"loss": 1.3179020881652832,
"step": 372
},
{
"epoch": 0.4284077892325315,
"grad_norm": 0.36115938425064087,
"learning_rate": 9.791536385820815e-06,
"loss": 1.330890417098999,
"step": 374
},
{
"epoch": 0.4306987399770905,
"grad_norm": 0.10276070982217789,
"learning_rate": 9.788103878441614e-06,
"loss": 0.9845031499862671,
"step": 376
},
{
"epoch": 0.4329896907216495,
"grad_norm": 0.5846933126449585,
"learning_rate": 9.784644023158118e-06,
"loss": 1.2902302742004395,
"step": 378
},
{
"epoch": 0.4352806414662085,
"grad_norm": 0.45211294293403625,
"learning_rate": 9.781156842035978e-06,
"loss": 1.2718219757080078,
"step": 380
},
{
"epoch": 0.43757159221076747,
"grad_norm": 0.4230896234512329,
"learning_rate": 9.777642357315115e-06,
"loss": 1.4458763599395752,
"step": 382
},
{
"epoch": 0.43986254295532645,
"grad_norm": 0.5726656913757324,
"learning_rate": 9.774100591409583e-06,
"loss": 0.866162896156311,
"step": 384
},
{
"epoch": 0.44215349369988544,
"grad_norm": 0.2680141031742096,
"learning_rate": 9.770531566907424e-06,
"loss": 1.1903131008148193,
"step": 386
},
{
"epoch": 0.4444444444444444,
"grad_norm": 0.21644671261310577,
"learning_rate": 9.766935306570528e-06,
"loss": 1.0767426490783691,
"step": 388
},
{
"epoch": 0.44673539518900346,
"grad_norm": 0.40251392126083374,
"learning_rate": 9.763311833334482e-06,
"loss": 1.206404209136963,
"step": 390
},
{
"epoch": 0.44902634593356244,
"grad_norm": 0.26591402292251587,
"learning_rate": 9.759661170308426e-06,
"loss": 1.0683099031448364,
"step": 392
},
{
"epoch": 0.4513172966781214,
"grad_norm": 0.6021142601966858,
"learning_rate": 9.75598334077491e-06,
"loss": 1.4385409355163574,
"step": 394
},
{
"epoch": 0.4536082474226804,
"grad_norm": 0.4870530068874359,
"learning_rate": 9.752278368189738e-06,
"loss": 1.2794133424758911,
"step": 396
},
{
"epoch": 0.4558991981672394,
"grad_norm": 0.5457776784896851,
"learning_rate": 9.748546276181824e-06,
"loss": 1.3290512561798096,
"step": 398
},
{
"epoch": 0.4581901489117984,
"grad_norm": 0.6035308837890625,
"learning_rate": 9.74478708855304e-06,
"loss": 1.1363365650177002,
"step": 400
},
{
"epoch": 0.46048109965635736,
"grad_norm": 1.4187884330749512,
"learning_rate": 9.74100082927806e-06,
"loss": 1.1181583404541016,
"step": 402
},
{
"epoch": 0.4627720504009164,
"grad_norm": 0.6478531360626221,
"learning_rate": 9.737187522504215e-06,
"loss": 1.292447805404663,
"step": 404
},
{
"epoch": 0.4650630011454754,
"grad_norm": 0.45188528299331665,
"learning_rate": 9.733347192551333e-06,
"loss": 1.341691493988037,
"step": 406
},
{
"epoch": 0.46735395189003437,
"grad_norm": 1.0683960914611816,
"learning_rate": 9.729479863911585e-06,
"loss": 0.9544723033905029,
"step": 408
},
{
"epoch": 0.46964490263459335,
"grad_norm": 0.40036776661872864,
"learning_rate": 9.725585561249331e-06,
"loss": 1.299605369567871,
"step": 410
},
{
"epoch": 0.47193585337915234,
"grad_norm": 1.0352967977523804,
"learning_rate": 9.72166430940096e-06,
"loss": 1.5143306255340576,
"step": 412
},
{
"epoch": 0.4742268041237113,
"grad_norm": 0.47124430537223816,
"learning_rate": 9.71771613337473e-06,
"loss": 1.2042604684829712,
"step": 414
},
{
"epoch": 0.4765177548682703,
"grad_norm": 0.40640538930892944,
"learning_rate": 9.713741058350618e-06,
"loss": 1.3638975620269775,
"step": 416
},
{
"epoch": 0.47880870561282934,
"grad_norm": 0.22503964602947235,
"learning_rate": 9.709739109680146e-06,
"loss": 1.0454559326171875,
"step": 418
},
{
"epoch": 0.48109965635738833,
"grad_norm": 0.7164255976676941,
"learning_rate": 9.70571031288623e-06,
"loss": 1.4321982860565186,
"step": 420
},
{
"epoch": 0.4833906071019473,
"grad_norm": 0.22116360068321228,
"learning_rate": 9.701654693663012e-06,
"loss": 1.2148916721343994,
"step": 422
},
{
"epoch": 0.4856815578465063,
"grad_norm": 0.40741488337516785,
"learning_rate": 9.697572277875696e-06,
"loss": 1.2174824476242065,
"step": 424
},
{
"epoch": 0.4879725085910653,
"grad_norm": 0.33800235390663147,
"learning_rate": 9.693463091560387e-06,
"loss": 1.0184588432312012,
"step": 426
},
{
"epoch": 0.49026345933562426,
"grad_norm": 1.2050279378890991,
"learning_rate": 9.689327160923918e-06,
"loss": 1.278519868850708,
"step": 428
},
{
"epoch": 0.4925544100801833,
"grad_norm": 0.3495669364929199,
"learning_rate": 9.685164512343694e-06,
"loss": 1.2338956594467163,
"step": 430
},
{
"epoch": 0.4948453608247423,
"grad_norm": 0.23146063089370728,
"learning_rate": 9.680975172367508e-06,
"loss": 1.40183687210083,
"step": 432
},
{
"epoch": 0.49713631156930127,
"grad_norm": 0.21940575540065765,
"learning_rate": 9.67675916771339e-06,
"loss": 1.1633131504058838,
"step": 434
},
{
"epoch": 0.49942726231386025,
"grad_norm": 0.47240111231803894,
"learning_rate": 9.67251652526942e-06,
"loss": 1.1047279834747314,
"step": 436
},
{
"epoch": 0.5017182130584192,
"grad_norm": 0.36501702666282654,
"learning_rate": 9.668247272093568e-06,
"loss": 1.256941795349121,
"step": 438
},
{
"epoch": 0.5040091638029782,
"grad_norm": 0.4658450782299042,
"learning_rate": 9.663951435413512e-06,
"loss": 1.454789400100708,
"step": 440
},
{
"epoch": 0.5063001145475372,
"grad_norm": 0.5242023468017578,
"learning_rate": 9.659629042626478e-06,
"loss": 1.2262189388275146,
"step": 442
},
{
"epoch": 0.5085910652920962,
"grad_norm": 1.0691118240356445,
"learning_rate": 9.655280121299049e-06,
"loss": 1.2670984268188477,
"step": 444
},
{
"epoch": 0.5108820160366552,
"grad_norm": 0.7168107032775879,
"learning_rate": 9.650904699167002e-06,
"loss": 1.2349098920822144,
"step": 446
},
{
"epoch": 0.5131729667812142,
"grad_norm": 0.3086417019367218,
"learning_rate": 9.646502804135125e-06,
"loss": 1.1503790616989136,
"step": 448
},
{
"epoch": 0.5154639175257731,
"grad_norm": 0.15410441160202026,
"learning_rate": 9.642074464277035e-06,
"loss": 1.050001859664917,
"step": 450
},
{
"epoch": 0.5177548682703322,
"grad_norm": 0.923253059387207,
"learning_rate": 9.637619707835011e-06,
"loss": 1.10781729221344,
"step": 452
},
{
"epoch": 0.5200458190148912,
"grad_norm": 0.38856205344200134,
"learning_rate": 9.633138563219805e-06,
"loss": 1.2779165506362915,
"step": 454
},
{
"epoch": 0.5223367697594502,
"grad_norm": 0.22952410578727722,
"learning_rate": 9.628631059010459e-06,
"loss": 1.1592912673950195,
"step": 456
},
{
"epoch": 0.5246277205040092,
"grad_norm": 0.42092597484588623,
"learning_rate": 9.624097223954132e-06,
"loss": 1.2500171661376953,
"step": 458
},
{
"epoch": 0.5269186712485682,
"grad_norm": 1.195084571838379,
"learning_rate": 9.619537086965909e-06,
"loss": 0.9345007538795471,
"step": 460
},
{
"epoch": 0.5292096219931272,
"grad_norm": 0.25558075308799744,
"learning_rate": 9.614950677128618e-06,
"loss": 1.0049808025360107,
"step": 462
},
{
"epoch": 0.5315005727376861,
"grad_norm": 0.31603124737739563,
"learning_rate": 9.610338023692644e-06,
"loss": 1.2373789548873901,
"step": 464
},
{
"epoch": 0.5337915234822451,
"grad_norm": 0.5582301616668701,
"learning_rate": 9.60569915607575e-06,
"loss": 0.9904681444168091,
"step": 466
},
{
"epoch": 0.5360824742268041,
"grad_norm": 0.20000745356082916,
"learning_rate": 9.601034103862875e-06,
"loss": 1.1860427856445312,
"step": 468
},
{
"epoch": 0.5383734249713631,
"grad_norm": 0.40028443932533264,
"learning_rate": 9.596342896805958e-06,
"loss": 1.07786226272583,
"step": 470
},
{
"epoch": 0.5406643757159221,
"grad_norm": 3.4503417015075684,
"learning_rate": 9.591625564823743e-06,
"loss": 0.6197147965431213,
"step": 472
},
{
"epoch": 0.5429553264604811,
"grad_norm": 0.3767026960849762,
"learning_rate": 9.58688213800159e-06,
"loss": 1.141902208328247,
"step": 474
},
{
"epoch": 0.54524627720504,
"grad_norm": 0.3436010181903839,
"learning_rate": 9.58211264659128e-06,
"loss": 1.2863194942474365,
"step": 476
},
{
"epoch": 0.5475372279495991,
"grad_norm": 0.582786500453949,
"learning_rate": 9.577317121010822e-06,
"loss": 1.0963010787963867,
"step": 478
},
{
"epoch": 0.5498281786941581,
"grad_norm": 0.31849485635757446,
"learning_rate": 9.572495591844268e-06,
"loss": 1.269704818725586,
"step": 480
},
{
"epoch": 0.5521191294387171,
"grad_norm": 0.20424437522888184,
"learning_rate": 9.567648089841504e-06,
"loss": 1.0950038433074951,
"step": 482
},
{
"epoch": 0.5544100801832761,
"grad_norm": 0.4071334898471832,
"learning_rate": 9.562774645918067e-06,
"loss": 1.1633511781692505,
"step": 484
},
{
"epoch": 0.5567010309278351,
"grad_norm": 0.14564399421215057,
"learning_rate": 9.557875291154937e-06,
"loss": 1.0785776376724243,
"step": 486
},
{
"epoch": 0.5589919816723941,
"grad_norm": 0.5690705180168152,
"learning_rate": 9.552950056798345e-06,
"loss": 1.397330403327942,
"step": 488
},
{
"epoch": 0.561282932416953,
"grad_norm": 0.715670108795166,
"learning_rate": 9.547998974259573e-06,
"loss": 1.26545250415802,
"step": 490
},
{
"epoch": 0.563573883161512,
"grad_norm": 0.293247789144516,
"learning_rate": 9.543022075114751e-06,
"loss": 1.1600966453552246,
"step": 492
},
{
"epoch": 0.565864833906071,
"grad_norm": 0.6295793652534485,
"learning_rate": 9.538019391104659e-06,
"loss": 1.1647446155548096,
"step": 494
},
{
"epoch": 0.56815578465063,
"grad_norm": 0.49164626002311707,
"learning_rate": 9.532990954134527e-06,
"loss": 1.2747015953063965,
"step": 496
},
{
"epoch": 0.570446735395189,
"grad_norm": 0.9389271140098572,
"learning_rate": 9.527936796273818e-06,
"loss": 1.3396897315979004,
"step": 498
},
{
"epoch": 0.572737686139748,
"grad_norm": 0.6148760318756104,
"learning_rate": 9.522856949756042e-06,
"loss": 1.3895974159240723,
"step": 500
},
{
"epoch": 0.5750286368843069,
"grad_norm": 0.257207989692688,
"learning_rate": 9.517751446978537e-06,
"loss": 1.1982309818267822,
"step": 502
},
{
"epoch": 0.5773195876288659,
"grad_norm": 0.2164141982793808,
"learning_rate": 9.51262032050227e-06,
"loss": 1.1153879165649414,
"step": 504
},
{
"epoch": 0.579610538373425,
"grad_norm": 0.4683654010295868,
"learning_rate": 9.507463603051624e-06,
"loss": 1.2678872346878052,
"step": 506
},
{
"epoch": 0.581901489117984,
"grad_norm": 1.9290564060211182,
"learning_rate": 9.502281327514192e-06,
"loss": 1.3685619831085205,
"step": 508
},
{
"epoch": 0.584192439862543,
"grad_norm": 0.3087227940559387,
"learning_rate": 9.497073526940564e-06,
"loss": 1.1183393001556396,
"step": 510
},
{
"epoch": 0.586483390607102,
"grad_norm": 0.39474597573280334,
"learning_rate": 9.491840234544127e-06,
"loss": 1.1198551654815674,
"step": 512
},
{
"epoch": 0.588774341351661,
"grad_norm": 0.4169851243495941,
"learning_rate": 9.486581483700836e-06,
"loss": 1.087975263595581,
"step": 514
},
{
"epoch": 0.5910652920962199,
"grad_norm": 0.6149902939796448,
"learning_rate": 9.481297307949016e-06,
"loss": 1.3570208549499512,
"step": 516
},
{
"epoch": 0.5933562428407789,
"grad_norm": 0.35397911071777344,
"learning_rate": 9.47598774098914e-06,
"loss": 1.1269992589950562,
"step": 518
},
{
"epoch": 0.5956471935853379,
"grad_norm": 0.9242658615112305,
"learning_rate": 9.470652816683619e-06,
"loss": 1.1659682989120483,
"step": 520
},
{
"epoch": 0.5979381443298969,
"grad_norm": 0.6208882331848145,
"learning_rate": 9.46529256905658e-06,
"loss": 1.3326858282089233,
"step": 522
},
{
"epoch": 0.6002290950744559,
"grad_norm": 0.48353537917137146,
"learning_rate": 9.459907032293654e-06,
"loss": 1.2706080675125122,
"step": 524
},
{
"epoch": 0.6025200458190149,
"grad_norm": 0.35826799273490906,
"learning_rate": 9.454496240741761e-06,
"loss": 0.9778612852096558,
"step": 526
},
{
"epoch": 0.6048109965635738,
"grad_norm": 0.2744974195957184,
"learning_rate": 9.44906022890888e-06,
"loss": 1.1908738613128662,
"step": 528
},
{
"epoch": 0.6071019473081328,
"grad_norm": 0.45797595381736755,
"learning_rate": 9.443599031463838e-06,
"loss": 1.4189608097076416,
"step": 530
},
{
"epoch": 0.6093928980526919,
"grad_norm": 1.2901155948638916,
"learning_rate": 9.438112683236086e-06,
"loss": 0.7630795836448669,
"step": 532
},
{
"epoch": 0.6116838487972509,
"grad_norm": 0.2464115470647812,
"learning_rate": 9.432601219215479e-06,
"loss": 1.1513936519622803,
"step": 534
},
{
"epoch": 0.6139747995418099,
"grad_norm": 0.2185329645872116,
"learning_rate": 9.427064674552046e-06,
"loss": 0.974953293800354,
"step": 536
},
{
"epoch": 0.6162657502863689,
"grad_norm": 2.7072620391845703,
"learning_rate": 9.421503084555778e-06,
"loss": 1.123814582824707,
"step": 538
},
{
"epoch": 0.6185567010309279,
"grad_norm": 0.25285181403160095,
"learning_rate": 9.41591648469639e-06,
"loss": 1.1648335456848145,
"step": 540
},
{
"epoch": 0.6208476517754868,
"grad_norm": 0.6194114685058594,
"learning_rate": 9.410304910603105e-06,
"loss": 0.9610656499862671,
"step": 542
},
{
"epoch": 0.6231386025200458,
"grad_norm": 0.1943773478269577,
"learning_rate": 9.404668398064415e-06,
"loss": 1.216538429260254,
"step": 544
},
{
"epoch": 0.6254295532646048,
"grad_norm": 0.5252053737640381,
"learning_rate": 9.399006983027869e-06,
"loss": 1.2029259204864502,
"step": 546
},
{
"epoch": 0.6277205040091638,
"grad_norm": 1.5968742370605469,
"learning_rate": 9.393320701599826e-06,
"loss": 1.3843748569488525,
"step": 548
},
{
"epoch": 0.6300114547537228,
"grad_norm": 1.534686803817749,
"learning_rate": 9.387609590045243e-06,
"loss": 1.522881269454956,
"step": 550
},
{
"epoch": 0.6323024054982818,
"grad_norm": 0.215438574552536,
"learning_rate": 9.381873684787424e-06,
"loss": 1.085440993309021,
"step": 552
},
{
"epoch": 0.6345933562428407,
"grad_norm": 0.7377695441246033,
"learning_rate": 9.376113022407806e-06,
"loss": 1.3917303085327148,
"step": 554
},
{
"epoch": 0.6368843069873997,
"grad_norm": 0.29986539483070374,
"learning_rate": 9.370327639645715e-06,
"loss": 1.3055980205535889,
"step": 556
},
{
"epoch": 0.6391752577319587,
"grad_norm": 0.5736729502677917,
"learning_rate": 9.364517573398128e-06,
"loss": 1.1343388557434082,
"step": 558
},
{
"epoch": 0.6414662084765178,
"grad_norm": 0.3546328842639923,
"learning_rate": 9.358682860719456e-06,
"loss": 1.2395200729370117,
"step": 560
},
{
"epoch": 0.6437571592210768,
"grad_norm": 0.8976812958717346,
"learning_rate": 9.352823538821286e-06,
"loss": 1.3456403017044067,
"step": 562
},
{
"epoch": 0.6460481099656358,
"grad_norm": 0.5660920143127441,
"learning_rate": 9.346939645072158e-06,
"loss": 1.2232890129089355,
"step": 564
},
{
"epoch": 0.6483390607101948,
"grad_norm": 0.5268477201461792,
"learning_rate": 9.341031216997318e-06,
"loss": 1.3191683292388916,
"step": 566
},
{
"epoch": 0.6506300114547537,
"grad_norm": 0.4400789141654968,
"learning_rate": 9.335098292278487e-06,
"loss": 1.060530185699463,
"step": 568
},
{
"epoch": 0.6529209621993127,
"grad_norm": 0.33317258954048157,
"learning_rate": 9.329140908753612e-06,
"loss": 1.2318658828735352,
"step": 570
},
{
"epoch": 0.6552119129438717,
"grad_norm": 1.2683199644088745,
"learning_rate": 9.323159104416637e-06,
"loss": 1.393534779548645,
"step": 572
},
{
"epoch": 0.6575028636884307,
"grad_norm": 0.47602105140686035,
"learning_rate": 9.31715291741724e-06,
"loss": 1.2734475135803223,
"step": 574
},
{
"epoch": 0.6597938144329897,
"grad_norm": 0.5834298729896545,
"learning_rate": 9.311122386060612e-06,
"loss": 1.109555959701538,
"step": 576
},
{
"epoch": 0.6620847651775487,
"grad_norm": 0.4927360713481903,
"learning_rate": 9.305067548807202e-06,
"loss": 1.1678287982940674,
"step": 578
},
{
"epoch": 0.6643757159221076,
"grad_norm": 0.6562817692756653,
"learning_rate": 9.29898844427247e-06,
"loss": 1.290650486946106,
"step": 580
},
{
"epoch": 0.6666666666666666,
"grad_norm": 0.25812843441963196,
"learning_rate": 9.292885111226647e-06,
"loss": 1.173723816871643,
"step": 582
},
{
"epoch": 0.6689576174112256,
"grad_norm": 0.35074305534362793,
"learning_rate": 9.286757588594479e-06,
"loss": 1.3392133712768555,
"step": 584
},
{
"epoch": 0.6712485681557846,
"grad_norm": 1.0929124355316162,
"learning_rate": 9.28060591545499e-06,
"loss": 0.5634982585906982,
"step": 586
},
{
"epoch": 0.6735395189003437,
"grad_norm": 1.2900134325027466,
"learning_rate": 9.274430131041224e-06,
"loss": 1.2917420864105225,
"step": 588
},
{
"epoch": 0.6758304696449027,
"grad_norm": 0.9884699583053589,
"learning_rate": 9.268230274739993e-06,
"loss": 1.6777942180633545,
"step": 590
},
{
"epoch": 0.6781214203894617,
"grad_norm": 0.3877362906932831,
"learning_rate": 9.262006386091643e-06,
"loss": 1.284725546836853,
"step": 592
},
{
"epoch": 0.6804123711340206,
"grad_norm": 0.44240447878837585,
"learning_rate": 9.255758504789773e-06,
"loss": 1.4049007892608643,
"step": 594
},
{
"epoch": 0.6827033218785796,
"grad_norm": 0.6455042362213135,
"learning_rate": 9.249486670681011e-06,
"loss": 1.3378018140792847,
"step": 596
},
{
"epoch": 0.6849942726231386,
"grad_norm": 2.295257091522217,
"learning_rate": 9.243190923764743e-06,
"loss": 0.9357068538665771,
"step": 598
},
{
"epoch": 0.6872852233676976,
"grad_norm": 0.6263779997825623,
"learning_rate": 9.236871304192857e-06,
"loss": 1.3586604595184326,
"step": 600
},
{
"epoch": 0.6895761741122566,
"grad_norm": 0.2799982726573944,
"learning_rate": 9.2305278522695e-06,
"loss": 1.185795545578003,
"step": 602
},
{
"epoch": 0.6918671248568156,
"grad_norm": 0.5048342347145081,
"learning_rate": 9.224160608450806e-06,
"loss": 1.218984603881836,
"step": 604
},
{
"epoch": 0.6941580756013745,
"grad_norm": 0.337823748588562,
"learning_rate": 9.217769613344647e-06,
"loss": 1.0943036079406738,
"step": 606
},
{
"epoch": 0.6964490263459335,
"grad_norm": 0.2669612169265747,
"learning_rate": 9.211354907710373e-06,
"loss": 1.2221721410751343,
"step": 608
},
{
"epoch": 0.6987399770904925,
"grad_norm": 0.5142564177513123,
"learning_rate": 9.204916532458552e-06,
"loss": 1.2275896072387695,
"step": 610
},
{
"epoch": 0.7010309278350515,
"grad_norm": 0.5470200181007385,
"learning_rate": 9.198454528650702e-06,
"loss": 1.284466028213501,
"step": 612
},
{
"epoch": 0.7033218785796106,
"grad_norm": 0.2950684130191803,
"learning_rate": 9.191968937499041e-06,
"loss": 1.1009457111358643,
"step": 614
},
{
"epoch": 0.7056128293241696,
"grad_norm": 0.29451650381088257,
"learning_rate": 9.185459800366212e-06,
"loss": 1.1608242988586426,
"step": 616
},
{
"epoch": 0.7079037800687286,
"grad_norm": 0.7399727702140808,
"learning_rate": 9.178927158765037e-06,
"loss": 1.1171135902404785,
"step": 618
},
{
"epoch": 0.7101947308132875,
"grad_norm": 0.646661639213562,
"learning_rate": 9.172371054358224e-06,
"loss": 1.211768627166748,
"step": 620
},
{
"epoch": 0.7124856815578465,
"grad_norm": 0.43126344680786133,
"learning_rate": 9.16579152895813e-06,
"loss": 1.0997836589813232,
"step": 622
},
{
"epoch": 0.7147766323024055,
"grad_norm": 0.4805259108543396,
"learning_rate": 9.15918862452648e-06,
"loss": 1.0585315227508545,
"step": 624
},
{
"epoch": 0.7170675830469645,
"grad_norm": 2.9120118618011475,
"learning_rate": 9.152562383174102e-06,
"loss": 0.6636776328086853,
"step": 626
},
{
"epoch": 0.7193585337915235,
"grad_norm": 0.7202600836753845,
"learning_rate": 9.145912847160652e-06,
"loss": 1.3869521617889404,
"step": 628
},
{
"epoch": 0.7216494845360825,
"grad_norm": 0.4563579559326172,
"learning_rate": 9.139240058894358e-06,
"loss": 1.2350819110870361,
"step": 630
},
{
"epoch": 0.7239404352806414,
"grad_norm": 0.4965321719646454,
"learning_rate": 9.132544060931738e-06,
"loss": 1.4039654731750488,
"step": 632
},
{
"epoch": 0.7262313860252004,
"grad_norm": 0.4682888984680176,
"learning_rate": 9.125824895977334e-06,
"loss": 0.7936815023422241,
"step": 634
},
{
"epoch": 0.7285223367697594,
"grad_norm": 1.1855103969573975,
"learning_rate": 9.11908260688344e-06,
"loss": 1.312853455543518,
"step": 636
},
{
"epoch": 0.7308132875143184,
"grad_norm": 0.3794934153556824,
"learning_rate": 9.112317236649822e-06,
"loss": 1.1484042406082153,
"step": 638
},
{
"epoch": 0.7331042382588774,
"grad_norm": 0.5372730493545532,
"learning_rate": 9.105528828423455e-06,
"loss": 1.3890142440795898,
"step": 640
},
{
"epoch": 0.7353951890034365,
"grad_norm": 0.30828553438186646,
"learning_rate": 9.098717425498237e-06,
"loss": 1.122969150543213,
"step": 642
},
{
"epoch": 0.7376861397479955,
"grad_norm": 0.8401303291320801,
"learning_rate": 9.09188307131472e-06,
"loss": 1.3887906074523926,
"step": 644
},
{
"epoch": 0.7399770904925544,
"grad_norm": 0.23984971642494202,
"learning_rate": 9.085025809459826e-06,
"loss": 1.1644659042358398,
"step": 646
},
{
"epoch": 0.7422680412371134,
"grad_norm": 0.3041433095932007,
"learning_rate": 9.078145683666582e-06,
"loss": 1.028432011604309,
"step": 648
},
{
"epoch": 0.7445589919816724,
"grad_norm": 0.6665249466896057,
"learning_rate": 9.071242737813824e-06,
"loss": 1.0644886493682861,
"step": 650
},
{
"epoch": 0.7468499427262314,
"grad_norm": 0.20975077152252197,
"learning_rate": 9.06431701592593e-06,
"loss": 1.1107673645019531,
"step": 652
},
{
"epoch": 0.7491408934707904,
"grad_norm": 0.45067790150642395,
"learning_rate": 9.057368562172535e-06,
"loss": 1.0495760440826416,
"step": 654
},
{
"epoch": 0.7514318442153494,
"grad_norm": 0.651390016078949,
"learning_rate": 9.050397420868246e-06,
"loss": 1.1846449375152588,
"step": 656
},
{
"epoch": 0.7537227949599083,
"grad_norm": 0.49499931931495667,
"learning_rate": 9.043403636472368e-06,
"loss": 0.9847328662872314,
"step": 658
},
{
"epoch": 0.7560137457044673,
"grad_norm": 0.8829602003097534,
"learning_rate": 9.036387253588611e-06,
"loss": 1.4708762168884277,
"step": 660
},
{
"epoch": 0.7583046964490263,
"grad_norm": 0.5728069543838501,
"learning_rate": 9.02934831696481e-06,
"loss": 0.6352672576904297,
"step": 662
},
{
"epoch": 0.7605956471935853,
"grad_norm": 0.23112104833126068,
"learning_rate": 9.022286871492641e-06,
"loss": 1.1570534706115723,
"step": 664
},
{
"epoch": 0.7628865979381443,
"grad_norm": 0.40691277384757996,
"learning_rate": 9.015202962207329e-06,
"loss": 1.2407546043395996,
"step": 666
},
{
"epoch": 0.7651775486827033,
"grad_norm": 0.6645967960357666,
"learning_rate": 9.008096634287372e-06,
"loss": 1.211501121520996,
"step": 668
},
{
"epoch": 0.7674684994272624,
"grad_norm": 0.7196452617645264,
"learning_rate": 9.000967933054236e-06,
"loss": 1.2097508907318115,
"step": 670
},
{
"epoch": 0.7697594501718213,
"grad_norm": 0.2818904221057892,
"learning_rate": 8.993816903972083e-06,
"loss": 1.3354947566986084,
"step": 672
},
{
"epoch": 0.7720504009163803,
"grad_norm": 0.42488253116607666,
"learning_rate": 8.986643592647473e-06,
"loss": 0.9214844703674316,
"step": 674
},
{
"epoch": 0.7743413516609393,
"grad_norm": 1.0121887922286987,
"learning_rate": 8.979448044829068e-06,
"loss": 1.2559878826141357,
"step": 676
},
{
"epoch": 0.7766323024054983,
"grad_norm": 0.48196929693222046,
"learning_rate": 8.972230306407354e-06,
"loss": 0.9865573048591614,
"step": 678
},
{
"epoch": 0.7789232531500573,
"grad_norm": 0.5423112511634827,
"learning_rate": 8.964990423414334e-06,
"loss": 1.2802906036376953,
"step": 680
},
{
"epoch": 0.7812142038946163,
"grad_norm": 0.7373243570327759,
"learning_rate": 8.957728442023243e-06,
"loss": 1.2485854625701904,
"step": 682
},
{
"epoch": 0.7835051546391752,
"grad_norm": 0.7472283244132996,
"learning_rate": 8.95044440854825e-06,
"loss": 1.4083813428878784,
"step": 684
},
{
"epoch": 0.7857961053837342,
"grad_norm": 0.425252228975296,
"learning_rate": 8.943138369444165e-06,
"loss": 1.2072433233261108,
"step": 686
},
{
"epoch": 0.7880870561282932,
"grad_norm": 0.28261420130729675,
"learning_rate": 8.935810371306143e-06,
"loss": 1.0523269176483154,
"step": 688
},
{
"epoch": 0.7903780068728522,
"grad_norm": 0.6173751950263977,
"learning_rate": 8.928460460869383e-06,
"loss": 1.3250004053115845,
"step": 690
},
{
"epoch": 0.7926689576174112,
"grad_norm": 0.7808064818382263,
"learning_rate": 8.921088685008833e-06,
"loss": 1.450615644454956,
"step": 692
},
{
"epoch": 0.7949599083619702,
"grad_norm": 0.28962916135787964,
"learning_rate": 8.913695090738891e-06,
"loss": 1.1594984531402588,
"step": 694
},
{
"epoch": 0.7972508591065293,
"grad_norm": 0.24257132411003113,
"learning_rate": 8.906279725213105e-06,
"loss": 1.0624207258224487,
"step": 696
},
{
"epoch": 0.7995418098510882,
"grad_norm": 0.5376023650169373,
"learning_rate": 8.898842635723868e-06,
"loss": 1.2058117389678955,
"step": 698
},
{
"epoch": 0.8018327605956472,
"grad_norm": 0.32557961344718933,
"learning_rate": 8.891383869702127e-06,
"loss": 1.2131527662277222,
"step": 700
},
{
"epoch": 0.8041237113402062,
"grad_norm": 0.4365437626838684,
"learning_rate": 8.883903474717067e-06,
"loss": 1.1353981494903564,
"step": 702
},
{
"epoch": 0.8064146620847652,
"grad_norm": 1.537105679512024,
"learning_rate": 8.876401498475818e-06,
"loss": 1.2067625522613525,
"step": 704
},
{
"epoch": 0.8087056128293242,
"grad_norm": 0.3798339068889618,
"learning_rate": 8.868877988823148e-06,
"loss": 1.2047245502471924,
"step": 706
},
{
"epoch": 0.8109965635738832,
"grad_norm": 0.4750516414642334,
"learning_rate": 8.861332993741155e-06,
"loss": 1.3064179420471191,
"step": 708
},
{
"epoch": 0.8132875143184422,
"grad_norm": 2.4261434078216553,
"learning_rate": 8.85376656134896e-06,
"loss": 1.1926203966140747,
"step": 710
},
{
"epoch": 0.8155784650630011,
"grad_norm": 0.767508864402771,
"learning_rate": 8.846178739902409e-06,
"loss": 1.511869192123413,
"step": 712
},
{
"epoch": 0.8178694158075601,
"grad_norm": 0.36493149399757385,
"learning_rate": 8.838569577793756e-06,
"loss": 1.021366834640503,
"step": 714
},
{
"epoch": 0.8201603665521191,
"grad_norm": 0.4858986437320709,
"learning_rate": 8.83093912355136e-06,
"loss": 1.1372207403182983,
"step": 716
},
{
"epoch": 0.8224513172966781,
"grad_norm": 0.353909432888031,
"learning_rate": 8.82328742583937e-06,
"loss": 1.1546082496643066,
"step": 718
},
{
"epoch": 0.8247422680412371,
"grad_norm": 0.27604976296424866,
"learning_rate": 8.815614533457419e-06,
"loss": 1.0815863609313965,
"step": 720
},
{
"epoch": 0.827033218785796,
"grad_norm": 0.4933943450450897,
"learning_rate": 8.807920495340313e-06,
"loss": 1.2626594305038452,
"step": 722
},
{
"epoch": 0.8293241695303551,
"grad_norm": 0.4562585651874542,
"learning_rate": 8.800205360557714e-06,
"loss": 1.305412769317627,
"step": 724
},
{
"epoch": 0.8316151202749141,
"grad_norm": 0.844664454460144,
"learning_rate": 8.792469178313835e-06,
"loss": 1.1361956596374512,
"step": 726
},
{
"epoch": 0.8339060710194731,
"grad_norm": 0.5677139759063721,
"learning_rate": 8.784711997947121e-06,
"loss": 1.2585844993591309,
"step": 728
},
{
"epoch": 0.8361970217640321,
"grad_norm": 0.3627360165119171,
"learning_rate": 8.776933868929929e-06,
"loss": 1.2242045402526855,
"step": 730
},
{
"epoch": 0.8384879725085911,
"grad_norm": 0.16927999258041382,
"learning_rate": 8.769134840868228e-06,
"loss": 1.1687601804733276,
"step": 732
},
{
"epoch": 0.8407789232531501,
"grad_norm": 0.33561190962791443,
"learning_rate": 8.761314963501265e-06,
"loss": 1.2507251501083374,
"step": 734
},
{
"epoch": 0.843069873997709,
"grad_norm": 0.22226959466934204,
"learning_rate": 8.753474286701263e-06,
"loss": 1.0193073749542236,
"step": 736
},
{
"epoch": 0.845360824742268,
"grad_norm": 1.0019639730453491,
"learning_rate": 8.74561286047309e-06,
"loss": 1.3349125385284424,
"step": 738
},
{
"epoch": 0.847651775486827,
"grad_norm": 0.4346841275691986,
"learning_rate": 8.737730734953949e-06,
"loss": 1.3727765083312988,
"step": 740
},
{
"epoch": 0.849942726231386,
"grad_norm": 0.1936601996421814,
"learning_rate": 8.729827960413054e-06,
"loss": 1.3784360885620117,
"step": 742
},
{
"epoch": 0.852233676975945,
"grad_norm": 0.36480435729026794,
"learning_rate": 8.721904587251315e-06,
"loss": 1.2935837507247925,
"step": 744
},
{
"epoch": 0.854524627720504,
"grad_norm": 0.38390636444091797,
"learning_rate": 8.713960666001e-06,
"loss": 0.992585301399231,
"step": 746
},
{
"epoch": 0.856815578465063,
"grad_norm": 0.21102914214134216,
"learning_rate": 8.705996247325443e-06,
"loss": 1.1434128284454346,
"step": 748
},
{
"epoch": 0.8591065292096219,
"grad_norm": 0.43868809938430786,
"learning_rate": 8.698011382018687e-06,
"loss": 1.4248082637786865,
"step": 750
},
{
"epoch": 0.861397479954181,
"grad_norm": 0.5018251538276672,
"learning_rate": 8.690006121005187e-06,
"loss": 1.4735299348831177,
"step": 752
},
{
"epoch": 0.86368843069874,
"grad_norm": 0.7337985038757324,
"learning_rate": 8.681980515339464e-06,
"loss": 1.163881778717041,
"step": 754
},
{
"epoch": 0.865979381443299,
"grad_norm": 0.47560250759124756,
"learning_rate": 8.6739346162058e-06,
"loss": 1.0997637510299683,
"step": 756
},
{
"epoch": 0.868270332187858,
"grad_norm": 0.3398919403553009,
"learning_rate": 8.66586847491789e-06,
"loss": 1.0989038944244385,
"step": 758
},
{
"epoch": 0.870561282932417,
"grad_norm": 0.38324427604675293,
"learning_rate": 8.657782142918537e-06,
"loss": 1.2464152574539185,
"step": 760
},
{
"epoch": 0.872852233676976,
"grad_norm": 0.39253002405166626,
"learning_rate": 8.649675671779304e-06,
"loss": 1.1929283142089844,
"step": 762
},
{
"epoch": 0.8751431844215349,
"grad_norm": 0.7560232877731323,
"learning_rate": 8.641549113200198e-06,
"loss": 1.1025712490081787,
"step": 764
},
{
"epoch": 0.8774341351660939,
"grad_norm": 0.29719632863998413,
"learning_rate": 8.633402519009337e-06,
"loss": 1.297807216644287,
"step": 766
},
{
"epoch": 0.8797250859106529,
"grad_norm": 0.5266314744949341,
"learning_rate": 8.625235941162615e-06,
"loss": 1.127463459968567,
"step": 768
},
{
"epoch": 0.8820160366552119,
"grad_norm": 3.540093421936035,
"learning_rate": 8.617049431743376e-06,
"loss": 1.0660855770111084,
"step": 770
},
{
"epoch": 0.8843069873997709,
"grad_norm": 0.38510146737098694,
"learning_rate": 8.60884304296208e-06,
"loss": 0.8022444248199463,
"step": 772
},
{
"epoch": 0.8865979381443299,
"grad_norm": 0.44131770730018616,
"learning_rate": 8.600616827155968e-06,
"loss": 1.2196941375732422,
"step": 774
},
{
"epoch": 0.8888888888888888,
"grad_norm": 0.4722777307033539,
"learning_rate": 8.592370836788738e-06,
"loss": 1.3086035251617432,
"step": 776
},
{
"epoch": 0.8911798396334479,
"grad_norm": 0.491374135017395,
"learning_rate": 8.584105124450192e-06,
"loss": 1.425337791442871,
"step": 778
},
{
"epoch": 0.8934707903780069,
"grad_norm": 1.0334980487823486,
"learning_rate": 8.575819742855918e-06,
"loss": 1.6096653938293457,
"step": 780
},
{
"epoch": 0.8957617411225659,
"grad_norm": 0.19674751162528992,
"learning_rate": 8.567514744846947e-06,
"loss": 1.1552717685699463,
"step": 782
},
{
"epoch": 0.8980526918671249,
"grad_norm": 0.5686236619949341,
"learning_rate": 8.559190183389411e-06,
"loss": 0.6807040572166443,
"step": 784
},
{
"epoch": 0.9003436426116839,
"grad_norm": 0.3094983994960785,
"learning_rate": 8.550846111574216e-06,
"loss": 0.990125298500061,
"step": 786
},
{
"epoch": 0.9026345933562429,
"grad_norm": 1.2661654949188232,
"learning_rate": 8.542482582616694e-06,
"loss": 1.1654900312423706,
"step": 788
},
{
"epoch": 0.9049255441008018,
"grad_norm": 0.20958510041236877,
"learning_rate": 8.53409964985627e-06,
"loss": 1.1051843166351318,
"step": 790
},
{
"epoch": 0.9072164948453608,
"grad_norm": 2.105391502380371,
"learning_rate": 8.525697366756117e-06,
"loss": 1.1716747283935547,
"step": 792
},
{
"epoch": 0.9095074455899198,
"grad_norm": 0.840846836566925,
"learning_rate": 8.51727578690282e-06,
"loss": 1.2360508441925049,
"step": 794
},
{
"epoch": 0.9117983963344788,
"grad_norm": 0.2139447033405304,
"learning_rate": 8.508834964006026e-06,
"loss": 1.1012890338897705,
"step": 796
},
{
"epoch": 0.9140893470790378,
"grad_norm": 0.6380893588066101,
"learning_rate": 8.500374951898111e-06,
"loss": 1.2604258060455322,
"step": 798
},
{
"epoch": 0.9163802978235968,
"grad_norm": 7.733399391174316,
"learning_rate": 8.491895804533834e-06,
"loss": 1.3236706256866455,
"step": 800
},
{
"epoch": 0.9186712485681557,
"grad_norm": 0.38475513458251953,
"learning_rate": 8.483397575989984e-06,
"loss": 1.2977417707443237,
"step": 802
},
{
"epoch": 0.9209621993127147,
"grad_norm": 0.7843623757362366,
"learning_rate": 8.474880320465054e-06,
"loss": 1.3058006763458252,
"step": 804
},
{
"epoch": 0.9232531500572738,
"grad_norm": 0.9258148670196533,
"learning_rate": 8.466344092278874e-06,
"loss": 1.4281094074249268,
"step": 806
},
{
"epoch": 0.9255441008018328,
"grad_norm": 0.456764280796051,
"learning_rate": 8.457788945872278e-06,
"loss": 1.2990665435791016,
"step": 808
},
{
"epoch": 0.9278350515463918,
"grad_norm": 0.9124121069908142,
"learning_rate": 8.449214935806754e-06,
"loss": 1.274019718170166,
"step": 810
},
{
"epoch": 0.9301260022909508,
"grad_norm": 0.8229918479919434,
"learning_rate": 8.440622116764095e-06,
"loss": 1.3842912912368774,
"step": 812
},
{
"epoch": 0.9324169530355098,
"grad_norm": 0.3444792628288269,
"learning_rate": 8.43201054354605e-06,
"loss": 1.054539442062378,
"step": 814
},
{
"epoch": 0.9347079037800687,
"grad_norm": 0.8576228618621826,
"learning_rate": 8.423380271073975e-06,
"loss": 1.0527242422103882,
"step": 816
},
{
"epoch": 0.9369988545246277,
"grad_norm": 1.4201858043670654,
"learning_rate": 8.41473135438848e-06,
"loss": 1.2259652614593506,
"step": 818
},
{
"epoch": 0.9392898052691867,
"grad_norm": 0.32819488644599915,
"learning_rate": 8.406063848649089e-06,
"loss": 1.1133246421813965,
"step": 820
},
{
"epoch": 0.9415807560137457,
"grad_norm": 0.22478768229484558,
"learning_rate": 8.397377809133872e-06,
"loss": 1.0828020572662354,
"step": 822
},
{
"epoch": 0.9438717067583047,
"grad_norm": 0.4433528780937195,
"learning_rate": 8.388673291239098e-06,
"loss": 1.2108690738677979,
"step": 824
},
{
"epoch": 0.9461626575028637,
"grad_norm": 0.4043940603733063,
"learning_rate": 8.379950350478899e-06,
"loss": 1.1333519220352173,
"step": 826
},
{
"epoch": 0.9484536082474226,
"grad_norm": 0.17816071212291718,
"learning_rate": 8.371209042484884e-06,
"loss": 1.1817986965179443,
"step": 828
},
{
"epoch": 0.9507445589919816,
"grad_norm": 1.2147220373153687,
"learning_rate": 8.362449423005811e-06,
"loss": 0.941910445690155,
"step": 830
},
{
"epoch": 0.9530355097365406,
"grad_norm": 0.828575611114502,
"learning_rate": 8.353671547907218e-06,
"loss": 1.2004121541976929,
"step": 832
},
{
"epoch": 0.9553264604810997,
"grad_norm": 0.24938572943210602,
"learning_rate": 8.344875473171072e-06,
"loss": 1.0861434936523438,
"step": 834
},
{
"epoch": 0.9576174112256587,
"grad_norm": 0.2983783483505249,
"learning_rate": 8.33606125489541e-06,
"loss": 1.1088454723358154,
"step": 836
},
{
"epoch": 0.9599083619702177,
"grad_norm": 0.26970595121383667,
"learning_rate": 8.327228949293983e-06,
"loss": 1.179089903831482,
"step": 838
},
{
"epoch": 0.9621993127147767,
"grad_norm": 0.30218416452407837,
"learning_rate": 8.318378612695893e-06,
"loss": 1.1748452186584473,
"step": 840
},
{
"epoch": 0.9644902634593356,
"grad_norm": 0.5073446035385132,
"learning_rate": 8.30951030154524e-06,
"loss": 1.1994376182556152,
"step": 842
},
{
"epoch": 0.9667812142038946,
"grad_norm": 0.4085674285888672,
"learning_rate": 8.300624072400757e-06,
"loss": 1.11344313621521,
"step": 844
},
{
"epoch": 0.9690721649484536,
"grad_norm": 0.511972963809967,
"learning_rate": 8.29171998193545e-06,
"loss": 1.2800376415252686,
"step": 846
},
{
"epoch": 0.9713631156930126,
"grad_norm": 0.9452718496322632,
"learning_rate": 8.28279808693624e-06,
"loss": 1.2644078731536865,
"step": 848
},
{
"epoch": 0.9736540664375716,
"grad_norm": 1.4905763864517212,
"learning_rate": 8.273858444303601e-06,
"loss": 1.2725701332092285,
"step": 850
},
{
"epoch": 0.9759450171821306,
"grad_norm": 0.47297149896621704,
"learning_rate": 8.26490111105119e-06,
"loss": 1.2751691341400146,
"step": 852
},
{
"epoch": 0.9782359679266895,
"grad_norm": 0.3333577811717987,
"learning_rate": 8.25592614430549e-06,
"loss": 1.4911508560180664,
"step": 854
},
{
"epoch": 0.9805269186712485,
"grad_norm": 1.9368016719818115,
"learning_rate": 8.246933601305441e-06,
"loss": 1.2820303440093994,
"step": 856
},
{
"epoch": 0.9828178694158075,
"grad_norm": 0.6824301481246948,
"learning_rate": 8.237923539402083e-06,
"loss": 1.1723785400390625,
"step": 858
},
{
"epoch": 0.9851088201603666,
"grad_norm": 0.1673649251461029,
"learning_rate": 8.228896016058182e-06,
"loss": 1.0617311000823975,
"step": 860
},
{
"epoch": 0.9873997709049256,
"grad_norm": 1.2521167993545532,
"learning_rate": 8.219851088847866e-06,
"loss": 0.845752477645874,
"step": 862
},
{
"epoch": 0.9896907216494846,
"grad_norm": 1.0000566244125366,
"learning_rate": 8.210788815456259e-06,
"loss": 0.5226180553436279,
"step": 864
},
{
"epoch": 0.9919816723940436,
"grad_norm": 0.2924060523509979,
"learning_rate": 8.201709253679113e-06,
"loss": 1.100063443183899,
"step": 866
},
{
"epoch": 0.9942726231386025,
"grad_norm": 0.6394639015197754,
"learning_rate": 8.192612461422436e-06,
"loss": 1.1543686389923096,
"step": 868
},
{
"epoch": 0.9965635738831615,
"grad_norm": 0.6142589449882507,
"learning_rate": 8.18349849670213e-06,
"loss": 1.1854156255722046,
"step": 870
},
{
"epoch": 0.9988545246277205,
"grad_norm": 0.3970600366592407,
"learning_rate": 8.174367417643614e-06,
"loss": 1.2869349718093872,
"step": 872
},
{
"epoch": 1.0011454753722795,
"grad_norm": 0.13177210092544556,
"learning_rate": 8.165219282481454e-06,
"loss": 1.085939645767212,
"step": 874
},
{
"epoch": 1.0034364261168385,
"grad_norm": 0.5160970091819763,
"learning_rate": 8.156054149558997e-06,
"loss": 1.2138652801513672,
"step": 876
},
{
"epoch": 1.0057273768613975,
"grad_norm": 0.2649502456188202,
"learning_rate": 8.146872077327992e-06,
"loss": 1.2543463706970215,
"step": 878
},
{
"epoch": 1.0080183276059564,
"grad_norm": 0.5178881883621216,
"learning_rate": 8.137673124348224e-06,
"loss": 1.2511537075042725,
"step": 880
},
{
"epoch": 1.0103092783505154,
"grad_norm": 0.9306449294090271,
"learning_rate": 8.128457349287134e-06,
"loss": 1.297982931137085,
"step": 882
},
{
"epoch": 1.0126002290950744,
"grad_norm": 1.041662573814392,
"learning_rate": 8.119224810919446e-06,
"loss": 1.1431050300598145,
"step": 884
},
{
"epoch": 1.0148911798396334,
"grad_norm": 0.6204004287719727,
"learning_rate": 8.1099755681268e-06,
"loss": 1.2724859714508057,
"step": 886
},
{
"epoch": 1.0171821305841924,
"grad_norm": 0.36028027534484863,
"learning_rate": 8.10070967989737e-06,
"loss": 1.2920498847961426,
"step": 888
},
{
"epoch": 1.0194730813287514,
"grad_norm": 0.2737988531589508,
"learning_rate": 8.091427205325481e-06,
"loss": 1.1133158206939697,
"step": 890
},
{
"epoch": 1.0217640320733103,
"grad_norm": 0.3814859688282013,
"learning_rate": 8.082128203611245e-06,
"loss": 1.1604249477386475,
"step": 892
},
{
"epoch": 1.0240549828178693,
"grad_norm": 0.7075831890106201,
"learning_rate": 8.07281273406018e-06,
"loss": 1.2947951555252075,
"step": 894
},
{
"epoch": 1.0263459335624283,
"grad_norm": 0.43392547965049744,
"learning_rate": 8.063480856082822e-06,
"loss": 1.3201167583465576,
"step": 896
},
{
"epoch": 1.0286368843069873,
"grad_norm": 0.6384571194648743,
"learning_rate": 8.054132629194363e-06,
"loss": 1.2386162281036377,
"step": 898
},
{
"epoch": 1.0309278350515463,
"grad_norm": 1.1632153987884521,
"learning_rate": 8.044768113014253e-06,
"loss": 1.2521460056304932,
"step": 900
},
{
"epoch": 1.0332187857961055,
"grad_norm": 1.460950255393982,
"learning_rate": 8.03538736726584e-06,
"loss": 0.9340270161628723,
"step": 902
},
{
"epoch": 1.0355097365406645,
"grad_norm": 0.32956740260124207,
"learning_rate": 8.025990451775963e-06,
"loss": 1.191983938217163,
"step": 904
},
{
"epoch": 1.0378006872852235,
"grad_norm": 0.39645904302597046,
"learning_rate": 8.016577426474602e-06,
"loss": 1.1712646484375,
"step": 906
},
{
"epoch": 1.0400916380297824,
"grad_norm": 0.7310532331466675,
"learning_rate": 8.007148351394465e-06,
"loss": 1.208326816558838,
"step": 908
},
{
"epoch": 1.0423825887743414,
"grad_norm": 0.3210288882255554,
"learning_rate": 7.99770328667063e-06,
"loss": 1.0705335140228271,
"step": 910
},
{
"epoch": 1.0446735395189004,
"grad_norm": 0.8159183263778687,
"learning_rate": 7.988242292540144e-06,
"loss": 1.2352807521820068,
"step": 912
},
{
"epoch": 1.0469644902634594,
"grad_norm": 1.189197063446045,
"learning_rate": 7.978765429341651e-06,
"loss": 1.1748485565185547,
"step": 914
},
{
"epoch": 1.0492554410080184,
"grad_norm": 0.30064672231674194,
"learning_rate": 7.969272757514997e-06,
"loss": 1.1746854782104492,
"step": 916
},
{
"epoch": 1.0515463917525774,
"grad_norm": 0.5656450986862183,
"learning_rate": 7.959764337600852e-06,
"loss": 1.2387423515319824,
"step": 918
},
{
"epoch": 1.0538373424971363,
"grad_norm": 0.32169046998023987,
"learning_rate": 7.950240230240323e-06,
"loss": 0.6924710273742676,
"step": 920
},
{
"epoch": 1.0561282932416953,
"grad_norm": 0.7423054575920105,
"learning_rate": 7.94070049617456e-06,
"loss": 1.2082648277282715,
"step": 922
},
{
"epoch": 1.0584192439862543,
"grad_norm": 0.7247334718704224,
"learning_rate": 7.93114519624438e-06,
"loss": 1.2238739728927612,
"step": 924
},
{
"epoch": 1.0607101947308133,
"grad_norm": 0.7445449233055115,
"learning_rate": 7.921574391389874e-06,
"loss": 1.161942720413208,
"step": 926
},
{
"epoch": 1.0630011454753723,
"grad_norm": 0.3218993544578552,
"learning_rate": 7.911988142650008e-06,
"loss": 1.0611815452575684,
"step": 928
},
{
"epoch": 1.0652920962199313,
"grad_norm": 2.6120293140411377,
"learning_rate": 7.902386511162257e-06,
"loss": 1.066215991973877,
"step": 930
},
{
"epoch": 1.0675830469644902,
"grad_norm": 0.4084852933883667,
"learning_rate": 7.892769558162188e-06,
"loss": 1.1941471099853516,
"step": 932
},
{
"epoch": 1.0698739977090492,
"grad_norm": 0.3998633325099945,
"learning_rate": 7.883137344983094e-06,
"loss": 0.7665129899978638,
"step": 934
},
{
"epoch": 1.0721649484536082,
"grad_norm": 0.40679097175598145,
"learning_rate": 7.873489933055586e-06,
"loss": 1.1121021509170532,
"step": 936
},
{
"epoch": 1.0744558991981672,
"grad_norm": 0.6797943711280823,
"learning_rate": 7.863827383907202e-06,
"loss": 1.3281772136688232,
"step": 938
},
{
"epoch": 1.0767468499427262,
"grad_norm": 0.510895848274231,
"learning_rate": 7.85414975916203e-06,
"loss": 1.2209117412567139,
"step": 940
},
{
"epoch": 1.0790378006872852,
"grad_norm": 0.21708233654499054,
"learning_rate": 7.8444571205403e-06,
"loss": 1.1070363521575928,
"step": 942
},
{
"epoch": 1.0813287514318441,
"grad_norm": 1.5223441123962402,
"learning_rate": 7.834749529857991e-06,
"loss": 1.1975922584533691,
"step": 944
},
{
"epoch": 1.0836197021764031,
"grad_norm": 0.400249183177948,
"learning_rate": 7.825027049026448e-06,
"loss": 1.2500934600830078,
"step": 946
},
{
"epoch": 1.0859106529209621,
"grad_norm": 0.48843708634376526,
"learning_rate": 7.81528974005197e-06,
"loss": 1.253622055053711,
"step": 948
},
{
"epoch": 1.088201603665521,
"grad_norm": 0.5512809157371521,
"learning_rate": 7.805537665035435e-06,
"loss": 1.1393404006958008,
"step": 950
},
{
"epoch": 1.09049255441008,
"grad_norm": 0.35912221670150757,
"learning_rate": 7.795770886171885e-06,
"loss": 1.2351994514465332,
"step": 952
},
{
"epoch": 1.0927835051546393,
"grad_norm": 0.2847960293292999,
"learning_rate": 7.785989465750144e-06,
"loss": 1.1068283319473267,
"step": 954
},
{
"epoch": 1.0950744558991983,
"grad_norm": 0.37099456787109375,
"learning_rate": 7.776193466152408e-06,
"loss": 1.2183043956756592,
"step": 956
},
{
"epoch": 1.0973654066437573,
"grad_norm": 0.2889074981212616,
"learning_rate": 7.766382949853856e-06,
"loss": 1.1053792238235474,
"step": 958
},
{
"epoch": 1.0996563573883162,
"grad_norm": 0.23075656592845917,
"learning_rate": 7.756557979422254e-06,
"loss": 1.1999101638793945,
"step": 960
},
{
"epoch": 1.1019473081328752,
"grad_norm": 0.3924773037433624,
"learning_rate": 7.746718617517541e-06,
"loss": 1.232895016670227,
"step": 962
},
{
"epoch": 1.1042382588774342,
"grad_norm": 0.5827606320381165,
"learning_rate": 7.73686492689145e-06,
"loss": 1.258862018585205,
"step": 964
},
{
"epoch": 1.1065292096219932,
"grad_norm": 0.5682635307312012,
"learning_rate": 7.726996970387087e-06,
"loss": 1.1788777112960815,
"step": 966
},
{
"epoch": 1.1088201603665522,
"grad_norm": 0.5773354172706604,
"learning_rate": 7.717114810938548e-06,
"loss": 1.2178382873535156,
"step": 968
},
{
"epoch": 1.1111111111111112,
"grad_norm": 1.0229160785675049,
"learning_rate": 7.707218511570506e-06,
"loss": 0.9783726930618286,
"step": 970
},
{
"epoch": 1.1134020618556701,
"grad_norm": 0.48322662711143494,
"learning_rate": 7.697308135397819e-06,
"loss": 1.293108344078064,
"step": 972
},
{
"epoch": 1.1156930126002291,
"grad_norm": 0.406630277633667,
"learning_rate": 7.687383745625113e-06,
"loss": 0.7449958324432373,
"step": 974
},
{
"epoch": 1.1179839633447881,
"grad_norm": 0.7270604372024536,
"learning_rate": 7.67744540554639e-06,
"loss": 1.3136848211288452,
"step": 976
},
{
"epoch": 1.120274914089347,
"grad_norm": 0.6745506525039673,
"learning_rate": 7.667493178544626e-06,
"loss": 1.030921459197998,
"step": 978
},
{
"epoch": 1.122565864833906,
"grad_norm": 0.5561051368713379,
"learning_rate": 7.65752712809136e-06,
"loss": 1.0987008810043335,
"step": 980
},
{
"epoch": 1.124856815578465,
"grad_norm": 0.7598734498023987,
"learning_rate": 7.64754731774629e-06,
"loss": 1.3208575248718262,
"step": 982
},
{
"epoch": 1.127147766323024,
"grad_norm": 0.5683021545410156,
"learning_rate": 7.637553811156871e-06,
"loss": 1.1968867778778076,
"step": 984
},
{
"epoch": 1.129438717067583,
"grad_norm": 0.41832515597343445,
"learning_rate": 7.627546672057908e-06,
"loss": 1.1648495197296143,
"step": 986
},
{
"epoch": 1.131729667812142,
"grad_norm": 0.6963409185409546,
"learning_rate": 7.617525964271149e-06,
"loss": 1.1976368427276611,
"step": 988
},
{
"epoch": 1.134020618556701,
"grad_norm": 0.3125353753566742,
"learning_rate": 7.607491751704876e-06,
"loss": 1.1840250492095947,
"step": 990
},
{
"epoch": 1.13631156930126,
"grad_norm": 0.18655972182750702,
"learning_rate": 7.5974440983535015e-06,
"loss": 0.9917019009590149,
"step": 992
},
{
"epoch": 1.138602520045819,
"grad_norm": 0.4015827178955078,
"learning_rate": 7.587383068297157e-06,
"loss": 1.1198314428329468,
"step": 994
},
{
"epoch": 1.140893470790378,
"grad_norm": 0.3271810710430145,
"learning_rate": 7.577308725701285e-06,
"loss": 0.9866247177124023,
"step": 996
},
{
"epoch": 1.143184421534937,
"grad_norm": 0.3470871150493622,
"learning_rate": 7.567221134816235e-06,
"loss": 1.1671900749206543,
"step": 998
},
{
"epoch": 1.145475372279496,
"grad_norm": 0.5754879713058472,
"learning_rate": 7.557120359976843e-06,
"loss": 1.1137046813964844,
"step": 1000
},
{
"epoch": 1.147766323024055,
"grad_norm": 0.8881956338882446,
"learning_rate": 7.547006465602026e-06,
"loss": 1.0529272556304932,
"step": 1002
},
{
"epoch": 1.1500572737686139,
"grad_norm": 0.3590288758277893,
"learning_rate": 7.5368795161943835e-06,
"loss": 1.3641059398651123,
"step": 1004
},
{
"epoch": 1.1523482245131729,
"grad_norm": 0.21775324642658234,
"learning_rate": 7.526739576339761e-06,
"loss": 1.063247561454773,
"step": 1006
},
{
"epoch": 1.1546391752577319,
"grad_norm": 0.27955248951911926,
"learning_rate": 7.516586710706862e-06,
"loss": 1.0667047500610352,
"step": 1008
},
{
"epoch": 1.1569301260022908,
"grad_norm": 0.35729917883872986,
"learning_rate": 7.506420984046823e-06,
"loss": 1.1514363288879395,
"step": 1010
},
{
"epoch": 1.1592210767468498,
"grad_norm": 0.4915781021118164,
"learning_rate": 7.496242461192801e-06,
"loss": 1.1422107219696045,
"step": 1012
},
{
"epoch": 1.161512027491409,
"grad_norm": 0.42877474427223206,
"learning_rate": 7.486051207059567e-06,
"loss": 0.9425073862075806,
"step": 1014
},
{
"epoch": 1.163802978235968,
"grad_norm": 0.7864556312561035,
"learning_rate": 7.475847286643081e-06,
"loss": 1.1090211868286133,
"step": 1016
},
{
"epoch": 1.166093928980527,
"grad_norm": 0.28250840306282043,
"learning_rate": 7.46563076502009e-06,
"loss": 1.1747761964797974,
"step": 1018
},
{
"epoch": 1.168384879725086,
"grad_norm": 0.3922346532344818,
"learning_rate": 7.4554017073477e-06,
"loss": 1.0735399723052979,
"step": 1020
},
{
"epoch": 1.170675830469645,
"grad_norm": 0.30916616320610046,
"learning_rate": 7.445160178862977e-06,
"loss": 0.9813652038574219,
"step": 1022
},
{
"epoch": 1.172966781214204,
"grad_norm": 0.5802461504936218,
"learning_rate": 7.434906244882508e-06,
"loss": 0.9689698219299316,
"step": 1024
},
{
"epoch": 1.175257731958763,
"grad_norm": 0.5440467596054077,
"learning_rate": 7.42463997080201e-06,
"loss": 0.9294706583023071,
"step": 1026
},
{
"epoch": 1.177548682703322,
"grad_norm": 0.2637293040752411,
"learning_rate": 7.414361422095894e-06,
"loss": 1.1335558891296387,
"step": 1028
},
{
"epoch": 1.179839633447881,
"grad_norm": 0.40597668290138245,
"learning_rate": 7.404070664316855e-06,
"loss": 1.4880869388580322,
"step": 1030
},
{
"epoch": 1.1821305841924399,
"grad_norm": 1.225714087486267,
"learning_rate": 7.393767763095452e-06,
"loss": 0.6794233322143555,
"step": 1032
},
{
"epoch": 1.1844215349369989,
"grad_norm": 0.4729476571083069,
"learning_rate": 7.383452784139694e-06,
"loss": 1.1326751708984375,
"step": 1034
},
{
"epoch": 1.1867124856815578,
"grad_norm": 0.7771596908569336,
"learning_rate": 7.37312579323461e-06,
"loss": 0.7189903259277344,
"step": 1036
},
{
"epoch": 1.1890034364261168,
"grad_norm": 0.7064245939254761,
"learning_rate": 7.362786856241845e-06,
"loss": 1.1460908651351929,
"step": 1038
},
{
"epoch": 1.1912943871706758,
"grad_norm": 0.6097919344902039,
"learning_rate": 7.3524360390992275e-06,
"loss": 1.0353729724884033,
"step": 1040
},
{
"epoch": 1.1935853379152348,
"grad_norm": 0.4319731295108795,
"learning_rate": 7.342073407820351e-06,
"loss": 0.6625040173530579,
"step": 1042
},
{
"epoch": 1.1958762886597938,
"grad_norm": 0.9910304546356201,
"learning_rate": 7.331699028494161e-06,
"loss": 1.1734858751296997,
"step": 1044
},
{
"epoch": 1.1981672394043528,
"grad_norm": 1.1285514831542969,
"learning_rate": 7.321312967284518e-06,
"loss": 1.3283900022506714,
"step": 1046
},
{
"epoch": 1.2004581901489118,
"grad_norm": 1.028917908668518,
"learning_rate": 7.310915290429799e-06,
"loss": 1.2122178077697754,
"step": 1048
},
{
"epoch": 1.2027491408934707,
"grad_norm": 0.46919187903404236,
"learning_rate": 7.300506064242448e-06,
"loss": 1.1828129291534424,
"step": 1050
},
{
"epoch": 1.2050400916380297,
"grad_norm": 0.3256652355194092,
"learning_rate": 7.290085355108573e-06,
"loss": 1.177816390991211,
"step": 1052
},
{
"epoch": 1.2073310423825887,
"grad_norm": 0.268441766500473,
"learning_rate": 7.279653229487517e-06,
"loss": 1.0987030267715454,
"step": 1054
},
{
"epoch": 1.2096219931271477,
"grad_norm": 0.3454778492450714,
"learning_rate": 7.269209753911426e-06,
"loss": 1.1410846710205078,
"step": 1056
},
{
"epoch": 1.2119129438717067,
"grad_norm": 1.5586289167404175,
"learning_rate": 7.258754994984839e-06,
"loss": 0.9169238805770874,
"step": 1058
},
{
"epoch": 1.2142038946162657,
"grad_norm": 0.24782694876194,
"learning_rate": 7.248289019384255e-06,
"loss": 1.0707839727401733,
"step": 1060
},
{
"epoch": 1.2164948453608249,
"grad_norm": 0.3990442156791687,
"learning_rate": 7.237811893857703e-06,
"loss": 1.201479434967041,
"step": 1062
},
{
"epoch": 1.2187857961053838,
"grad_norm": 0.2834305465221405,
"learning_rate": 7.227323685224329e-06,
"loss": 1.0706955194473267,
"step": 1064
},
{
"epoch": 1.2210767468499428,
"grad_norm": 0.6435824036598206,
"learning_rate": 7.216824460373959e-06,
"loss": 1.0990760326385498,
"step": 1066
},
{
"epoch": 1.2233676975945018,
"grad_norm": 3.215423822402954,
"learning_rate": 7.206314286266676e-06,
"loss": 0.9824307560920715,
"step": 1068
},
{
"epoch": 1.2256586483390608,
"grad_norm": 0.4849179983139038,
"learning_rate": 7.195793229932397e-06,
"loss": 1.1502206325531006,
"step": 1070
},
{
"epoch": 1.2279495990836198,
"grad_norm": 2.8261635303497314,
"learning_rate": 7.185261358470436e-06,
"loss": 1.0585294961929321,
"step": 1072
},
{
"epoch": 1.2302405498281788,
"grad_norm": 0.9895238280296326,
"learning_rate": 7.174718739049087e-06,
"loss": 1.0261123180389404,
"step": 1074
},
{
"epoch": 1.2325315005727377,
"grad_norm": 0.7041395902633667,
"learning_rate": 7.164165438905186e-06,
"loss": 0.9829214811325073,
"step": 1076
},
{
"epoch": 1.2348224513172967,
"grad_norm": 0.5983326435089111,
"learning_rate": 7.153601525343692e-06,
"loss": 1.1627838611602783,
"step": 1078
},
{
"epoch": 1.2371134020618557,
"grad_norm": 0.6902378797531128,
"learning_rate": 7.143027065737247e-06,
"loss": 1.0614171028137207,
"step": 1080
},
{
"epoch": 1.2394043528064147,
"grad_norm": 0.22289074957370758,
"learning_rate": 7.132442127525754e-06,
"loss": 0.9967485070228577,
"step": 1082
},
{
"epoch": 1.2416953035509737,
"grad_norm": 0.3911464512348175,
"learning_rate": 7.121846778215946e-06,
"loss": 1.1069986820220947,
"step": 1084
},
{
"epoch": 1.2439862542955327,
"grad_norm": 0.5967843532562256,
"learning_rate": 7.111241085380951e-06,
"loss": 1.0382379293441772,
"step": 1086
},
{
"epoch": 1.2462772050400917,
"grad_norm": 0.46687158942222595,
"learning_rate": 7.100625116659867e-06,
"loss": 1.1105608940124512,
"step": 1088
},
{
"epoch": 1.2485681557846506,
"grad_norm": 0.17509859800338745,
"learning_rate": 7.089998939757323e-06,
"loss": 1.0866317749023438,
"step": 1090
},
{
"epoch": 1.2508591065292096,
"grad_norm": 0.709657609462738,
"learning_rate": 7.07936262244306e-06,
"loss": 1.2985808849334717,
"step": 1092
},
{
"epoch": 1.2531500572737686,
"grad_norm": 0.298130601644516,
"learning_rate": 7.068716232551484e-06,
"loss": 1.1646193265914917,
"step": 1094
},
{
"epoch": 1.2554410080183276,
"grad_norm": 0.5657819509506226,
"learning_rate": 7.05805983798124e-06,
"loss": 1.226839303970337,
"step": 1096
},
{
"epoch": 1.2577319587628866,
"grad_norm": 0.2218884825706482,
"learning_rate": 7.047393506694784e-06,
"loss": 1.1182987689971924,
"step": 1098
},
{
"epoch": 1.2600229095074456,
"grad_norm": 0.5544705986976624,
"learning_rate": 7.036717306717941e-06,
"loss": 1.0833134651184082,
"step": 1100
},
{
"epoch": 1.2623138602520045,
"grad_norm": 0.8727872371673584,
"learning_rate": 7.026031306139476e-06,
"loss": 1.1887166500091553,
"step": 1102
},
{
"epoch": 1.2646048109965635,
"grad_norm": 0.31878840923309326,
"learning_rate": 7.015335573110655e-06,
"loss": 1.0471493005752563,
"step": 1104
},
{
"epoch": 1.2668957617411225,
"grad_norm": 1.7133678197860718,
"learning_rate": 7.004630175844821e-06,
"loss": 1.2911756038665771,
"step": 1106
},
{
"epoch": 1.2691867124856815,
"grad_norm": 0.4181298017501831,
"learning_rate": 6.9939151826169435e-06,
"loss": 1.1934268474578857,
"step": 1108
},
{
"epoch": 1.2714776632302405,
"grad_norm": 0.4590524137020111,
"learning_rate": 6.9831906617632015e-06,
"loss": 1.0381492376327515,
"step": 1110
},
{
"epoch": 1.2737686139747995,
"grad_norm": 0.35924550890922546,
"learning_rate": 6.972456681680526e-06,
"loss": 1.1387617588043213,
"step": 1112
},
{
"epoch": 1.2760595647193584,
"grad_norm": 0.284288614988327,
"learning_rate": 6.961713310826186e-06,
"loss": 1.2039177417755127,
"step": 1114
},
{
"epoch": 1.2783505154639174,
"grad_norm": 0.5767810344696045,
"learning_rate": 6.950960617717331e-06,
"loss": 1.193761944770813,
"step": 1116
},
{
"epoch": 1.2806414662084764,
"grad_norm": 0.5597623586654663,
"learning_rate": 6.940198670930575e-06,
"loss": 1.1562107801437378,
"step": 1118
},
{
"epoch": 1.2829324169530354,
"grad_norm": 1.776769995689392,
"learning_rate": 6.929427539101542e-06,
"loss": 1.1486637592315674,
"step": 1120
},
{
"epoch": 1.2852233676975944,
"grad_norm": 2.092482089996338,
"learning_rate": 6.918647290924431e-06,
"loss": 0.7264701128005981,
"step": 1122
},
{
"epoch": 1.2875143184421534,
"grad_norm": 0.5090541243553162,
"learning_rate": 6.907857995151593e-06,
"loss": 1.331784725189209,
"step": 1124
},
{
"epoch": 1.2898052691867126,
"grad_norm": 1.2451262474060059,
"learning_rate": 6.897059720593072e-06,
"loss": 1.2851134538650513,
"step": 1126
},
{
"epoch": 1.2920962199312716,
"grad_norm": 1.4176346063613892,
"learning_rate": 6.886252536116178e-06,
"loss": 0.7333021759986877,
"step": 1128
},
{
"epoch": 1.2943871706758305,
"grad_norm": 0.6894351840019226,
"learning_rate": 6.875436510645046e-06,
"loss": 1.1808993816375732,
"step": 1130
},
{
"epoch": 1.2966781214203895,
"grad_norm": 0.5326879620552063,
"learning_rate": 6.864611713160195e-06,
"loss": 0.4958106279373169,
"step": 1132
},
{
"epoch": 1.2989690721649485,
"grad_norm": 0.5978541374206543,
"learning_rate": 6.853778212698085e-06,
"loss": 1.1578189134597778,
"step": 1134
},
{
"epoch": 1.3012600229095075,
"grad_norm": 0.27673229575157166,
"learning_rate": 6.842936078350687e-06,
"loss": 1.1419076919555664,
"step": 1136
},
{
"epoch": 1.3035509736540665,
"grad_norm": 0.6036549806594849,
"learning_rate": 6.832085379265032e-06,
"loss": 1.0068280696868896,
"step": 1138
},
{
"epoch": 1.3058419243986255,
"grad_norm": 1.3319073915481567,
"learning_rate": 6.821226184642769e-06,
"loss": 1.1533417701721191,
"step": 1140
},
{
"epoch": 1.3081328751431844,
"grad_norm": 1.2838786840438843,
"learning_rate": 6.810358563739741e-06,
"loss": 0.9974702000617981,
"step": 1142
},
{
"epoch": 1.3104238258877434,
"grad_norm": 0.25137636065483093,
"learning_rate": 6.799482585865517e-06,
"loss": 1.1108713150024414,
"step": 1144
},
{
"epoch": 1.3127147766323024,
"grad_norm": 0.9687121510505676,
"learning_rate": 6.788598320382969e-06,
"loss": 0.7017291784286499,
"step": 1146
},
{
"epoch": 1.3150057273768614,
"grad_norm": 0.6507570743560791,
"learning_rate": 6.777705836707827e-06,
"loss": 0.6254103183746338,
"step": 1148
},
{
"epoch": 1.3172966781214204,
"grad_norm": 0.24653151631355286,
"learning_rate": 6.766805204308232e-06,
"loss": 1.0639063119888306,
"step": 1150
},
{
"epoch": 1.3195876288659794,
"grad_norm": 0.8183524012565613,
"learning_rate": 6.755896492704292e-06,
"loss": 0.7704499363899231,
"step": 1152
},
{
"epoch": 1.3218785796105383,
"grad_norm": 1.3019349575042725,
"learning_rate": 6.7449797714676446e-06,
"loss": 1.0398871898651123,
"step": 1154
},
{
"epoch": 1.3241695303550973,
"grad_norm": 0.5958473682403564,
"learning_rate": 6.734055110221004e-06,
"loss": 0.9920108914375305,
"step": 1156
},
{
"epoch": 1.3264604810996563,
"grad_norm": 0.34555068612098694,
"learning_rate": 6.72312257863773e-06,
"loss": 0.7964114546775818,
"step": 1158
},
{
"epoch": 1.3287514318442153,
"grad_norm": 1.3161815404891968,
"learning_rate": 6.712182246441372e-06,
"loss": 1.170238971710205,
"step": 1160
},
{
"epoch": 1.3310423825887743,
"grad_norm": 0.4162517189979553,
"learning_rate": 6.701234183405228e-06,
"loss": 1.1864508390426636,
"step": 1162
},
{
"epoch": 1.3333333333333333,
"grad_norm": 0.682020902633667,
"learning_rate": 6.690278459351907e-06,
"loss": 1.0800316333770752,
"step": 1164
},
{
"epoch": 1.3356242840778922,
"grad_norm": 0.8428366780281067,
"learning_rate": 6.679315144152867e-06,
"loss": 1.0788335800170898,
"step": 1166
},
{
"epoch": 1.3379152348224514,
"grad_norm": 0.5951038599014282,
"learning_rate": 6.6683443077279885e-06,
"loss": 0.9050284028053284,
"step": 1168
},
{
"epoch": 1.3402061855670104,
"grad_norm": 14.55526351928711,
"learning_rate": 6.6573660200451155e-06,
"loss": 0.8427594900131226,
"step": 1170
},
{
"epoch": 1.3424971363115694,
"grad_norm": 0.4018735885620117,
"learning_rate": 6.646380351119612e-06,
"loss": 0.9908223152160645,
"step": 1172
},
{
"epoch": 1.3447880870561284,
"grad_norm": 0.44170042872428894,
"learning_rate": 6.6353873710139185e-06,
"loss": 1.1631572246551514,
"step": 1174
},
{
"epoch": 1.3470790378006874,
"grad_norm": 0.4431830644607544,
"learning_rate": 6.624387149837105e-06,
"loss": 1.1010160446166992,
"step": 1176
},
{
"epoch": 1.3493699885452464,
"grad_norm": 0.20745538175106049,
"learning_rate": 6.613379757744419e-06,
"loss": 1.083961844444275,
"step": 1178
},
{
"epoch": 1.3516609392898054,
"grad_norm": 0.23701387643814087,
"learning_rate": 6.602365264936843e-06,
"loss": 0.91713547706604,
"step": 1180
},
{
"epoch": 1.3539518900343643,
"grad_norm": 0.24008503556251526,
"learning_rate": 6.5913437416606485e-06,
"loss": 1.1394959688186646,
"step": 1182
},
{
"epoch": 1.3562428407789233,
"grad_norm": 0.5425772070884705,
"learning_rate": 6.5803152582069365e-06,
"loss": 1.0138827562332153,
"step": 1184
},
{
"epoch": 1.3585337915234823,
"grad_norm": 0.5500049591064453,
"learning_rate": 6.569279884911205e-06,
"loss": 1.1640450954437256,
"step": 1186
},
{
"epoch": 1.3608247422680413,
"grad_norm": 0.23986561596393585,
"learning_rate": 6.558237692152889e-06,
"loss": 1.0996382236480713,
"step": 1188
},
{
"epoch": 1.3631156930126003,
"grad_norm": 0.33864185214042664,
"learning_rate": 6.547188750354917e-06,
"loss": 1.1096324920654297,
"step": 1190
},
{
"epoch": 1.3654066437571593,
"grad_norm": 1.339373230934143,
"learning_rate": 6.536133129983261e-06,
"loss": 1.0554709434509277,
"step": 1192
},
{
"epoch": 1.3676975945017182,
"grad_norm": 0.5389803051948547,
"learning_rate": 6.5250709015464834e-06,
"loss": 1.1650443077087402,
"step": 1194
},
{
"epoch": 1.3699885452462772,
"grad_norm": 0.42033982276916504,
"learning_rate": 6.5140021355952935e-06,
"loss": 1.177642583847046,
"step": 1196
},
{
"epoch": 1.3722794959908362,
"grad_norm": 0.9180698990821838,
"learning_rate": 6.502926902722092e-06,
"loss": 1.0504584312438965,
"step": 1198
},
{
"epoch": 1.3745704467353952,
"grad_norm": 0.48793044686317444,
"learning_rate": 6.4918452735605245e-06,
"loss": 1.1541026830673218,
"step": 1200
},
{
"epoch": 1.3768613974799542,
"grad_norm": 0.14638172090053558,
"learning_rate": 6.4807573187850295e-06,
"loss": 0.9544370174407959,
"step": 1202
},
{
"epoch": 1.3791523482245132,
"grad_norm": 1.1175258159637451,
"learning_rate": 6.469663109110389e-06,
"loss": 0.94257652759552,
"step": 1204
},
{
"epoch": 1.3814432989690721,
"grad_norm": 0.5161193609237671,
"learning_rate": 6.458562715291273e-06,
"loss": 1.1248518228530884,
"step": 1206
},
{
"epoch": 1.3837342497136311,
"grad_norm": 0.3805331289768219,
"learning_rate": 6.4474562081217975e-06,
"loss": 0.9945235252380371,
"step": 1208
},
{
"epoch": 1.38602520045819,
"grad_norm": 0.4004988670349121,
"learning_rate": 6.436343658435059e-06,
"loss": 1.2267534732818604,
"step": 1210
},
{
"epoch": 1.388316151202749,
"grad_norm": 0.5942703485488892,
"learning_rate": 6.4252251371026984e-06,
"loss": 1.1504127979278564,
"step": 1212
},
{
"epoch": 1.390607101947308,
"grad_norm": 0.6351091265678406,
"learning_rate": 6.414100715034437e-06,
"loss": 1.2171190977096558,
"step": 1214
},
{
"epoch": 1.392898052691867,
"grad_norm": 0.3700306713581085,
"learning_rate": 6.4029704631776334e-06,
"loss": 1.153677225112915,
"step": 1216
},
{
"epoch": 1.395189003436426,
"grad_norm": 0.5730115175247192,
"learning_rate": 6.3918344525168176e-06,
"loss": 1.2128088474273682,
"step": 1218
},
{
"epoch": 1.397479954180985,
"grad_norm": 0.8086961507797241,
"learning_rate": 6.380692754073257e-06,
"loss": 1.1901538372039795,
"step": 1220
},
{
"epoch": 1.399770904925544,
"grad_norm": 0.2177879959344864,
"learning_rate": 6.3695454389044885e-06,
"loss": 1.050641655921936,
"step": 1222
},
{
"epoch": 1.402061855670103,
"grad_norm": 0.48566994071006775,
"learning_rate": 6.35839257810387e-06,
"loss": 1.1169625520706177,
"step": 1224
},
{
"epoch": 1.404352806414662,
"grad_norm": 0.5073995590209961,
"learning_rate": 6.347234242800131e-06,
"loss": 1.0769405364990234,
"step": 1226
},
{
"epoch": 1.406643757159221,
"grad_norm": 1.046706199645996,
"learning_rate": 6.336070504156907e-06,
"loss": 0.498112291097641,
"step": 1228
},
{
"epoch": 1.40893470790378,
"grad_norm": 0.571991503238678,
"learning_rate": 6.324901433372307e-06,
"loss": 1.0861037969589233,
"step": 1230
},
{
"epoch": 1.411225658648339,
"grad_norm": 0.3702922463417053,
"learning_rate": 6.313727101678433e-06,
"loss": 1.1421905755996704,
"step": 1232
},
{
"epoch": 1.413516609392898,
"grad_norm": 0.8220073580741882,
"learning_rate": 6.302547580340949e-06,
"loss": 1.2322626113891602,
"step": 1234
},
{
"epoch": 1.4158075601374571,
"grad_norm": 0.5405969619750977,
"learning_rate": 6.291362940658612e-06,
"loss": 0.9657451510429382,
"step": 1236
},
{
"epoch": 1.418098510882016,
"grad_norm": 1.5954561233520508,
"learning_rate": 6.2801732539628205e-06,
"loss": 1.1020958423614502,
"step": 1238
},
{
"epoch": 1.420389461626575,
"grad_norm": 0.42759695649147034,
"learning_rate": 6.2689785916171656e-06,
"loss": 1.098386526107788,
"step": 1240
},
{
"epoch": 1.422680412371134,
"grad_norm": 0.19068364799022675,
"learning_rate": 6.257779025016967e-06,
"loss": 1.1080601215362549,
"step": 1242
},
{
"epoch": 1.424971363115693,
"grad_norm": 0.6803228855133057,
"learning_rate": 6.246574625588824e-06,
"loss": 1.1234514713287354,
"step": 1244
},
{
"epoch": 1.427262313860252,
"grad_norm": 0.3850613236427307,
"learning_rate": 6.235365464790158e-06,
"loss": 1.191713571548462,
"step": 1246
},
{
"epoch": 1.429553264604811,
"grad_norm": 0.5191882252693176,
"learning_rate": 6.224151614108755e-06,
"loss": 1.113290548324585,
"step": 1248
},
{
"epoch": 1.43184421534937,
"grad_norm": 0.32683122158050537,
"learning_rate": 6.212933145062313e-06,
"loss": 0.977238655090332,
"step": 1250
},
{
"epoch": 1.434135166093929,
"grad_norm": 0.3372647762298584,
"learning_rate": 6.201710129197984e-06,
"loss": 1.0865776538848877,
"step": 1252
},
{
"epoch": 1.436426116838488,
"grad_norm": 1.7988835573196411,
"learning_rate": 6.190482638091917e-06,
"loss": 1.2454376220703125,
"step": 1254
},
{
"epoch": 1.438717067583047,
"grad_norm": 0.5903953313827515,
"learning_rate": 6.179250743348801e-06,
"loss": 0.7004958391189575,
"step": 1256
},
{
"epoch": 1.441008018327606,
"grad_norm": 1.3094062805175781,
"learning_rate": 6.1680145166014135e-06,
"loss": 1.1775736808776855,
"step": 1258
},
{
"epoch": 1.443298969072165,
"grad_norm": 0.15861783921718597,
"learning_rate": 6.156774029510158e-06,
"loss": 1.0173115730285645,
"step": 1260
},
{
"epoch": 1.445589919816724,
"grad_norm": 0.31646618247032166,
"learning_rate": 6.145529353762608e-06,
"loss": 1.0695950984954834,
"step": 1262
},
{
"epoch": 1.447880870561283,
"grad_norm": 0.2128850221633911,
"learning_rate": 6.1342805610730515e-06,
"loss": 1.0297033786773682,
"step": 1264
},
{
"epoch": 1.4501718213058419,
"grad_norm": 0.43015554547309875,
"learning_rate": 6.123027723182034e-06,
"loss": 1.0786964893341064,
"step": 1266
},
{
"epoch": 1.4524627720504009,
"grad_norm": 0.5410537719726562,
"learning_rate": 6.111770911855895e-06,
"loss": 1.2960965633392334,
"step": 1268
},
{
"epoch": 1.4547537227949598,
"grad_norm": 0.40598443150520325,
"learning_rate": 6.100510198886324e-06,
"loss": 1.150607943534851,
"step": 1270
},
{
"epoch": 1.4570446735395188,
"grad_norm": 0.25774961709976196,
"learning_rate": 6.089245656089882e-06,
"loss": 1.0670363903045654,
"step": 1272
},
{
"epoch": 1.4593356242840778,
"grad_norm": 1.2430238723754883,
"learning_rate": 6.077977355307564e-06,
"loss": 0.929009199142456,
"step": 1274
},
{
"epoch": 1.461626575028637,
"grad_norm": 0.819111704826355,
"learning_rate": 6.066705368404332e-06,
"loss": 1.0384223461151123,
"step": 1276
},
{
"epoch": 1.463917525773196,
"grad_norm": 0.6329976320266724,
"learning_rate": 6.0554297672686515e-06,
"loss": 1.146793246269226,
"step": 1278
},
{
"epoch": 1.466208476517755,
"grad_norm": 0.5748082399368286,
"learning_rate": 6.044150623812041e-06,
"loss": 1.21446692943573,
"step": 1280
},
{
"epoch": 1.468499427262314,
"grad_norm": 0.37093842029571533,
"learning_rate": 6.032868009968611e-06,
"loss": 0.6608114242553711,
"step": 1282
},
{
"epoch": 1.470790378006873,
"grad_norm": 0.4401775598526001,
"learning_rate": 6.021581997694604e-06,
"loss": 1.1975040435791016,
"step": 1284
},
{
"epoch": 1.473081328751432,
"grad_norm": 0.198956698179245,
"learning_rate": 6.010292658967937e-06,
"loss": 0.8298950791358948,
"step": 1286
},
{
"epoch": 1.475372279495991,
"grad_norm": 1.1646336317062378,
"learning_rate": 5.999000065787741e-06,
"loss": 1.121684193611145,
"step": 1288
},
{
"epoch": 1.47766323024055,
"grad_norm": 0.19620084762573242,
"learning_rate": 5.987704290173904e-06,
"loss": 1.113356113433838,
"step": 1290
},
{
"epoch": 1.479954180985109,
"grad_norm": 0.8701173663139343,
"learning_rate": 5.976405404166609e-06,
"loss": 1.0445246696472168,
"step": 1292
},
{
"epoch": 1.4822451317296679,
"grad_norm": 0.8870877623558044,
"learning_rate": 5.965103479825874e-06,
"loss": 1.3073999881744385,
"step": 1294
},
{
"epoch": 1.4845360824742269,
"grad_norm": 0.6979129910469055,
"learning_rate": 5.953798589231102e-06,
"loss": 1.0925014019012451,
"step": 1296
},
{
"epoch": 1.4868270332187858,
"grad_norm": 0.3692026734352112,
"learning_rate": 5.942490804480605e-06,
"loss": 0.999457597732544,
"step": 1298
},
{
"epoch": 1.4891179839633448,
"grad_norm": 0.6890159845352173,
"learning_rate": 5.931180197691155e-06,
"loss": 1.0261509418487549,
"step": 1300
},
{
"epoch": 1.4914089347079038,
"grad_norm": 0.40297093987464905,
"learning_rate": 5.919866840997528e-06,
"loss": 0.8686540126800537,
"step": 1302
},
{
"epoch": 1.4936998854524628,
"grad_norm": 1.5102224349975586,
"learning_rate": 5.908550806552027e-06,
"loss": 1.269791841506958,
"step": 1304
},
{
"epoch": 1.4959908361970218,
"grad_norm": 0.3148708939552307,
"learning_rate": 5.897232166524044e-06,
"loss": 1.1037763357162476,
"step": 1306
},
{
"epoch": 1.4982817869415808,
"grad_norm": 1.3338543176651,
"learning_rate": 5.885910993099581e-06,
"loss": 0.8658826351165771,
"step": 1308
},
{
"epoch": 1.5005727376861397,
"grad_norm": 0.594698429107666,
"learning_rate": 5.874587358480798e-06,
"loss": 1.0999151468276978,
"step": 1310
},
{
"epoch": 1.5028636884306987,
"grad_norm": 0.8571167588233948,
"learning_rate": 5.863261334885553e-06,
"loss": 1.3375155925750732,
"step": 1312
},
{
"epoch": 1.5051546391752577,
"grad_norm": 0.3977302610874176,
"learning_rate": 5.851932994546941e-06,
"loss": 1.1367615461349487,
"step": 1314
},
{
"epoch": 1.5074455899198167,
"grad_norm": 1.0233980417251587,
"learning_rate": 5.840602409712831e-06,
"loss": 1.1128208637237549,
"step": 1316
},
{
"epoch": 1.5097365406643757,
"grad_norm": 0.42678776383399963,
"learning_rate": 5.829269652645404e-06,
"loss": 0.9502562880516052,
"step": 1318
},
{
"epoch": 1.5120274914089347,
"grad_norm": 0.6844682097434998,
"learning_rate": 5.817934795620702e-06,
"loss": 0.842738151550293,
"step": 1320
},
{
"epoch": 1.5143184421534936,
"grad_norm": 0.17597302794456482,
"learning_rate": 5.8065979109281515e-06,
"loss": 1.1172516345977783,
"step": 1322
},
{
"epoch": 1.5166093928980526,
"grad_norm": 1.0520700216293335,
"learning_rate": 5.795259070870114e-06,
"loss": 1.019298791885376,
"step": 1324
},
{
"epoch": 1.5189003436426116,
"grad_norm": 0.301179975271225,
"learning_rate": 5.783918347761424e-06,
"loss": 1.0366371870040894,
"step": 1326
},
{
"epoch": 1.5211912943871706,
"grad_norm": 1.4107294082641602,
"learning_rate": 5.77257581392892e-06,
"loss": 1.0230910778045654,
"step": 1328
},
{
"epoch": 1.5234822451317296,
"grad_norm": 3.041111946105957,
"learning_rate": 5.761231541710994e-06,
"loss": 1.078089714050293,
"step": 1330
},
{
"epoch": 1.5257731958762886,
"grad_norm": 0.39890924096107483,
"learning_rate": 5.7498856034571235e-06,
"loss": 0.9173415303230286,
"step": 1332
},
{
"epoch": 1.5280641466208476,
"grad_norm": 1.2413023710250854,
"learning_rate": 5.7385380715274075e-06,
"loss": 0.7650175094604492,
"step": 1334
},
{
"epoch": 1.5303550973654065,
"grad_norm": 0.40077197551727295,
"learning_rate": 5.727189018292115e-06,
"loss": 1.0416836738586426,
"step": 1336
},
{
"epoch": 1.5326460481099655,
"grad_norm": 1.2296018600463867,
"learning_rate": 5.715838516131212e-06,
"loss": 0.7319103479385376,
"step": 1338
},
{
"epoch": 1.5349369988545245,
"grad_norm": 0.5661179423332214,
"learning_rate": 5.704486637433907e-06,
"loss": 1.1215949058532715,
"step": 1340
},
{
"epoch": 1.5372279495990835,
"grad_norm": 1.0080679655075073,
"learning_rate": 5.6931334545981876e-06,
"loss": 1.0732643604278564,
"step": 1342
},
{
"epoch": 1.5395189003436425,
"grad_norm": 0.46233344078063965,
"learning_rate": 5.68177904003036e-06,
"loss": 0.5498278737068176,
"step": 1344
},
{
"epoch": 1.5418098510882015,
"grad_norm": 1.1307381391525269,
"learning_rate": 5.670423466144585e-06,
"loss": 0.8027646541595459,
"step": 1346
},
{
"epoch": 1.5441008018327604,
"grad_norm": 0.2457914799451828,
"learning_rate": 5.659066805362416e-06,
"loss": 1.099556803703308,
"step": 1348
},
{
"epoch": 1.5463917525773194,
"grad_norm": 0.1576959639787674,
"learning_rate": 5.647709130112339e-06,
"loss": 1.0494180917739868,
"step": 1350
},
{
"epoch": 1.5486827033218786,
"grad_norm": 1.4281489849090576,
"learning_rate": 5.6363505128293105e-06,
"loss": 1.0572658777236938,
"step": 1352
},
{
"epoch": 1.5509736540664376,
"grad_norm": 0.3911938965320587,
"learning_rate": 5.624991025954296e-06,
"loss": 1.112137794494629,
"step": 1354
},
{
"epoch": 1.5532646048109966,
"grad_norm": 0.7160745859146118,
"learning_rate": 5.613630741933801e-06,
"loss": 1.0537481307983398,
"step": 1356
},
{
"epoch": 1.5555555555555556,
"grad_norm": 1.9618110656738281,
"learning_rate": 5.602269733219422e-06,
"loss": 0.8826972246170044,
"step": 1358
},
{
"epoch": 1.5578465063001146,
"grad_norm": 0.6684516668319702,
"learning_rate": 5.590908072267376e-06,
"loss": 1.1781320571899414,
"step": 1360
},
{
"epoch": 1.5601374570446735,
"grad_norm": 0.9931756854057312,
"learning_rate": 5.579545831538036e-06,
"loss": 1.003735065460205,
"step": 1362
},
{
"epoch": 1.5624284077892325,
"grad_norm": 0.35055479407310486,
"learning_rate": 5.568183083495476e-06,
"loss": 1.188180923461914,
"step": 1364
},
{
"epoch": 1.5647193585337915,
"grad_norm": 0.2840157747268677,
"learning_rate": 5.556819900607004e-06,
"loss": 1.0069973468780518,
"step": 1366
},
{
"epoch": 1.5670103092783505,
"grad_norm": 0.2898299992084503,
"learning_rate": 5.545456355342702e-06,
"loss": 1.0996860265731812,
"step": 1368
},
{
"epoch": 1.5693012600229095,
"grad_norm": 0.44947919249534607,
"learning_rate": 5.53409252017496e-06,
"loss": 1.120392918586731,
"step": 1370
},
{
"epoch": 1.5715922107674685,
"grad_norm": 1.1559861898422241,
"learning_rate": 5.522728467578024e-06,
"loss": 1.3419742584228516,
"step": 1372
},
{
"epoch": 1.5738831615120275,
"grad_norm": 0.7148109674453735,
"learning_rate": 5.51136427002752e-06,
"loss": 1.1241408586502075,
"step": 1374
},
{
"epoch": 1.5761741122565864,
"grad_norm": 1.7949758768081665,
"learning_rate": 5.500000000000001e-06,
"loss": 1.0823068618774414,
"step": 1376
},
{
"epoch": 1.5784650630011456,
"grad_norm": 0.5439109206199646,
"learning_rate": 5.488635729972482e-06,
"loss": 1.0603818893432617,
"step": 1378
},
{
"epoch": 1.5807560137457046,
"grad_norm": 0.6118595004081726,
"learning_rate": 5.477271532421978e-06,
"loss": 1.1041834354400635,
"step": 1380
},
{
"epoch": 1.5830469644902636,
"grad_norm": 0.6366990804672241,
"learning_rate": 5.465907479825041e-06,
"loss": 1.2534537315368652,
"step": 1382
},
{
"epoch": 1.5853379152348226,
"grad_norm": 0.3186343014240265,
"learning_rate": 5.454543644657302e-06,
"loss": 1.00857412815094,
"step": 1384
},
{
"epoch": 1.5876288659793816,
"grad_norm": 0.23584212362766266,
"learning_rate": 5.4431800993929985e-06,
"loss": 0.9454631805419922,
"step": 1386
},
{
"epoch": 1.5899198167239406,
"grad_norm": 0.7114871740341187,
"learning_rate": 5.431816916504526e-06,
"loss": 1.1222310066223145,
"step": 1388
},
{
"epoch": 1.5922107674684995,
"grad_norm": 0.3654209077358246,
"learning_rate": 5.420454168461966e-06,
"loss": 1.1215158700942993,
"step": 1390
},
{
"epoch": 1.5945017182130585,
"grad_norm": 0.7768158912658691,
"learning_rate": 5.4090919277326255e-06,
"loss": 0.8392388820648193,
"step": 1392
},
{
"epoch": 1.5967926689576175,
"grad_norm": 0.2861047387123108,
"learning_rate": 5.39773026678058e-06,
"loss": 0.7805435061454773,
"step": 1394
},
{
"epoch": 1.5990836197021765,
"grad_norm": 0.6634432673454285,
"learning_rate": 5.386369258066201e-06,
"loss": 1.295003890991211,
"step": 1396
},
{
"epoch": 1.6013745704467355,
"grad_norm": 0.3201494812965393,
"learning_rate": 5.3750089740457075e-06,
"loss": 1.061468243598938,
"step": 1398
},
{
"epoch": 1.6036655211912945,
"grad_norm": 0.5713483095169067,
"learning_rate": 5.363649487170691e-06,
"loss": 0.9963769912719727,
"step": 1400
},
{
"epoch": 1.6059564719358534,
"grad_norm": 0.9836020469665527,
"learning_rate": 5.352290869887663e-06,
"loss": 1.194199800491333,
"step": 1402
},
{
"epoch": 1.6082474226804124,
"grad_norm": 0.1987893432378769,
"learning_rate": 5.340933194637586e-06,
"loss": 0.6283943057060242,
"step": 1404
},
{
"epoch": 1.6105383734249714,
"grad_norm": 0.580379843711853,
"learning_rate": 5.329576533855415e-06,
"loss": 1.2401604652404785,
"step": 1406
},
{
"epoch": 1.6128293241695304,
"grad_norm": 0.48753857612609863,
"learning_rate": 5.3182209599696415e-06,
"loss": 0.9093868732452393,
"step": 1408
},
{
"epoch": 1.6151202749140894,
"grad_norm": 1.1758943796157837,
"learning_rate": 5.306866545401813e-06,
"loss": 1.1071420907974243,
"step": 1410
},
{
"epoch": 1.6174112256586484,
"grad_norm": 0.4928566515445709,
"learning_rate": 5.295513362566096e-06,
"loss": 0.7807958722114563,
"step": 1412
},
{
"epoch": 1.6197021764032073,
"grad_norm": 0.4523963928222656,
"learning_rate": 5.28416148386879e-06,
"loss": 1.0439610481262207,
"step": 1414
},
{
"epoch": 1.6219931271477663,
"grad_norm": 0.7955020070075989,
"learning_rate": 5.272810981707886e-06,
"loss": 1.0640166997909546,
"step": 1416
},
{
"epoch": 1.6242840778923253,
"grad_norm": 0.7474223375320435,
"learning_rate": 5.261461928472593e-06,
"loss": 0.889735758304596,
"step": 1418
},
{
"epoch": 1.6265750286368843,
"grad_norm": 0.3920307457447052,
"learning_rate": 5.250114396542877e-06,
"loss": 1.061119794845581,
"step": 1420
},
{
"epoch": 1.6288659793814433,
"grad_norm": 1.0079574584960938,
"learning_rate": 5.238768458289007e-06,
"loss": 1.3186607360839844,
"step": 1422
},
{
"epoch": 1.6311569301260023,
"grad_norm": 0.4474199712276459,
"learning_rate": 5.227424186071081e-06,
"loss": 1.0147289037704468,
"step": 1424
},
{
"epoch": 1.6334478808705613,
"grad_norm": 0.480244904756546,
"learning_rate": 5.216081652238579e-06,
"loss": 1.1268162727355957,
"step": 1426
},
{
"epoch": 1.6357388316151202,
"grad_norm": 0.9012577533721924,
"learning_rate": 5.204740929129887e-06,
"loss": 1.120902180671692,
"step": 1428
},
{
"epoch": 1.6380297823596792,
"grad_norm": 0.21147756278514862,
"learning_rate": 5.193402089071852e-06,
"loss": 1.1125855445861816,
"step": 1430
},
{
"epoch": 1.6403207331042382,
"grad_norm": 0.3258548378944397,
"learning_rate": 5.1820652043793e-06,
"loss": 1.011163592338562,
"step": 1432
},
{
"epoch": 1.6426116838487972,
"grad_norm": 0.39005911350250244,
"learning_rate": 5.1707303473545955e-06,
"loss": 1.2018420696258545,
"step": 1434
},
{
"epoch": 1.6449026345933562,
"grad_norm": 0.5742803812026978,
"learning_rate": 5.1593975902871705e-06,
"loss": 1.1393346786499023,
"step": 1436
},
{
"epoch": 1.6471935853379152,
"grad_norm": 0.41290178894996643,
"learning_rate": 5.1480670054530605e-06,
"loss": 1.1971912384033203,
"step": 1438
},
{
"epoch": 1.6494845360824741,
"grad_norm": 0.23765195906162262,
"learning_rate": 5.136738665114449e-06,
"loss": 1.047823190689087,
"step": 1440
},
{
"epoch": 1.6517754868270331,
"grad_norm": 0.7021539211273193,
"learning_rate": 5.125412641519204e-06,
"loss": 1.1116011142730713,
"step": 1442
},
{
"epoch": 1.654066437571592,
"grad_norm": 1.0828512907028198,
"learning_rate": 5.114089006900422e-06,
"loss": 1.3231321573257446,
"step": 1444
},
{
"epoch": 1.656357388316151,
"grad_norm": 0.4457782208919525,
"learning_rate": 5.102767833475958e-06,
"loss": 1.1356236934661865,
"step": 1446
},
{
"epoch": 1.65864833906071,
"grad_norm": 0.862468421459198,
"learning_rate": 5.091449193447974e-06,
"loss": 1.0704690217971802,
"step": 1448
},
{
"epoch": 1.660939289805269,
"grad_norm": 0.3140993118286133,
"learning_rate": 5.080133159002474e-06,
"loss": 0.9936932325363159,
"step": 1450
},
{
"epoch": 1.663230240549828,
"grad_norm": 0.4344022274017334,
"learning_rate": 5.068819802308845e-06,
"loss": 1.1224863529205322,
"step": 1452
},
{
"epoch": 1.665521191294387,
"grad_norm": 0.5366426706314087,
"learning_rate": 5.057509195519398e-06,
"loss": 1.1559295654296875,
"step": 1454
},
{
"epoch": 1.667812142038946,
"grad_norm": 0.6061311960220337,
"learning_rate": 5.046201410768899e-06,
"loss": 1.1576176881790161,
"step": 1456
},
{
"epoch": 1.670103092783505,
"grad_norm": 0.9727919697761536,
"learning_rate": 5.034896520174126e-06,
"loss": 0.7988566160202026,
"step": 1458
},
{
"epoch": 1.6723940435280642,
"grad_norm": 0.1961892545223236,
"learning_rate": 5.023594595833393e-06,
"loss": 0.4671269953250885,
"step": 1460
},
{
"epoch": 1.6746849942726232,
"grad_norm": 0.8168536424636841,
"learning_rate": 5.0122957098260975e-06,
"loss": 1.3973238468170166,
"step": 1462
},
{
"epoch": 1.6769759450171822,
"grad_norm": 0.5047257542610168,
"learning_rate": 5.0009999342122606e-06,
"loss": 1.2853825092315674,
"step": 1464
},
{
"epoch": 1.6792668957617412,
"grad_norm": 0.6044178605079651,
"learning_rate": 4.989707341032064e-06,
"loss": 1.174715280532837,
"step": 1466
},
{
"epoch": 1.6815578465063001,
"grad_norm": 0.46799567341804504,
"learning_rate": 4.978418002305399e-06,
"loss": 1.2637076377868652,
"step": 1468
},
{
"epoch": 1.6838487972508591,
"grad_norm": 0.27708613872528076,
"learning_rate": 4.967131990031391e-06,
"loss": 1.092246174812317,
"step": 1470
},
{
"epoch": 1.686139747995418,
"grad_norm": 0.5905908346176147,
"learning_rate": 4.955849376187961e-06,
"loss": 0.8725787997245789,
"step": 1472
},
{
"epoch": 1.688430698739977,
"grad_norm": 0.49280673265457153,
"learning_rate": 4.944570232731351e-06,
"loss": 1.1290966272354126,
"step": 1474
},
{
"epoch": 1.690721649484536,
"grad_norm": 0.6904608011245728,
"learning_rate": 4.9332946315956695e-06,
"loss": 0.8955701589584351,
"step": 1476
},
{
"epoch": 1.693012600229095,
"grad_norm": 0.6627287268638611,
"learning_rate": 4.922022644692438e-06,
"loss": 1.1323654651641846,
"step": 1478
},
{
"epoch": 1.695303550973654,
"grad_norm": 0.7109958529472351,
"learning_rate": 4.910754343910121e-06,
"loss": 1.025420904159546,
"step": 1480
},
{
"epoch": 1.697594501718213,
"grad_norm": 0.43384861946105957,
"learning_rate": 4.89948980111368e-06,
"loss": 1.111706256866455,
"step": 1482
},
{
"epoch": 1.699885452462772,
"grad_norm": 0.5754409432411194,
"learning_rate": 4.888229088144106e-06,
"loss": 1.1376874446868896,
"step": 1484
},
{
"epoch": 1.702176403207331,
"grad_norm": 0.31369054317474365,
"learning_rate": 4.8769722768179686e-06,
"loss": 1.0659782886505127,
"step": 1486
},
{
"epoch": 1.7044673539518902,
"grad_norm": 1.1461002826690674,
"learning_rate": 4.86571943892695e-06,
"loss": 1.1083794832229614,
"step": 1488
},
{
"epoch": 1.7067583046964492,
"grad_norm": 0.5363646149635315,
"learning_rate": 4.854470646237392e-06,
"loss": 1.103102684020996,
"step": 1490
},
{
"epoch": 1.7090492554410082,
"grad_norm": 0.22202357649803162,
"learning_rate": 4.843225970489843e-06,
"loss": 0.7583179473876953,
"step": 1492
},
{
"epoch": 1.7113402061855671,
"grad_norm": 0.19881503283977509,
"learning_rate": 4.831985483398587e-06,
"loss": 1.2146061658859253,
"step": 1494
},
{
"epoch": 1.7136311569301261,
"grad_norm": 0.6612064838409424,
"learning_rate": 4.820749256651202e-06,
"loss": 0.949057936668396,
"step": 1496
},
{
"epoch": 1.7159221076746851,
"grad_norm": 0.25396284461021423,
"learning_rate": 4.809517361908086e-06,
"loss": 1.0635230541229248,
"step": 1498
},
{
"epoch": 1.718213058419244,
"grad_norm": 0.2346191257238388,
"learning_rate": 4.798289870802018e-06,
"loss": 0.6289651393890381,
"step": 1500
},
{
"epoch": 1.720504009163803,
"grad_norm": 0.8989472389221191,
"learning_rate": 4.787066854937689e-06,
"loss": 1.1217808723449707,
"step": 1502
},
{
"epoch": 1.722794959908362,
"grad_norm": 2.643550395965576,
"learning_rate": 4.775848385891245e-06,
"loss": 1.2906912565231323,
"step": 1504
},
{
"epoch": 1.725085910652921,
"grad_norm": 0.731056272983551,
"learning_rate": 4.7646345352098445e-06,
"loss": 0.6282181739807129,
"step": 1506
},
{
"epoch": 1.72737686139748,
"grad_norm": 0.3717769682407379,
"learning_rate": 4.753425374411177e-06,
"loss": 1.0892517566680908,
"step": 1508
},
{
"epoch": 1.729667812142039,
"grad_norm": 0.2676532566547394,
"learning_rate": 4.742220974983036e-06,
"loss": 1.0730390548706055,
"step": 1510
},
{
"epoch": 1.731958762886598,
"grad_norm": 0.16937556862831116,
"learning_rate": 4.731021408382837e-06,
"loss": 1.120283603668213,
"step": 1512
},
{
"epoch": 1.734249713631157,
"grad_norm": 0.30678778886795044,
"learning_rate": 4.7198267460371826e-06,
"loss": 1.369347095489502,
"step": 1514
},
{
"epoch": 1.736540664375716,
"grad_norm": 0.4600529670715332,
"learning_rate": 4.70863705934139e-06,
"loss": 1.0413298606872559,
"step": 1516
},
{
"epoch": 1.738831615120275,
"grad_norm": 0.4162696301937103,
"learning_rate": 4.697452419659052e-06,
"loss": 1.2059823274612427,
"step": 1518
},
{
"epoch": 1.741122565864834,
"grad_norm": 0.36466091871261597,
"learning_rate": 4.686272898321568e-06,
"loss": 1.0452512502670288,
"step": 1520
},
{
"epoch": 1.743413516609393,
"grad_norm": 0.4611301124095917,
"learning_rate": 4.675098566627695e-06,
"loss": 0.9318735599517822,
"step": 1522
},
{
"epoch": 1.745704467353952,
"grad_norm": 0.6732238531112671,
"learning_rate": 4.6639294958430946e-06,
"loss": 0.9963729977607727,
"step": 1524
},
{
"epoch": 1.747995418098511,
"grad_norm": 0.310858815908432,
"learning_rate": 4.652765757199872e-06,
"loss": 1.0314257144927979,
"step": 1526
},
{
"epoch": 1.7502863688430699,
"grad_norm": 0.2551518678665161,
"learning_rate": 4.641607421896132e-06,
"loss": 0.9765444993972778,
"step": 1528
},
{
"epoch": 1.7525773195876289,
"grad_norm": 0.890846848487854,
"learning_rate": 4.630454561095513e-06,
"loss": 0.8637576699256897,
"step": 1530
},
{
"epoch": 1.7548682703321878,
"grad_norm": 0.3232835829257965,
"learning_rate": 4.619307245926743e-06,
"loss": 1.139219880104065,
"step": 1532
},
{
"epoch": 1.7571592210767468,
"grad_norm": 0.35957878828048706,
"learning_rate": 4.608165547483183e-06,
"loss": 1.202222466468811,
"step": 1534
},
{
"epoch": 1.7594501718213058,
"grad_norm": 2.1918275356292725,
"learning_rate": 4.597029536822368e-06,
"loss": 0.6219435930252075,
"step": 1536
},
{
"epoch": 1.7617411225658648,
"grad_norm": 0.4178512692451477,
"learning_rate": 4.585899284965563e-06,
"loss": 1.0911495685577393,
"step": 1538
},
{
"epoch": 1.7640320733104238,
"grad_norm": 0.5344310402870178,
"learning_rate": 4.574774862897302e-06,
"loss": 1.1106891632080078,
"step": 1540
},
{
"epoch": 1.7663230240549828,
"grad_norm": 0.41900160908699036,
"learning_rate": 4.563656341564941e-06,
"loss": 1.0348694324493408,
"step": 1542
},
{
"epoch": 1.7686139747995417,
"grad_norm": 2.1765964031219482,
"learning_rate": 4.552543791878205e-06,
"loss": 1.2949724197387695,
"step": 1544
},
{
"epoch": 1.7709049255441007,
"grad_norm": 0.28183940052986145,
"learning_rate": 4.541437284708727e-06,
"loss": 0.9392028450965881,
"step": 1546
},
{
"epoch": 1.7731958762886597,
"grad_norm": 0.3175741732120514,
"learning_rate": 4.530336890889613e-06,
"loss": 1.0502426624298096,
"step": 1548
},
{
"epoch": 1.7754868270332187,
"grad_norm": 1.4262926578521729,
"learning_rate": 4.519242681214971e-06,
"loss": 0.7205455303192139,
"step": 1550
},
{
"epoch": 1.7777777777777777,
"grad_norm": 1.2114626169204712,
"learning_rate": 4.508154726439478e-06,
"loss": 1.0237972736358643,
"step": 1552
},
{
"epoch": 1.7800687285223367,
"grad_norm": 0.8481336832046509,
"learning_rate": 4.49707309727791e-06,
"loss": 1.1520320177078247,
"step": 1554
},
{
"epoch": 1.7823596792668956,
"grad_norm": 0.3496226966381073,
"learning_rate": 4.485997864404709e-06,
"loss": 1.1129924058914185,
"step": 1556
},
{
"epoch": 1.7846506300114546,
"grad_norm": 0.2375352829694748,
"learning_rate": 4.474929098453519e-06,
"loss": 1.211329460144043,
"step": 1558
},
{
"epoch": 1.7869415807560136,
"grad_norm": 0.2744106650352478,
"learning_rate": 4.463866870016741e-06,
"loss": 1.0723626613616943,
"step": 1560
},
{
"epoch": 1.7892325315005726,
"grad_norm": 1.0329090356826782,
"learning_rate": 4.452811249645085e-06,
"loss": 1.088670015335083,
"step": 1562
},
{
"epoch": 1.7915234822451316,
"grad_norm": 0.6205560564994812,
"learning_rate": 4.441762307847113e-06,
"loss": 1.2151033878326416,
"step": 1564
},
{
"epoch": 1.7938144329896906,
"grad_norm": 0.6296006441116333,
"learning_rate": 4.430720115088797e-06,
"loss": 1.1723359823226929,
"step": 1566
},
{
"epoch": 1.7961053837342495,
"grad_norm": 1.2101175785064697,
"learning_rate": 4.419684741793066e-06,
"loss": 0.9560754299163818,
"step": 1568
},
{
"epoch": 1.7983963344788088,
"grad_norm": 0.33595010638237,
"learning_rate": 4.408656258339355e-06,
"loss": 1.0018739700317383,
"step": 1570
},
{
"epoch": 1.8006872852233677,
"grad_norm": 0.34481048583984375,
"learning_rate": 4.397634735063158e-06,
"loss": 1.0918338298797607,
"step": 1572
},
{
"epoch": 1.8029782359679267,
"grad_norm": 0.506317138671875,
"learning_rate": 4.386620242255583e-06,
"loss": 1.082423448562622,
"step": 1574
},
{
"epoch": 1.8052691867124857,
"grad_norm": 1.5229862928390503,
"learning_rate": 4.375612850162897e-06,
"loss": 1.20133638381958,
"step": 1576
},
{
"epoch": 1.8075601374570447,
"grad_norm": 0.4049842357635498,
"learning_rate": 4.364612628986082e-06,
"loss": 0.9879454970359802,
"step": 1578
},
{
"epoch": 1.8098510882016037,
"grad_norm": 0.7496724128723145,
"learning_rate": 4.35361964888039e-06,
"loss": 0.9724379777908325,
"step": 1580
},
{
"epoch": 1.8121420389461627,
"grad_norm": 0.7204214930534363,
"learning_rate": 4.342633979954887e-06,
"loss": 1.1547467708587646,
"step": 1582
},
{
"epoch": 1.8144329896907216,
"grad_norm": 1.05171537399292,
"learning_rate": 4.331655692272011e-06,
"loss": 1.0399408340454102,
"step": 1584
},
{
"epoch": 1.8167239404352806,
"grad_norm": 0.36879026889801025,
"learning_rate": 4.320684855847135e-06,
"loss": 1.2286815643310547,
"step": 1586
},
{
"epoch": 1.8190148911798396,
"grad_norm": 0.27584725618362427,
"learning_rate": 4.309721540648094e-06,
"loss": 0.946727991104126,
"step": 1588
},
{
"epoch": 1.8213058419243986,
"grad_norm": 0.2803502678871155,
"learning_rate": 4.298765816594773e-06,
"loss": 1.0936319828033447,
"step": 1590
},
{
"epoch": 1.8235967926689576,
"grad_norm": 0.46186119318008423,
"learning_rate": 4.2878177535586294e-06,
"loss": 1.0281469821929932,
"step": 1592
},
{
"epoch": 1.8258877434135166,
"grad_norm": 0.8824283480644226,
"learning_rate": 4.276877421362273e-06,
"loss": 1.1012415885925293,
"step": 1594
},
{
"epoch": 1.8281786941580758,
"grad_norm": 0.563686192035675,
"learning_rate": 4.265944889778998e-06,
"loss": 1.2015641927719116,
"step": 1596
},
{
"epoch": 1.8304696449026348,
"grad_norm": 0.45275604724884033,
"learning_rate": 4.255020228532357e-06,
"loss": 1.168736457824707,
"step": 1598
},
{
"epoch": 1.8327605956471937,
"grad_norm": 0.5189002156257629,
"learning_rate": 4.244103507295709e-06,
"loss": 0.8564615249633789,
"step": 1600
},
{
"epoch": 1.8350515463917527,
"grad_norm": 0.4436849355697632,
"learning_rate": 4.233194795691767e-06,
"loss": 1.1107850074768066,
"step": 1602
},
{
"epoch": 1.8373424971363117,
"grad_norm": 0.5029271841049194,
"learning_rate": 4.222294163292173e-06,
"loss": 1.1380796432495117,
"step": 1604
},
{
"epoch": 1.8396334478808707,
"grad_norm": 0.38482290506362915,
"learning_rate": 4.211401679617031e-06,
"loss": 1.0681747198104858,
"step": 1606
},
{
"epoch": 1.8419243986254297,
"grad_norm": 0.3398834466934204,
"learning_rate": 4.200517414134487e-06,
"loss": 1.1440093517303467,
"step": 1608
},
{
"epoch": 1.8442153493699887,
"grad_norm": 0.3168809115886688,
"learning_rate": 4.1896414362602615e-06,
"loss": 0.9699110984802246,
"step": 1610
},
{
"epoch": 1.8465063001145476,
"grad_norm": 0.608744204044342,
"learning_rate": 4.178773815357231e-06,
"loss": 1.3050799369812012,
"step": 1612
},
{
"epoch": 1.8487972508591066,
"grad_norm": 1.0546008348464966,
"learning_rate": 4.167914620734971e-06,
"loss": 1.372516393661499,
"step": 1614
},
{
"epoch": 1.8510882016036656,
"grad_norm": 0.6382185220718384,
"learning_rate": 4.157063921649314e-06,
"loss": 1.0993999242782593,
"step": 1616
},
{
"epoch": 1.8533791523482246,
"grad_norm": 0.24312907457351685,
"learning_rate": 4.146221787301916e-06,
"loss": 1.1152019500732422,
"step": 1618
},
{
"epoch": 1.8556701030927836,
"grad_norm": 0.5500279664993286,
"learning_rate": 4.135388286839806e-06,
"loss": 1.0518808364868164,
"step": 1620
},
{
"epoch": 1.8579610538373426,
"grad_norm": 0.6796494126319885,
"learning_rate": 4.1245634893549545e-06,
"loss": 1.1548529863357544,
"step": 1622
},
{
"epoch": 1.8602520045819015,
"grad_norm": 0.5995921492576599,
"learning_rate": 4.113747463883823e-06,
"loss": 1.352916955947876,
"step": 1624
},
{
"epoch": 1.8625429553264605,
"grad_norm": 0.24284528195858002,
"learning_rate": 4.102940279406928e-06,
"loss": 1.0819270610809326,
"step": 1626
},
{
"epoch": 1.8648339060710195,
"grad_norm": 0.6232653260231018,
"learning_rate": 4.092142004848408e-06,
"loss": 1.1328476667404175,
"step": 1628
},
{
"epoch": 1.8671248568155785,
"grad_norm": 0.27039143443107605,
"learning_rate": 4.08135270907557e-06,
"loss": 1.0456500053405762,
"step": 1630
},
{
"epoch": 1.8694158075601375,
"grad_norm": 0.5682404637336731,
"learning_rate": 4.070572460898462e-06,
"loss": 1.0884917974472046,
"step": 1632
},
{
"epoch": 1.8717067583046965,
"grad_norm": 0.28413209319114685,
"learning_rate": 4.059801329069426e-06,
"loss": 1.0192196369171143,
"step": 1634
},
{
"epoch": 1.8739977090492554,
"grad_norm": 0.20320376753807068,
"learning_rate": 4.04903938228267e-06,
"loss": 1.1282374858856201,
"step": 1636
},
{
"epoch": 1.8762886597938144,
"grad_norm": 0.5271807909011841,
"learning_rate": 4.038286689173817e-06,
"loss": 1.1059315204620361,
"step": 1638
},
{
"epoch": 1.8785796105383734,
"grad_norm": 0.18544262647628784,
"learning_rate": 4.027543318319474e-06,
"loss": 1.027941346168518,
"step": 1640
},
{
"epoch": 1.8808705612829324,
"grad_norm": 0.6330454349517822,
"learning_rate": 4.0168093382368e-06,
"loss": 1.1752293109893799,
"step": 1642
},
{
"epoch": 1.8831615120274914,
"grad_norm": 0.4557496905326843,
"learning_rate": 4.006084817383056e-06,
"loss": 0.3644358515739441,
"step": 1644
},
{
"epoch": 1.8854524627720504,
"grad_norm": 1.3597580194473267,
"learning_rate": 3.995369824155181e-06,
"loss": 1.1257150173187256,
"step": 1646
},
{
"epoch": 1.8877434135166093,
"grad_norm": 0.4406583607196808,
"learning_rate": 3.984664426889346e-06,
"loss": 1.1774303913116455,
"step": 1648
},
{
"epoch": 1.8900343642611683,
"grad_norm": 0.43711021542549133,
"learning_rate": 3.973968693860527e-06,
"loss": 1.2031350135803223,
"step": 1650
},
{
"epoch": 1.8923253150057273,
"grad_norm": 0.4777853190898895,
"learning_rate": 3.96328269328206e-06,
"loss": 1.2918949127197266,
"step": 1652
},
{
"epoch": 1.8946162657502863,
"grad_norm": 0.6269339323043823,
"learning_rate": 3.952606493305216e-06,
"loss": 1.4281535148620605,
"step": 1654
},
{
"epoch": 1.8969072164948453,
"grad_norm": 0.3071335256099701,
"learning_rate": 3.941940162018761e-06,
"loss": 1.010391116142273,
"step": 1656
},
{
"epoch": 1.8991981672394043,
"grad_norm": 0.18624217808246613,
"learning_rate": 3.931283767448518e-06,
"loss": 0.5205022096633911,
"step": 1658
},
{
"epoch": 1.9014891179839633,
"grad_norm": 1.4488636255264282,
"learning_rate": 3.920637377556942e-06,
"loss": 0.844072699546814,
"step": 1660
},
{
"epoch": 1.9037800687285222,
"grad_norm": 0.5701979994773865,
"learning_rate": 3.910001060242677e-06,
"loss": 0.9334324598312378,
"step": 1662
},
{
"epoch": 1.9060710194730812,
"grad_norm": 0.7097718119621277,
"learning_rate": 3.899374883340137e-06,
"loss": 1.1183058023452759,
"step": 1664
},
{
"epoch": 1.9083619702176402,
"grad_norm": 0.3661045730113983,
"learning_rate": 3.88875891461905e-06,
"loss": 0.9330714344978333,
"step": 1666
},
{
"epoch": 1.9106529209621992,
"grad_norm": 0.29769042134284973,
"learning_rate": 3.878153221784054e-06,
"loss": 1.1027047634124756,
"step": 1668
},
{
"epoch": 1.9129438717067582,
"grad_norm": 0.8713205456733704,
"learning_rate": 3.867557872474248e-06,
"loss": 1.0835906267166138,
"step": 1670
},
{
"epoch": 1.9152348224513172,
"grad_norm": 0.4693801701068878,
"learning_rate": 3.856972934262755e-06,
"loss": 1.1927740573883057,
"step": 1672
},
{
"epoch": 1.9175257731958761,
"grad_norm": 0.4635125696659088,
"learning_rate": 3.84639847465631e-06,
"loss": 1.2767350673675537,
"step": 1674
},
{
"epoch": 1.9198167239404351,
"grad_norm": 0.48501482605934143,
"learning_rate": 3.835834561094815e-06,
"loss": 1.130952000617981,
"step": 1676
},
{
"epoch": 1.9221076746849943,
"grad_norm": 0.9517636895179749,
"learning_rate": 3.825281260950916e-06,
"loss": 1.3384122848510742,
"step": 1678
},
{
"epoch": 1.9243986254295533,
"grad_norm": 1.5027124881744385,
"learning_rate": 3.814738641529566e-06,
"loss": 1.2897770404815674,
"step": 1680
},
{
"epoch": 1.9266895761741123,
"grad_norm": 0.17051805555820465,
"learning_rate": 3.8042067700676044e-06,
"loss": 0.9062872529029846,
"step": 1682
},
{
"epoch": 1.9289805269186713,
"grad_norm": 0.3399532437324524,
"learning_rate": 3.793685713733326e-06,
"loss": 1.1948344707489014,
"step": 1684
},
{
"epoch": 1.9312714776632303,
"grad_norm": 1.0268810987472534,
"learning_rate": 3.7831755396260416e-06,
"loss": 1.1116868257522583,
"step": 1686
},
{
"epoch": 1.9335624284077892,
"grad_norm": 0.25544682145118713,
"learning_rate": 3.772676314775674e-06,
"loss": 0.9785124063491821,
"step": 1688
},
{
"epoch": 1.9358533791523482,
"grad_norm": 0.2780124545097351,
"learning_rate": 3.762188106142298e-06,
"loss": 0.8251289129257202,
"step": 1690
},
{
"epoch": 1.9381443298969072,
"grad_norm": 0.6381999254226685,
"learning_rate": 3.751710980615748e-06,
"loss": 1.154158353805542,
"step": 1692
},
{
"epoch": 1.9404352806414662,
"grad_norm": 0.36244654655456543,
"learning_rate": 3.741245005015162e-06,
"loss": 0.9876756072044373,
"step": 1694
},
{
"epoch": 1.9427262313860252,
"grad_norm": 0.5184593796730042,
"learning_rate": 3.7307902460885746e-06,
"loss": 1.1275906562805176,
"step": 1696
},
{
"epoch": 1.9450171821305842,
"grad_norm": 0.51296067237854,
"learning_rate": 3.720346770512486e-06,
"loss": 1.025963544845581,
"step": 1698
},
{
"epoch": 1.9473081328751431,
"grad_norm": 0.5093493461608887,
"learning_rate": 3.709914644891427e-06,
"loss": 1.1869707107543945,
"step": 1700
},
{
"epoch": 1.9495990836197021,
"grad_norm": 2.1370980739593506,
"learning_rate": 3.6994939357575533e-06,
"loss": 0.6620631217956543,
"step": 1702
},
{
"epoch": 1.9518900343642611,
"grad_norm": 0.5478183031082153,
"learning_rate": 3.689084709570202e-06,
"loss": 1.123410701751709,
"step": 1704
},
{
"epoch": 1.9541809851088203,
"grad_norm": 0.4446258842945099,
"learning_rate": 3.6786870327154832e-06,
"loss": 1.1439985036849976,
"step": 1706
},
{
"epoch": 1.9564719358533793,
"grad_norm": 0.5509153008460999,
"learning_rate": 3.6683009715058416e-06,
"loss": 1.0780513286590576,
"step": 1708
},
{
"epoch": 1.9587628865979383,
"grad_norm": 0.48575830459594727,
"learning_rate": 3.657926592179649e-06,
"loss": 1.0788171291351318,
"step": 1710
},
{
"epoch": 1.9610538373424973,
"grad_norm": 0.21410536766052246,
"learning_rate": 3.647563960900774e-06,
"loss": 1.091663122177124,
"step": 1712
},
{
"epoch": 1.9633447880870563,
"grad_norm": 0.25092241168022156,
"learning_rate": 3.637213143758156e-06,
"loss": 1.0151268243789673,
"step": 1714
},
{
"epoch": 1.9656357388316152,
"grad_norm": 0.7077304124832153,
"learning_rate": 3.626874206765392e-06,
"loss": 1.1096687316894531,
"step": 1716
},
{
"epoch": 1.9679266895761742,
"grad_norm": 0.4164002537727356,
"learning_rate": 3.616547215860309e-06,
"loss": 1.0841827392578125,
"step": 1718
},
{
"epoch": 1.9702176403207332,
"grad_norm": 0.8496370911598206,
"learning_rate": 3.606232236904549e-06,
"loss": 1.2367123365402222,
"step": 1720
},
{
"epoch": 1.9725085910652922,
"grad_norm": 0.4818964898586273,
"learning_rate": 3.595929335683146e-06,
"loss": 0.9790089130401611,
"step": 1722
},
{
"epoch": 1.9747995418098512,
"grad_norm": 0.33810457587242126,
"learning_rate": 3.585638577904107e-06,
"loss": 1.145442247390747,
"step": 1724
},
{
"epoch": 1.9770904925544102,
"grad_norm": 1.0560132265090942,
"learning_rate": 3.5753600291979917e-06,
"loss": 1.3760294914245605,
"step": 1726
},
{
"epoch": 1.9793814432989691,
"grad_norm": 0.710391104221344,
"learning_rate": 3.5650937551174914e-06,
"loss": 1.148451328277588,
"step": 1728
},
{
"epoch": 1.9816723940435281,
"grad_norm": 1.249986171722412,
"learning_rate": 3.554839821137026e-06,
"loss": 0.9829602837562561,
"step": 1730
},
{
"epoch": 1.9839633447880871,
"grad_norm": 0.2360655516386032,
"learning_rate": 3.5445982926523006e-06,
"loss": 1.1001036167144775,
"step": 1732
},
{
"epoch": 1.986254295532646,
"grad_norm": 0.8706629276275635,
"learning_rate": 3.534369234979914e-06,
"loss": 1.0795059204101562,
"step": 1734
},
{
"epoch": 1.988545246277205,
"grad_norm": 0.3764958381652832,
"learning_rate": 3.5241527133569197e-06,
"loss": 0.0997859463095665,
"step": 1736
},
{
"epoch": 1.990836197021764,
"grad_norm": 0.8101068139076233,
"learning_rate": 3.5139487929404335e-06,
"loss": 0.8904887437820435,
"step": 1738
},
{
"epoch": 1.993127147766323,
"grad_norm": 0.5189135074615479,
"learning_rate": 3.5037575388071997e-06,
"loss": 1.174041509628296,
"step": 1740
},
{
"epoch": 1.995418098510882,
"grad_norm": 1.1633189916610718,
"learning_rate": 3.4935790159531786e-06,
"loss": 1.0208793878555298,
"step": 1742
},
{
"epoch": 1.997709049255441,
"grad_norm": 0.41635146737098694,
"learning_rate": 3.4834132892931405e-06,
"loss": 1.1487672328948975,
"step": 1744
},
{
"epoch": 2.0,
"grad_norm": 0.46879705786705017,
"learning_rate": 3.47326042366024e-06,
"loss": 1.1573315858840942,
"step": 1746
},
{
"epoch": 2.002290950744559,
"grad_norm": 0.42209404706954956,
"learning_rate": 3.46312048380562e-06,
"loss": 1.0264599323272705,
"step": 1748
},
{
"epoch": 2.004581901489118,
"grad_norm": 0.5175220370292664,
"learning_rate": 3.452993534397975e-06,
"loss": 1.196134090423584,
"step": 1750
},
{
"epoch": 2.006872852233677,
"grad_norm": 0.4058225750923157,
"learning_rate": 3.4428796400231595e-06,
"loss": 1.1240901947021484,
"step": 1752
},
{
"epoch": 2.009163802978236,
"grad_norm": 1.5380558967590332,
"learning_rate": 3.4327788651837658e-06,
"loss": 1.1184686422348022,
"step": 1754
},
{
"epoch": 2.011454753722795,
"grad_norm": 0.24166224896907806,
"learning_rate": 3.422691274298714e-06,
"loss": 1.0136035680770874,
"step": 1756
},
{
"epoch": 2.013745704467354,
"grad_norm": 0.3540262281894684,
"learning_rate": 3.4126169317028445e-06,
"loss": 1.099753975868225,
"step": 1758
},
{
"epoch": 2.016036655211913,
"grad_norm": 4.086158752441406,
"learning_rate": 3.4025559016465003e-06,
"loss": 1.2752199172973633,
"step": 1760
},
{
"epoch": 2.018327605956472,
"grad_norm": 0.2709142863750458,
"learning_rate": 3.3925082482951253e-06,
"loss": 1.0859545469284058,
"step": 1762
},
{
"epoch": 2.020618556701031,
"grad_norm": 1.6579638719558716,
"learning_rate": 3.382474035728852e-06,
"loss": 0.9678882360458374,
"step": 1764
},
{
"epoch": 2.02290950744559,
"grad_norm": 0.9905798435211182,
"learning_rate": 3.372453327942092e-06,
"loss": 1.2576557397842407,
"step": 1766
},
{
"epoch": 2.025200458190149,
"grad_norm": 0.586355984210968,
"learning_rate": 3.3624461888431315e-06,
"loss": 1.0932533740997314,
"step": 1768
},
{
"epoch": 2.027491408934708,
"grad_norm": 0.7022833824157715,
"learning_rate": 3.352452682253713e-06,
"loss": 1.1744213104248047,
"step": 1770
},
{
"epoch": 2.029782359679267,
"grad_norm": 0.5173978805541992,
"learning_rate": 3.3424728719086424e-06,
"loss": 1.1649706363677979,
"step": 1772
},
{
"epoch": 2.0320733104238258,
"grad_norm": 0.4220377206802368,
"learning_rate": 3.332506821455376e-06,
"loss": 0.9526954293251038,
"step": 1774
},
{
"epoch": 2.0343642611683848,
"grad_norm": 0.8884670734405518,
"learning_rate": 3.3225545944536132e-06,
"loss": 0.8466014862060547,
"step": 1776
},
{
"epoch": 2.0366552119129437,
"grad_norm": 0.7669530510902405,
"learning_rate": 3.3126162543748906e-06,
"loss": 1.048229694366455,
"step": 1778
},
{
"epoch": 2.0389461626575027,
"grad_norm": 0.45862677693367004,
"learning_rate": 3.302691864602183e-06,
"loss": 1.0864906311035156,
"step": 1780
},
{
"epoch": 2.0412371134020617,
"grad_norm": 0.6318526268005371,
"learning_rate": 3.292781488429494e-06,
"loss": 1.1123660802841187,
"step": 1782
},
{
"epoch": 2.0435280641466207,
"grad_norm": 0.6299246549606323,
"learning_rate": 3.2828851890614534e-06,
"loss": 1.052357792854309,
"step": 1784
},
{
"epoch": 2.0458190148911797,
"grad_norm": 0.23600554466247559,
"learning_rate": 3.2730030296129157e-06,
"loss": 1.099745750427246,
"step": 1786
},
{
"epoch": 2.0481099656357387,
"grad_norm": 0.29890838265419006,
"learning_rate": 3.2631350731085526e-06,
"loss": 1.165632963180542,
"step": 1788
},
{
"epoch": 2.0504009163802976,
"grad_norm": 0.38656070828437805,
"learning_rate": 3.25328138248246e-06,
"loss": 1.0975539684295654,
"step": 1790
},
{
"epoch": 2.0526918671248566,
"grad_norm": 0.5550158023834229,
"learning_rate": 3.2434420205777473e-06,
"loss": 0.6457653045654297,
"step": 1792
},
{
"epoch": 2.0549828178694156,
"grad_norm": 1.2036755084991455,
"learning_rate": 3.2336170501461433e-06,
"loss": 0.9297122955322266,
"step": 1794
},
{
"epoch": 2.0572737686139746,
"grad_norm": 0.7063114047050476,
"learning_rate": 3.2238065338475944e-06,
"loss": 0.9581747055053711,
"step": 1796
},
{
"epoch": 2.0595647193585336,
"grad_norm": 0.3019329309463501,
"learning_rate": 3.2140105342498577e-06,
"loss": 1.0715091228485107,
"step": 1798
},
{
"epoch": 2.0618556701030926,
"grad_norm": 0.43459922075271606,
"learning_rate": 3.2042291138281155e-06,
"loss": 1.1576061248779297,
"step": 1800
},
{
"epoch": 2.0641466208476515,
"grad_norm": 1.0746312141418457,
"learning_rate": 3.194462334964566e-06,
"loss": 0.6688909530639648,
"step": 1802
},
{
"epoch": 2.066437571592211,
"grad_norm": 1.6781188249588013,
"learning_rate": 3.184710259948032e-06,
"loss": 1.0624680519104004,
"step": 1804
},
{
"epoch": 2.06872852233677,
"grad_norm": 0.8279687762260437,
"learning_rate": 3.1749729509735555e-06,
"loss": 0.5724896192550659,
"step": 1806
},
{
"epoch": 2.071019473081329,
"grad_norm": 0.8611645698547363,
"learning_rate": 3.165250470142009e-06,
"loss": 1.0552845001220703,
"step": 1808
},
{
"epoch": 2.073310423825888,
"grad_norm": 0.731144905090332,
"learning_rate": 3.155542879459702e-06,
"loss": 1.301546573638916,
"step": 1810
},
{
"epoch": 2.075601374570447,
"grad_norm": 0.564445436000824,
"learning_rate": 3.145850240837971e-06,
"loss": 1.087944746017456,
"step": 1812
},
{
"epoch": 2.077892325315006,
"grad_norm": 0.6972827911376953,
"learning_rate": 3.136172616092801e-06,
"loss": 1.1200871467590332,
"step": 1814
},
{
"epoch": 2.080183276059565,
"grad_norm": 0.8714810013771057,
"learning_rate": 3.1265100669444184e-06,
"loss": 0.9296277761459351,
"step": 1816
},
{
"epoch": 2.082474226804124,
"grad_norm": 0.9884515404701233,
"learning_rate": 3.116862655016907e-06,
"loss": 1.1618545055389404,
"step": 1818
},
{
"epoch": 2.084765177548683,
"grad_norm": 0.506373405456543,
"learning_rate": 3.107230441837812e-06,
"loss": 1.1905784606933594,
"step": 1820
},
{
"epoch": 2.087056128293242,
"grad_norm": 0.30389317870140076,
"learning_rate": 3.097613488837744e-06,
"loss": 1.0908293724060059,
"step": 1822
},
{
"epoch": 2.089347079037801,
"grad_norm": 1.8103982210159302,
"learning_rate": 3.0880118573499928e-06,
"loss": 1.1718205213546753,
"step": 1824
},
{
"epoch": 2.09163802978236,
"grad_norm": 0.436989963054657,
"learning_rate": 3.0784256086101265e-06,
"loss": 1.092139482498169,
"step": 1826
},
{
"epoch": 2.093928980526919,
"grad_norm": 0.34637048840522766,
"learning_rate": 3.0688548037556203e-06,
"loss": 1.0642328262329102,
"step": 1828
},
{
"epoch": 2.0962199312714778,
"grad_norm": 0.8757824897766113,
"learning_rate": 3.0592995038254413e-06,
"loss": 1.127882480621338,
"step": 1830
},
{
"epoch": 2.0985108820160367,
"grad_norm": 0.30737531185150146,
"learning_rate": 3.04975976975968e-06,
"loss": 1.0287266969680786,
"step": 1832
},
{
"epoch": 2.1008018327605957,
"grad_norm": 0.5797215700149536,
"learning_rate": 3.0402356623991487e-06,
"loss": 1.2051303386688232,
"step": 1834
},
{
"epoch": 2.1030927835051547,
"grad_norm": 0.6220235228538513,
"learning_rate": 3.0307272424850043e-06,
"loss": 1.1138908863067627,
"step": 1836
},
{
"epoch": 2.1053837342497137,
"grad_norm": 0.24019214510917664,
"learning_rate": 3.021234570658351e-06,
"loss": 1.0616822242736816,
"step": 1838
},
{
"epoch": 2.1076746849942727,
"grad_norm": 0.47777384519577026,
"learning_rate": 3.011757707459857e-06,
"loss": 1.16475248336792,
"step": 1840
},
{
"epoch": 2.1099656357388317,
"grad_norm": 0.12783193588256836,
"learning_rate": 3.0022967133293716e-06,
"loss": 0.9693043231964111,
"step": 1842
},
{
"epoch": 2.1122565864833907,
"grad_norm": 0.743993878364563,
"learning_rate": 2.992851648605536e-06,
"loss": 1.1348570585250854,
"step": 1844
},
{
"epoch": 2.1145475372279496,
"grad_norm": 0.3964126706123352,
"learning_rate": 2.9834225735254e-06,
"loss": 0.6107242107391357,
"step": 1846
},
{
"epoch": 2.1168384879725086,
"grad_norm": 0.7012664079666138,
"learning_rate": 2.9740095482240382e-06,
"loss": 1.1791443824768066,
"step": 1848
},
{
"epoch": 2.1191294387170676,
"grad_norm": 0.18129964172840118,
"learning_rate": 2.964612632734163e-06,
"loss": 1.1110697984695435,
"step": 1850
},
{
"epoch": 2.1214203894616266,
"grad_norm": 0.2507210075855255,
"learning_rate": 2.9552318869857464e-06,
"loss": 0.8801964521408081,
"step": 1852
},
{
"epoch": 2.1237113402061856,
"grad_norm": 0.30734899640083313,
"learning_rate": 2.9458673708056385e-06,
"loss": 1.13587486743927,
"step": 1854
},
{
"epoch": 2.1260022909507446,
"grad_norm": 0.4164876341819763,
"learning_rate": 2.936519143917179e-06,
"loss": 1.0800116062164307,
"step": 1856
},
{
"epoch": 2.1282932416953035,
"grad_norm": 0.5680100321769714,
"learning_rate": 2.9271872659398224e-06,
"loss": 1.193451166152954,
"step": 1858
},
{
"epoch": 2.1305841924398625,
"grad_norm": 0.5644039511680603,
"learning_rate": 2.9178717963887557e-06,
"loss": 1.0699570178985596,
"step": 1860
},
{
"epoch": 2.1328751431844215,
"grad_norm": 0.7427708506584167,
"learning_rate": 2.90857279467452e-06,
"loss": 1.1289746761322021,
"step": 1862
},
{
"epoch": 2.1351660939289805,
"grad_norm": 0.47718605399131775,
"learning_rate": 2.899290320102631e-06,
"loss": 0.9117916226387024,
"step": 1864
},
{
"epoch": 2.1374570446735395,
"grad_norm": 1.0634793043136597,
"learning_rate": 2.890024431873201e-06,
"loss": 0.8808594942092896,
"step": 1866
},
{
"epoch": 2.1397479954180985,
"grad_norm": 0.2982898950576782,
"learning_rate": 2.880775189080555e-06,
"loss": 1.001103401184082,
"step": 1868
},
{
"epoch": 2.1420389461626574,
"grad_norm": 0.19586509466171265,
"learning_rate": 2.8715426507128687e-06,
"loss": 1.054373025894165,
"step": 1870
},
{
"epoch": 2.1443298969072164,
"grad_norm": 0.4863986074924469,
"learning_rate": 2.862326875651777e-06,
"loss": 1.1137691736221313,
"step": 1872
},
{
"epoch": 2.1466208476517754,
"grad_norm": 0.41139012575149536,
"learning_rate": 2.853127922672008e-06,
"loss": 1.0064350366592407,
"step": 1874
},
{
"epoch": 2.1489117983963344,
"grad_norm": 1.2407772541046143,
"learning_rate": 2.8439458504410044e-06,
"loss": 1.0638196468353271,
"step": 1876
},
{
"epoch": 2.1512027491408934,
"grad_norm": 0.6728169322013855,
"learning_rate": 2.834780717518547e-06,
"loss": 1.0377411842346191,
"step": 1878
},
{
"epoch": 2.1534936998854524,
"grad_norm": 0.2670747935771942,
"learning_rate": 2.825632582356387e-06,
"loss": 0.9914096593856812,
"step": 1880
},
{
"epoch": 2.1557846506300113,
"grad_norm": 0.24730202555656433,
"learning_rate": 2.8165015032978703e-06,
"loss": 1.0537339448928833,
"step": 1882
},
{
"epoch": 2.1580756013745703,
"grad_norm": 0.33252057433128357,
"learning_rate": 2.8073875385775663e-06,
"loss": 1.1693732738494873,
"step": 1884
},
{
"epoch": 2.1603665521191293,
"grad_norm": 0.23521366715431213,
"learning_rate": 2.7982907463208896e-06,
"loss": 0.8833960890769958,
"step": 1886
},
{
"epoch": 2.1626575028636883,
"grad_norm": 0.5896117091178894,
"learning_rate": 2.7892111845437416e-06,
"loss": 1.0457485914230347,
"step": 1888
},
{
"epoch": 2.1649484536082473,
"grad_norm": 0.4665268659591675,
"learning_rate": 2.780148911152135e-06,
"loss": 1.0155045986175537,
"step": 1890
},
{
"epoch": 2.1672394043528063,
"grad_norm": 0.43222132325172424,
"learning_rate": 2.7711039839418187e-06,
"loss": 0.9927984476089478,
"step": 1892
},
{
"epoch": 2.1695303550973652,
"grad_norm": 0.2742535471916199,
"learning_rate": 2.762076460597919e-06,
"loss": 1.0519921779632568,
"step": 1894
},
{
"epoch": 2.1718213058419242,
"grad_norm": 0.6122022867202759,
"learning_rate": 2.7530663986945605e-06,
"loss": 0.804986834526062,
"step": 1896
},
{
"epoch": 2.174112256586483,
"grad_norm": 0.3205280601978302,
"learning_rate": 2.7440738556945122e-06,
"loss": 0.9710903763771057,
"step": 1898
},
{
"epoch": 2.176403207331042,
"grad_norm": 0.5308868288993835,
"learning_rate": 2.735098888948811e-06,
"loss": 0.823322594165802,
"step": 1900
},
{
"epoch": 2.178694158075601,
"grad_norm": 1.2458512783050537,
"learning_rate": 2.726141555696399e-06,
"loss": 1.3204128742218018,
"step": 1902
},
{
"epoch": 2.18098510882016,
"grad_norm": 1.1205463409423828,
"learning_rate": 2.7172019130637605e-06,
"loss": 1.1062953472137451,
"step": 1904
},
{
"epoch": 2.183276059564719,
"grad_norm": 0.49585333466529846,
"learning_rate": 2.708280018064551e-06,
"loss": 0.5895834565162659,
"step": 1906
},
{
"epoch": 2.1855670103092786,
"grad_norm": 0.5067374110221863,
"learning_rate": 2.699375927599245e-06,
"loss": 1.037330985069275,
"step": 1908
},
{
"epoch": 2.1878579610538376,
"grad_norm": 0.9823523759841919,
"learning_rate": 2.6904896984547614e-06,
"loss": 0.5440244078636169,
"step": 1910
},
{
"epoch": 2.1901489117983965,
"grad_norm": 0.9039705991744995,
"learning_rate": 2.6816213873041086e-06,
"loss": 1.1244115829467773,
"step": 1912
},
{
"epoch": 2.1924398625429555,
"grad_norm": 1.6192059516906738,
"learning_rate": 2.6727710507060166e-06,
"loss": 0.4239127039909363,
"step": 1914
},
{
"epoch": 2.1947308132875145,
"grad_norm": 0.4529842436313629,
"learning_rate": 2.6639387451045894e-06,
"loss": 1.093721628189087,
"step": 1916
},
{
"epoch": 2.1970217640320735,
"grad_norm": 0.6420417428016663,
"learning_rate": 2.6551245268289293e-06,
"loss": 1.1479039192199707,
"step": 1918
},
{
"epoch": 2.1993127147766325,
"grad_norm": 0.8427514433860779,
"learning_rate": 2.6463284520927834e-06,
"loss": 1.0050065517425537,
"step": 1920
},
{
"epoch": 2.2016036655211915,
"grad_norm": 0.6339349150657654,
"learning_rate": 2.63755057699419e-06,
"loss": 0.9615905284881592,
"step": 1922
},
{
"epoch": 2.2038946162657505,
"grad_norm": 0.8383830189704895,
"learning_rate": 2.6287909575151166e-06,
"loss": 1.1090657711029053,
"step": 1924
},
{
"epoch": 2.2061855670103094,
"grad_norm": 1.1984935998916626,
"learning_rate": 2.620049649521103e-06,
"loss": 0.953253984451294,
"step": 1926
},
{
"epoch": 2.2084765177548684,
"grad_norm": 0.34425899386405945,
"learning_rate": 2.6113267087609018e-06,
"loss": 1.0870815515518188,
"step": 1928
},
{
"epoch": 2.2107674684994274,
"grad_norm": 1.677963376045227,
"learning_rate": 2.6026221908661307e-06,
"loss": 1.0642032623291016,
"step": 1930
},
{
"epoch": 2.2130584192439864,
"grad_norm": 0.594247043132782,
"learning_rate": 2.5939361513509124e-06,
"loss": 0.6523274183273315,
"step": 1932
},
{
"epoch": 2.2153493699885454,
"grad_norm": 0.458957314491272,
"learning_rate": 2.5852686456115194e-06,
"loss": 1.0752052068710327,
"step": 1934
},
{
"epoch": 2.2176403207331044,
"grad_norm": 0.31083378195762634,
"learning_rate": 2.5766197289260277e-06,
"loss": 1.0612359046936035,
"step": 1936
},
{
"epoch": 2.2199312714776633,
"grad_norm": 0.27045944333076477,
"learning_rate": 2.5679894564539513e-06,
"loss": 1.068912386894226,
"step": 1938
},
{
"epoch": 2.2222222222222223,
"grad_norm": 0.44911208748817444,
"learning_rate": 2.5593778832359062e-06,
"loss": 0.990754246711731,
"step": 1940
},
{
"epoch": 2.2245131729667813,
"grad_norm": 0.21117167174816132,
"learning_rate": 2.550785064193246e-06,
"loss": 0.8659780621528625,
"step": 1942
},
{
"epoch": 2.2268041237113403,
"grad_norm": 0.46905815601348877,
"learning_rate": 2.5422110541277225e-06,
"loss": 1.1481306552886963,
"step": 1944
},
{
"epoch": 2.2290950744558993,
"grad_norm": 2.0418505668640137,
"learning_rate": 2.5336559077211285e-06,
"loss": 0.8427814841270447,
"step": 1946
},
{
"epoch": 2.2313860252004583,
"grad_norm": 0.7220437526702881,
"learning_rate": 2.5251196795349476e-06,
"loss": 0.7763566374778748,
"step": 1948
},
{
"epoch": 2.2336769759450172,
"grad_norm": 0.3684818148612976,
"learning_rate": 2.5166024240100175e-06,
"loss": 1.0701335668563843,
"step": 1950
},
{
"epoch": 2.2359679266895762,
"grad_norm": 0.17287565767765045,
"learning_rate": 2.508104195466169e-06,
"loss": 1.0804567337036133,
"step": 1952
},
{
"epoch": 2.238258877434135,
"grad_norm": 0.6882048845291138,
"learning_rate": 2.4996250481018917e-06,
"loss": 0.618576169013977,
"step": 1954
},
{
"epoch": 2.240549828178694,
"grad_norm": 0.9598178863525391,
"learning_rate": 2.491165035993977e-06,
"loss": 1.0526471138000488,
"step": 1956
},
{
"epoch": 2.242840778923253,
"grad_norm": 0.22992214560508728,
"learning_rate": 2.4827242130971822e-06,
"loss": 0.9994850158691406,
"step": 1958
},
{
"epoch": 2.245131729667812,
"grad_norm": 0.21707294881343842,
"learning_rate": 2.4743026332438835e-06,
"loss": 1.0506229400634766,
"step": 1960
},
{
"epoch": 2.247422680412371,
"grad_norm": 0.894084095954895,
"learning_rate": 2.4659003501437302e-06,
"loss": 1.073114037513733,
"step": 1962
},
{
"epoch": 2.24971363115693,
"grad_norm": 0.47775256633758545,
"learning_rate": 2.4575174173833076e-06,
"loss": 1.0495290756225586,
"step": 1964
},
{
"epoch": 2.252004581901489,
"grad_norm": 0.6680848598480225,
"learning_rate": 2.449153888425786e-06,
"loss": 1.1938798427581787,
"step": 1966
},
{
"epoch": 2.254295532646048,
"grad_norm": 0.8610414862632751,
"learning_rate": 2.4408098166105905e-06,
"loss": 1.1015326976776123,
"step": 1968
},
{
"epoch": 2.256586483390607,
"grad_norm": 0.2719790041446686,
"learning_rate": 2.4324852551530546e-06,
"loss": 1.1828646659851074,
"step": 1970
},
{
"epoch": 2.258877434135166,
"grad_norm": 0.3849657475948334,
"learning_rate": 2.424180257144082e-06,
"loss": 0.9668605327606201,
"step": 1972
},
{
"epoch": 2.261168384879725,
"grad_norm": 0.24315226078033447,
"learning_rate": 2.4158948755498097e-06,
"loss": 1.0409603118896484,
"step": 1974
},
{
"epoch": 2.263459335624284,
"grad_norm": 0.3323868215084076,
"learning_rate": 2.407629163211264e-06,
"loss": 1.0804579257965088,
"step": 1976
},
{
"epoch": 2.265750286368843,
"grad_norm": 0.7172708511352539,
"learning_rate": 2.399383172844033e-06,
"loss": 1.0808374881744385,
"step": 1978
},
{
"epoch": 2.268041237113402,
"grad_norm": 0.48791539669036865,
"learning_rate": 2.3911569570379226e-06,
"loss": 1.0492467880249023,
"step": 1980
},
{
"epoch": 2.270332187857961,
"grad_norm": 0.6731001138687134,
"learning_rate": 2.3829505682566274e-06,
"loss": 0.8396996259689331,
"step": 1982
},
{
"epoch": 2.27262313860252,
"grad_norm": 0.2888670563697815,
"learning_rate": 2.374764058837388e-06,
"loss": 1.0629734992980957,
"step": 1984
},
{
"epoch": 2.274914089347079,
"grad_norm": 0.5932695269584656,
"learning_rate": 2.366597480990663e-06,
"loss": 1.1070618629455566,
"step": 1986
},
{
"epoch": 2.277205040091638,
"grad_norm": 0.4856787621974945,
"learning_rate": 2.3584508867998023e-06,
"loss": 1.13844633102417,
"step": 1988
},
{
"epoch": 2.279495990836197,
"grad_norm": 0.487051784992218,
"learning_rate": 2.350324328220697e-06,
"loss": 1.1154626607894897,
"step": 1990
},
{
"epoch": 2.281786941580756,
"grad_norm": 0.29695621132850647,
"learning_rate": 2.3422178570814656e-06,
"loss": 1.1702851057052612,
"step": 1992
},
{
"epoch": 2.284077892325315,
"grad_norm": 0.23192717134952545,
"learning_rate": 2.3341315250821102e-06,
"loss": 0.9374110698699951,
"step": 1994
},
{
"epoch": 2.286368843069874,
"grad_norm": 1.2225816249847412,
"learning_rate": 2.326065383794202e-06,
"loss": 0.7910254001617432,
"step": 1996
},
{
"epoch": 2.288659793814433,
"grad_norm": 0.7794654369354248,
"learning_rate": 2.3180194846605367e-06,
"loss": 1.1501424312591553,
"step": 1998
},
{
"epoch": 2.290950744558992,
"grad_norm": 0.6967248916625977,
"learning_rate": 2.3099938789948147e-06,
"loss": 0.9952400326728821,
"step": 2000
},
{
"epoch": 2.293241695303551,
"grad_norm": 1.1127731800079346,
"learning_rate": 2.301988617981314e-06,
"loss": 0.6200200915336609,
"step": 2002
},
{
"epoch": 2.29553264604811,
"grad_norm": 1.9862406253814697,
"learning_rate": 2.294003752674557e-06,
"loss": 0.8297408223152161,
"step": 2004
},
{
"epoch": 2.297823596792669,
"grad_norm": 1.128572702407837,
"learning_rate": 2.2860393339990005e-06,
"loss": 0.48388516902923584,
"step": 2006
},
{
"epoch": 2.3001145475372278,
"grad_norm": 0.25328806042671204,
"learning_rate": 2.2780954127486883e-06,
"loss": 1.1547516584396362,
"step": 2008
},
{
"epoch": 2.3024054982817868,
"grad_norm": 14.397195816040039,
"learning_rate": 2.270172039586948e-06,
"loss": 0.8220672607421875,
"step": 2010
},
{
"epoch": 2.3046964490263457,
"grad_norm": 0.5289891958236694,
"learning_rate": 2.2622692650460516e-06,
"loss": 1.0384976863861084,
"step": 2012
},
{
"epoch": 2.3069873997709047,
"grad_norm": 0.2456163614988327,
"learning_rate": 2.25438713952691e-06,
"loss": 0.9674023389816284,
"step": 2014
},
{
"epoch": 2.3092783505154637,
"grad_norm": 0.6010926961898804,
"learning_rate": 2.2465257132987388e-06,
"loss": 0.9032200574874878,
"step": 2016
},
{
"epoch": 2.3115693012600227,
"grad_norm": 0.42077237367630005,
"learning_rate": 2.2386850364987357e-06,
"loss": 1.0571644306182861,
"step": 2018
},
{
"epoch": 2.3138602520045817,
"grad_norm": 0.4678718149662018,
"learning_rate": 2.230865159131774e-06,
"loss": 0.12346747517585754,
"step": 2020
},
{
"epoch": 2.3161512027491407,
"grad_norm": 0.7129454016685486,
"learning_rate": 2.2230661310700717e-06,
"loss": 1.0253472328186035,
"step": 2022
},
{
"epoch": 2.3184421534936996,
"grad_norm": 1.1142655611038208,
"learning_rate": 2.215288002052882e-06,
"loss": 0.5160566568374634,
"step": 2024
},
{
"epoch": 2.320733104238259,
"grad_norm": 0.5436863899230957,
"learning_rate": 2.207530821686166e-06,
"loss": 1.054995059967041,
"step": 2026
},
{
"epoch": 2.323024054982818,
"grad_norm": 9.801424980163574,
"learning_rate": 2.199794639442287e-06,
"loss": 0.83745938539505,
"step": 2028
},
{
"epoch": 2.325315005727377,
"grad_norm": 0.7549030184745789,
"learning_rate": 2.1920795046596887e-06,
"loss": 1.2911202907562256,
"step": 2030
},
{
"epoch": 2.327605956471936,
"grad_norm": 0.5055285096168518,
"learning_rate": 2.1843854665425816e-06,
"loss": 0.5712074041366577,
"step": 2032
},
{
"epoch": 2.329896907216495,
"grad_norm": 0.44524145126342773,
"learning_rate": 2.176712574160632e-06,
"loss": 0.9774595499038696,
"step": 2034
},
{
"epoch": 2.332187857961054,
"grad_norm": 0.8350621461868286,
"learning_rate": 2.1690608764486417e-06,
"loss": 1.0842552185058594,
"step": 2036
},
{
"epoch": 2.334478808705613,
"grad_norm": 0.4929974377155304,
"learning_rate": 2.1614304222062447e-06,
"loss": 0.7975589036941528,
"step": 2038
},
{
"epoch": 2.336769759450172,
"grad_norm": 0.21416163444519043,
"learning_rate": 2.1538212600975927e-06,
"loss": 1.016804575920105,
"step": 2040
},
{
"epoch": 2.339060710194731,
"grad_norm": 2.911141872406006,
"learning_rate": 2.146233438651042e-06,
"loss": 0.9362708330154419,
"step": 2042
},
{
"epoch": 2.34135166093929,
"grad_norm": 0.2715796232223511,
"learning_rate": 2.138667006258849e-06,
"loss": 0.7574849128723145,
"step": 2044
},
{
"epoch": 2.343642611683849,
"grad_norm": 1.3292779922485352,
"learning_rate": 2.131122011176854e-06,
"loss": 1.0263173580169678,
"step": 2046
},
{
"epoch": 2.345933562428408,
"grad_norm": 2.113558530807495,
"learning_rate": 2.1235985015241832e-06,
"loss": 0.8715229034423828,
"step": 2048
},
{
"epoch": 2.348224513172967,
"grad_norm": 0.36145395040512085,
"learning_rate": 2.1160965252829344e-06,
"loss": 1.086829423904419,
"step": 2050
},
{
"epoch": 2.350515463917526,
"grad_norm": 2.3738577365875244,
"learning_rate": 2.108616130297876e-06,
"loss": 0.6945520639419556,
"step": 2052
},
{
"epoch": 2.352806414662085,
"grad_norm": 0.572725772857666,
"learning_rate": 2.101157364276134e-06,
"loss": 1.1936531066894531,
"step": 2054
},
{
"epoch": 2.355097365406644,
"grad_norm": 0.31756481528282166,
"learning_rate": 2.0937202747868974e-06,
"loss": 1.0508008003234863,
"step": 2056
},
{
"epoch": 2.357388316151203,
"grad_norm": 0.651594340801239,
"learning_rate": 2.08630490926111e-06,
"loss": 0.9217154383659363,
"step": 2058
},
{
"epoch": 2.359679266895762,
"grad_norm": 0.4735032618045807,
"learning_rate": 2.0789113149911678e-06,
"loss": 1.1117918491363525,
"step": 2060
},
{
"epoch": 2.361970217640321,
"grad_norm": 0.9481447339057922,
"learning_rate": 2.071539539130619e-06,
"loss": 0.8254319429397583,
"step": 2062
},
{
"epoch": 2.3642611683848798,
"grad_norm": 0.1962810903787613,
"learning_rate": 2.0641896286938586e-06,
"loss": 1.1184422969818115,
"step": 2064
},
{
"epoch": 2.3665521191294387,
"grad_norm": 1.010116696357727,
"learning_rate": 2.0568616305558362e-06,
"loss": 0.9059109687805176,
"step": 2066
},
{
"epoch": 2.3688430698739977,
"grad_norm": 0.5738312602043152,
"learning_rate": 2.0495555914517517e-06,
"loss": 1.0886991024017334,
"step": 2068
},
{
"epoch": 2.3711340206185567,
"grad_norm": 0.34965646266937256,
"learning_rate": 2.042271557976758e-06,
"loss": 1.0940256118774414,
"step": 2070
},
{
"epoch": 2.3734249713631157,
"grad_norm": 6.681120872497559,
"learning_rate": 2.0350095765856674e-06,
"loss": 0.7195948362350464,
"step": 2072
},
{
"epoch": 2.3757159221076747,
"grad_norm": 0.38432735204696655,
"learning_rate": 2.0277696935926465e-06,
"loss": 1.0875439643859863,
"step": 2074
},
{
"epoch": 2.3780068728522337,
"grad_norm": 0.3983290493488312,
"learning_rate": 2.020551955170932e-06,
"loss": 0.9240410327911377,
"step": 2076
},
{
"epoch": 2.3802978235967927,
"grad_norm": 0.3969801664352417,
"learning_rate": 2.0133564073525283e-06,
"loss": 0.7710465788841248,
"step": 2078
},
{
"epoch": 2.3825887743413516,
"grad_norm": 0.5255761742591858,
"learning_rate": 2.0061830960279182e-06,
"loss": 1.0027873516082764,
"step": 2080
},
{
"epoch": 2.3848797250859106,
"grad_norm": 0.7259711623191833,
"learning_rate": 1.9990320669457664e-06,
"loss": 1.1011751890182495,
"step": 2082
},
{
"epoch": 2.3871706758304696,
"grad_norm": 0.20675049722194672,
"learning_rate": 1.9919033657126297e-06,
"loss": 1.081270694732666,
"step": 2084
},
{
"epoch": 2.3894616265750286,
"grad_norm": 0.7646600008010864,
"learning_rate": 1.9847970377926723e-06,
"loss": 1.105683445930481,
"step": 2086
},
{
"epoch": 2.3917525773195876,
"grad_norm": 0.39679569005966187,
"learning_rate": 1.9777131285073608e-06,
"loss": 1.0713748931884766,
"step": 2088
},
{
"epoch": 2.3940435280641466,
"grad_norm": 0.24506868422031403,
"learning_rate": 1.9706516830351915e-06,
"loss": 1.2177720069885254,
"step": 2090
},
{
"epoch": 2.3963344788087055,
"grad_norm": 0.7486739158630371,
"learning_rate": 1.963612746411389e-06,
"loss": 1.154736042022705,
"step": 2092
},
{
"epoch": 2.3986254295532645,
"grad_norm": 0.31168675422668457,
"learning_rate": 1.9565963635276326e-06,
"loss": 1.0588105916976929,
"step": 2094
},
{
"epoch": 2.4009163802978235,
"grad_norm": 0.920814573764801,
"learning_rate": 1.949602579131754e-06,
"loss": 1.024669885635376,
"step": 2096
},
{
"epoch": 2.4032073310423825,
"grad_norm": 0.468282014131546,
"learning_rate": 1.9426314378274663e-06,
"loss": 1.0923664569854736,
"step": 2098
},
{
"epoch": 2.4054982817869415,
"grad_norm": 0.7058407068252563,
"learning_rate": 1.9356829840740705e-06,
"loss": 0.5762814283370972,
"step": 2100
},
{
"epoch": 2.4077892325315005,
"grad_norm": 0.5558688044548035,
"learning_rate": 1.928757262186177e-06,
"loss": 0.5579245090484619,
"step": 2102
},
{
"epoch": 2.4100801832760594,
"grad_norm": 0.5086036920547485,
"learning_rate": 1.9218543163334198e-06,
"loss": 1.1277086734771729,
"step": 2104
},
{
"epoch": 2.4123711340206184,
"grad_norm": 0.41747531294822693,
"learning_rate": 1.914974190540174e-06,
"loss": 1.0628433227539062,
"step": 2106
},
{
"epoch": 2.4146620847651774,
"grad_norm": 0.41488274931907654,
"learning_rate": 1.9081169286852827e-06,
"loss": 0.9923676252365112,
"step": 2108
},
{
"epoch": 2.4169530355097364,
"grad_norm": 0.709517776966095,
"learning_rate": 1.9012825745017633e-06,
"loss": 0.5995164513587952,
"step": 2110
},
{
"epoch": 2.4192439862542954,
"grad_norm": 2.2270102500915527,
"learning_rate": 1.8944711715765446e-06,
"loss": 1.2495015859603882,
"step": 2112
},
{
"epoch": 2.4215349369988544,
"grad_norm": 0.5543949604034424,
"learning_rate": 1.8876827633501784e-06,
"loss": 1.1288328170776367,
"step": 2114
},
{
"epoch": 2.4238258877434133,
"grad_norm": 0.46857401728630066,
"learning_rate": 1.8809173931165614e-06,
"loss": 0.9634852409362793,
"step": 2116
},
{
"epoch": 2.4261168384879723,
"grad_norm": 0.5771682262420654,
"learning_rate": 1.8741751040226663e-06,
"loss": 1.1221890449523926,
"step": 2118
},
{
"epoch": 2.4284077892325313,
"grad_norm": 1.220275640487671,
"learning_rate": 1.8674559390682629e-06,
"loss": 1.10588538646698,
"step": 2120
},
{
"epoch": 2.4306987399770907,
"grad_norm": 0.17475640773773193,
"learning_rate": 1.8607599411056444e-06,
"loss": 0.8858648538589478,
"step": 2122
},
{
"epoch": 2.4329896907216497,
"grad_norm": 0.6616489887237549,
"learning_rate": 1.8540871528393497e-06,
"loss": 0.996634304523468,
"step": 2124
},
{
"epoch": 2.4352806414662087,
"grad_norm": 1.1846482753753662,
"learning_rate": 1.8474376168259003e-06,
"loss": 1.125913143157959,
"step": 2126
},
{
"epoch": 2.4375715922107677,
"grad_norm": 0.5403264760971069,
"learning_rate": 1.84081137547352e-06,
"loss": 1.128383755683899,
"step": 2128
},
{
"epoch": 2.4398625429553267,
"grad_norm": 0.36700180172920227,
"learning_rate": 1.83420847104187e-06,
"loss": 0.5875706672668457,
"step": 2130
},
{
"epoch": 2.4421534936998857,
"grad_norm": 0.22898639738559723,
"learning_rate": 1.8276289456417784e-06,
"loss": 1.0603092908859253,
"step": 2132
},
{
"epoch": 2.4444444444444446,
"grad_norm": 0.23132291436195374,
"learning_rate": 1.821072841234966e-06,
"loss": 1.020564079284668,
"step": 2134
},
{
"epoch": 2.4467353951890036,
"grad_norm": 0.31519776582717896,
"learning_rate": 1.8145401996337877e-06,
"loss": 1.0668442249298096,
"step": 2136
},
{
"epoch": 2.4490263459335626,
"grad_norm": 0.2912147641181946,
"learning_rate": 1.80803106250096e-06,
"loss": 0.9594171047210693,
"step": 2138
},
{
"epoch": 2.4513172966781216,
"grad_norm": 1.1906369924545288,
"learning_rate": 1.8015454713492985e-06,
"loss": 1.1961896419525146,
"step": 2140
},
{
"epoch": 2.4536082474226806,
"grad_norm": 0.44794419407844543,
"learning_rate": 1.7950834675414498e-06,
"loss": 1.0605556964874268,
"step": 2142
},
{
"epoch": 2.4558991981672396,
"grad_norm": 0.4688771665096283,
"learning_rate": 1.788645092289627e-06,
"loss": 1.1090596914291382,
"step": 2144
},
{
"epoch": 2.4581901489117985,
"grad_norm": 0.5896684527397156,
"learning_rate": 1.7822303866553536e-06,
"loss": 1.0037330389022827,
"step": 2146
},
{
"epoch": 2.4604810996563575,
"grad_norm": 1.4058597087860107,
"learning_rate": 1.7758393915491957e-06,
"loss": 0.5776699185371399,
"step": 2148
},
{
"epoch": 2.4627720504009165,
"grad_norm": 0.20505838096141815,
"learning_rate": 1.7694721477305026e-06,
"loss": 1.0694668292999268,
"step": 2150
},
{
"epoch": 2.4650630011454755,
"grad_norm": 0.49451765418052673,
"learning_rate": 1.7631286958071444e-06,
"loss": 1.1405375003814697,
"step": 2152
},
{
"epoch": 2.4673539518900345,
"grad_norm": 0.407114177942276,
"learning_rate": 1.7568090762352591e-06,
"loss": 0.5926585793495178,
"step": 2154
},
{
"epoch": 2.4696449026345935,
"grad_norm": 0.3963899612426758,
"learning_rate": 1.7505133293189898e-06,
"loss": 1.0959696769714355,
"step": 2156
},
{
"epoch": 2.4719358533791524,
"grad_norm": 1.4024752378463745,
"learning_rate": 1.7442414952102279e-06,
"loss": 0.6533874273300171,
"step": 2158
},
{
"epoch": 2.4742268041237114,
"grad_norm": 0.31477341055870056,
"learning_rate": 1.7379936139083604e-06,
"loss": 1.1235566139221191,
"step": 2160
},
{
"epoch": 2.4765177548682704,
"grad_norm": 0.45664510130882263,
"learning_rate": 1.731769725260008e-06,
"loss": 0.9252137541770935,
"step": 2162
},
{
"epoch": 2.4788087056128294,
"grad_norm": 0.1995607316493988,
"learning_rate": 1.7255698689587774e-06,
"loss": 0.999367356300354,
"step": 2164
},
{
"epoch": 2.4810996563573884,
"grad_norm": 0.6067396402359009,
"learning_rate": 1.719394084545011e-06,
"loss": 1.191161870956421,
"step": 2166
},
{
"epoch": 2.4833906071019474,
"grad_norm": 0.2916881740093231,
"learning_rate": 1.7132424114055212e-06,
"loss": 1.0785753726959229,
"step": 2168
},
{
"epoch": 2.4856815578465064,
"grad_norm": 0.27279433608055115,
"learning_rate": 1.7071148887733552e-06,
"loss": 1.031059980392456,
"step": 2170
},
{
"epoch": 2.4879725085910653,
"grad_norm": 0.592989444732666,
"learning_rate": 1.7010115557275297e-06,
"loss": 0.9627765417098999,
"step": 2172
},
{
"epoch": 2.4902634593356243,
"grad_norm": 8.736328125,
"learning_rate": 1.6949324511927993e-06,
"loss": 0.6805863380432129,
"step": 2174
},
{
"epoch": 2.4925544100801833,
"grad_norm": 0.3850440979003906,
"learning_rate": 1.6888776139393892e-06,
"loss": 1.0184849500656128,
"step": 2176
},
{
"epoch": 2.4948453608247423,
"grad_norm": 0.26290857791900635,
"learning_rate": 1.6828470825827626e-06,
"loss": 1.1698343753814697,
"step": 2178
},
{
"epoch": 2.4971363115693013,
"grad_norm": 1.0720175504684448,
"learning_rate": 1.6768408955833653e-06,
"loss": 1.0607209205627441,
"step": 2180
},
{
"epoch": 2.4994272623138603,
"grad_norm": 0.7246781587600708,
"learning_rate": 1.6708590912463878e-06,
"loss": 0.7118844389915466,
"step": 2182
},
{
"epoch": 2.5017182130584192,
"grad_norm": 0.519919216632843,
"learning_rate": 1.6649017077215146e-06,
"loss": 1.1157710552215576,
"step": 2184
},
{
"epoch": 2.504009163802978,
"grad_norm": 0.5193498134613037,
"learning_rate": 1.6589687830026835e-06,
"loss": 1.1989827156066895,
"step": 2186
},
{
"epoch": 2.506300114547537,
"grad_norm": 0.5242924690246582,
"learning_rate": 1.653060354927844e-06,
"loss": 1.0901808738708496,
"step": 2188
},
{
"epoch": 2.508591065292096,
"grad_norm": 1.705298900604248,
"learning_rate": 1.647176461178714e-06,
"loss": 0.5451116561889648,
"step": 2190
},
{
"epoch": 2.510882016036655,
"grad_norm": 0.7543349862098694,
"learning_rate": 1.6413171392805444e-06,
"loss": 0.8635871410369873,
"step": 2192
},
{
"epoch": 2.513172966781214,
"grad_norm": 0.4033338725566864,
"learning_rate": 1.6354824266018726e-06,
"loss": 1.0020837783813477,
"step": 2194
},
{
"epoch": 2.515463917525773,
"grad_norm": 0.5562148094177246,
"learning_rate": 1.6296723603542874e-06,
"loss": 0.9753223657608032,
"step": 2196
},
{
"epoch": 2.517754868270332,
"grad_norm": 0.8762247562408447,
"learning_rate": 1.6238869775921947e-06,
"loss": 0.9873074293136597,
"step": 2198
},
{
"epoch": 2.520045819014891,
"grad_norm": 0.38216540217399597,
"learning_rate": 1.6181263152125761e-06,
"loss": 1.1414031982421875,
"step": 2200
},
{
"epoch": 2.52233676975945,
"grad_norm": 0.3067186176776886,
"learning_rate": 1.6123904099547593e-06,
"loss": 0.742071270942688,
"step": 2202
},
{
"epoch": 2.524627720504009,
"grad_norm": 0.38689979910850525,
"learning_rate": 1.606679298400175e-06,
"loss": 1.0735660791397095,
"step": 2204
},
{
"epoch": 2.526918671248568,
"grad_norm": 0.8441147804260254,
"learning_rate": 1.6009930169721332e-06,
"loss": 0.389965295791626,
"step": 2206
},
{
"epoch": 2.529209621993127,
"grad_norm": 0.39243248105049133,
"learning_rate": 1.595331601935586e-06,
"loss": 0.9524307250976562,
"step": 2208
},
{
"epoch": 2.531500572737686,
"grad_norm": 0.27771660685539246,
"learning_rate": 1.5896950893968968e-06,
"loss": 1.1004953384399414,
"step": 2210
},
{
"epoch": 2.533791523482245,
"grad_norm": 0.509863555431366,
"learning_rate": 1.584083515303611e-06,
"loss": 0.6490231156349182,
"step": 2212
},
{
"epoch": 2.536082474226804,
"grad_norm": 0.1981062889099121,
"learning_rate": 1.5784969154442228e-06,
"loss": 1.1196446418762207,
"step": 2214
},
{
"epoch": 2.538373424971363,
"grad_norm": 0.3844130337238312,
"learning_rate": 1.5729353254479544e-06,
"loss": 0.9391459226608276,
"step": 2216
},
{
"epoch": 2.540664375715922,
"grad_norm": 1.6460623741149902,
"learning_rate": 1.5673987807845224e-06,
"loss": 0.4223039746284485,
"step": 2218
},
{
"epoch": 2.542955326460481,
"grad_norm": 0.9630303382873535,
"learning_rate": 1.561887316763914e-06,
"loss": 0.7680240273475647,
"step": 2220
},
{
"epoch": 2.54524627720504,
"grad_norm": 0.8750276565551758,
"learning_rate": 1.5564009685361631e-06,
"loss": 1.1402933597564697,
"step": 2222
},
{
"epoch": 2.547537227949599,
"grad_norm": 0.29777172207832336,
"learning_rate": 1.5509397710911206e-06,
"loss": 1.0347285270690918,
"step": 2224
},
{
"epoch": 2.549828178694158,
"grad_norm": 0.5505490899085999,
"learning_rate": 1.545503759258239e-06,
"loss": 0.9146395921707153,
"step": 2226
},
{
"epoch": 2.552119129438717,
"grad_norm": 0.372688889503479,
"learning_rate": 1.5400929677063453e-06,
"loss": 0.9713397026062012,
"step": 2228
},
{
"epoch": 2.554410080183276,
"grad_norm": 0.9383296966552734,
"learning_rate": 1.534707430943422e-06,
"loss": 1.0299038887023926,
"step": 2230
},
{
"epoch": 2.556701030927835,
"grad_norm": 0.24810487031936646,
"learning_rate": 1.5293471833163836e-06,
"loss": 0.8337885141372681,
"step": 2232
},
{
"epoch": 2.558991981672394,
"grad_norm": 0.6177897453308105,
"learning_rate": 1.5240122590108615e-06,
"loss": 1.1169462203979492,
"step": 2234
},
{
"epoch": 2.561282932416953,
"grad_norm": 0.3793282210826874,
"learning_rate": 1.5187026920509857e-06,
"loss": 0.8278145790100098,
"step": 2236
},
{
"epoch": 2.563573883161512,
"grad_norm": 0.764354944229126,
"learning_rate": 1.5134185162991653e-06,
"loss": 1.032823920249939,
"step": 2238
},
{
"epoch": 2.565864833906071,
"grad_norm": 0.2363336831331253,
"learning_rate": 1.5081597654558753e-06,
"loss": 1.0410435199737549,
"step": 2240
},
{
"epoch": 2.5681557846506298,
"grad_norm": 0.6717326045036316,
"learning_rate": 1.5029264730594366e-06,
"loss": 1.0079201459884644,
"step": 2242
},
{
"epoch": 2.5704467353951888,
"grad_norm": 0.8275331258773804,
"learning_rate": 1.49771867248581e-06,
"loss": 1.1241486072540283,
"step": 2244
},
{
"epoch": 2.5727376861397477,
"grad_norm": 0.5479764938354492,
"learning_rate": 1.4925363969483764e-06,
"loss": 1.2064895629882812,
"step": 2246
},
{
"epoch": 2.5750286368843067,
"grad_norm": 0.29610878229141235,
"learning_rate": 1.48737967949773e-06,
"loss": 1.080405831336975,
"step": 2248
},
{
"epoch": 2.5773195876288657,
"grad_norm": 0.35893121361732483,
"learning_rate": 1.4822485530214636e-06,
"loss": 0.9980157017707825,
"step": 2250
},
{
"epoch": 2.579610538373425,
"grad_norm": 0.545345664024353,
"learning_rate": 1.4771430502439585e-06,
"loss": 1.0427041053771973,
"step": 2252
},
{
"epoch": 2.581901489117984,
"grad_norm": 0.6085090637207031,
"learning_rate": 1.4720632037261837e-06,
"loss": 1.0622684955596924,
"step": 2254
},
{
"epoch": 2.584192439862543,
"grad_norm": 0.4852236211299896,
"learning_rate": 1.4670090458654752e-06,
"loss": 1.002990484237671,
"step": 2256
},
{
"epoch": 2.586483390607102,
"grad_norm": 0.37894105911254883,
"learning_rate": 1.4619806088953423e-06,
"loss": 1.0157802104949951,
"step": 2258
},
{
"epoch": 2.588774341351661,
"grad_norm": 0.3049786686897278,
"learning_rate": 1.4569779248852514e-06,
"loss": 1.032530426979065,
"step": 2260
},
{
"epoch": 2.59106529209622,
"grad_norm": 0.3798280954360962,
"learning_rate": 1.4520010257404282e-06,
"loss": 1.0160043239593506,
"step": 2262
},
{
"epoch": 2.593356242840779,
"grad_norm": 0.41504234075546265,
"learning_rate": 1.4470499432016564e-06,
"loss": 1.0070414543151855,
"step": 2264
},
{
"epoch": 2.595647193585338,
"grad_norm": 0.5342398285865784,
"learning_rate": 1.4421247088450638e-06,
"loss": 0.3559042811393738,
"step": 2266
},
{
"epoch": 2.597938144329897,
"grad_norm": 1.0337114334106445,
"learning_rate": 1.4372253540819342e-06,
"loss": 1.1760070323944092,
"step": 2268
},
{
"epoch": 2.600229095074456,
"grad_norm": 0.6372706890106201,
"learning_rate": 1.4323519101584956e-06,
"loss": 1.1206852197647095,
"step": 2270
},
{
"epoch": 2.602520045819015,
"grad_norm": 0.23677659034729004,
"learning_rate": 1.4275044081557336e-06,
"loss": 0.9349355697631836,
"step": 2272
},
{
"epoch": 2.604810996563574,
"grad_norm": 0.23186595737934113,
"learning_rate": 1.422682878989179e-06,
"loss": 1.0508509874343872,
"step": 2274
},
{
"epoch": 2.607101947308133,
"grad_norm": 0.5015450119972229,
"learning_rate": 1.4178873534087224e-06,
"loss": 1.1740843057632446,
"step": 2276
},
{
"epoch": 2.609392898052692,
"grad_norm": 0.5724326372146606,
"learning_rate": 1.413117861998411e-06,
"loss": 0.5848703384399414,
"step": 2278
},
{
"epoch": 2.611683848797251,
"grad_norm": 0.26921916007995605,
"learning_rate": 1.4083744351762568e-06,
"loss": 1.0210750102996826,
"step": 2280
},
{
"epoch": 2.61397479954181,
"grad_norm": 0.24203850328922272,
"learning_rate": 1.4036571031940433e-06,
"loss": 0.932150661945343,
"step": 2282
},
{
"epoch": 2.616265750286369,
"grad_norm": 1.6473098993301392,
"learning_rate": 1.398965896137126e-06,
"loss": 0.7068231105804443,
"step": 2284
},
{
"epoch": 2.618556701030928,
"grad_norm": 0.41283610463142395,
"learning_rate": 1.394300843924251e-06,
"loss": 1.108357310295105,
"step": 2286
},
{
"epoch": 2.620847651775487,
"grad_norm": 0.3008173108100891,
"learning_rate": 1.3896619763073558e-06,
"loss": 0.8680709600448608,
"step": 2288
},
{
"epoch": 2.623138602520046,
"grad_norm": 0.3753491938114166,
"learning_rate": 1.3850493228713829e-06,
"loss": 1.1014978885650635,
"step": 2290
},
{
"epoch": 2.625429553264605,
"grad_norm": 0.4968052804470062,
"learning_rate": 1.3804629130340922e-06,
"loss": 0.7989722490310669,
"step": 2292
},
{
"epoch": 2.627720504009164,
"grad_norm": 1.7983485460281372,
"learning_rate": 1.375902776045869e-06,
"loss": 1.049964427947998,
"step": 2294
},
{
"epoch": 2.6300114547537228,
"grad_norm": 1.96951425075531,
"learning_rate": 1.3713689409895423e-06,
"loss": 0.9360839128494263,
"step": 2296
},
{
"epoch": 2.6323024054982818,
"grad_norm": 0.43087831139564514,
"learning_rate": 1.3668614367801968e-06,
"loss": 1.0364996194839478,
"step": 2298
},
{
"epoch": 2.6345933562428407,
"grad_norm": 1.26382577419281,
"learning_rate": 1.3623802921649903e-06,
"loss": 1.0188556909561157,
"step": 2300
},
{
"epoch": 2.6368843069873997,
"grad_norm": 0.20739609003067017,
"learning_rate": 1.3579255357229665e-06,
"loss": 1.0033255815505981,
"step": 2302
},
{
"epoch": 2.6391752577319587,
"grad_norm": 0.3471895754337311,
"learning_rate": 1.3534971958648763e-06,
"loss": 1.0234272480010986,
"step": 2304
},
{
"epoch": 2.6414662084765177,
"grad_norm": 0.7787106037139893,
"learning_rate": 1.3490953008329976e-06,
"loss": 1.0777220726013184,
"step": 2306
},
{
"epoch": 2.6437571592210767,
"grad_norm": 1.1166505813598633,
"learning_rate": 1.34471987870095e-06,
"loss": 1.075850009918213,
"step": 2308
},
{
"epoch": 2.6460481099656357,
"grad_norm": 0.6703611016273499,
"learning_rate": 1.3403709573735226e-06,
"loss": 1.053313136100769,
"step": 2310
},
{
"epoch": 2.6483390607101946,
"grad_norm": 0.7650402188301086,
"learning_rate": 1.336048564586489e-06,
"loss": 1.1778388023376465,
"step": 2312
},
{
"epoch": 2.6506300114547536,
"grad_norm": 1.5347474813461304,
"learning_rate": 1.3317527279064346e-06,
"loss": 0.9594372510910034,
"step": 2314
},
{
"epoch": 2.6529209621993126,
"grad_norm": 0.3883764147758484,
"learning_rate": 1.3274834747305816e-06,
"loss": 1.1035804748535156,
"step": 2316
},
{
"epoch": 2.6552119129438716,
"grad_norm": 0.30928364396095276,
"learning_rate": 1.3232408322866112e-06,
"loss": 1.2312179803848267,
"step": 2318
},
{
"epoch": 2.6575028636884306,
"grad_norm": 0.4253048598766327,
"learning_rate": 1.319024827632493e-06,
"loss": 1.101532220840454,
"step": 2320
},
{
"epoch": 2.6597938144329896,
"grad_norm": 0.26640430092811584,
"learning_rate": 1.3148354876563078e-06,
"loss": 0.8757455945014954,
"step": 2322
},
{
"epoch": 2.6620847651775486,
"grad_norm": 0.42339736223220825,
"learning_rate": 1.3106728390760822e-06,
"loss": 0.9791202545166016,
"step": 2324
},
{
"epoch": 2.6643757159221075,
"grad_norm": 0.7778316736221313,
"learning_rate": 1.3065369084396143e-06,
"loss": 1.1166675090789795,
"step": 2326
},
{
"epoch": 2.6666666666666665,
"grad_norm": 0.41248399019241333,
"learning_rate": 1.302427722124305e-06,
"loss": 1.0665652751922607,
"step": 2328
},
{
"epoch": 2.6689576174112255,
"grad_norm": 2.0227549076080322,
"learning_rate": 1.2983453063369886e-06,
"loss": 1.1030595302581787,
"step": 2330
},
{
"epoch": 2.6712485681557845,
"grad_norm": 0.12428312003612518,
"learning_rate": 1.2942896871137703e-06,
"loss": 0.1532021164894104,
"step": 2332
},
{
"epoch": 2.673539518900344,
"grad_norm": 1.4765028953552246,
"learning_rate": 1.290260890319854e-06,
"loss": 1.0765687227249146,
"step": 2334
},
{
"epoch": 2.675830469644903,
"grad_norm": 1.4312635660171509,
"learning_rate": 1.2862589416493828e-06,
"loss": 1.2551358938217163,
"step": 2336
},
{
"epoch": 2.678121420389462,
"grad_norm": 0.49947378039360046,
"learning_rate": 1.282283866625271e-06,
"loss": 1.1093734502792358,
"step": 2338
},
{
"epoch": 2.680412371134021,
"grad_norm": 0.8598469495773315,
"learning_rate": 1.278335690599042e-06,
"loss": 1.1796274185180664,
"step": 2340
},
{
"epoch": 2.68270332187858,
"grad_norm": 3.3485891819000244,
"learning_rate": 1.27441443875067e-06,
"loss": 1.1467323303222656,
"step": 2342
},
{
"epoch": 2.684994272623139,
"grad_norm": 1.1836572885513306,
"learning_rate": 1.2705201360884154e-06,
"loss": 0.5701313018798828,
"step": 2344
},
{
"epoch": 2.687285223367698,
"grad_norm": 0.9335172176361084,
"learning_rate": 1.2666528074486675e-06,
"loss": 1.0899162292480469,
"step": 2346
},
{
"epoch": 2.689576174112257,
"grad_norm": 0.257024884223938,
"learning_rate": 1.2628124774957862e-06,
"loss": 1.023752212524414,
"step": 2348
},
{
"epoch": 2.691867124856816,
"grad_norm": 0.5289345979690552,
"learning_rate": 1.2589991707219407e-06,
"loss": 0.8916153907775879,
"step": 2350
},
{
"epoch": 2.6941580756013748,
"grad_norm": 0.32408076524734497,
"learning_rate": 1.255212911446962e-06,
"loss": 1.040621042251587,
"step": 2352
},
{
"epoch": 2.6964490263459338,
"grad_norm": 0.39971452951431274,
"learning_rate": 1.2514537238181764e-06,
"loss": 0.8947229385375977,
"step": 2354
},
{
"epoch": 2.6987399770904927,
"grad_norm": 0.23209726810455322,
"learning_rate": 1.2477216318102638e-06,
"loss": 1.1119811534881592,
"step": 2356
},
{
"epoch": 2.7010309278350517,
"grad_norm": 1.079315185546875,
"learning_rate": 1.2440166592250908e-06,
"loss": 1.105717420578003,
"step": 2358
},
{
"epoch": 2.7033218785796107,
"grad_norm": 0.41968509554862976,
"learning_rate": 1.2403388296915748e-06,
"loss": 0.9688929915428162,
"step": 2360
},
{
"epoch": 2.7056128293241697,
"grad_norm": 0.34683892130851746,
"learning_rate": 1.2366881666655197e-06,
"loss": 1.0027223825454712,
"step": 2362
},
{
"epoch": 2.7079037800687287,
"grad_norm": 1.22856867313385,
"learning_rate": 1.2330646934294733e-06,
"loss": 0.8159598112106323,
"step": 2364
},
{
"epoch": 2.7101947308132877,
"grad_norm": 0.4460030198097229,
"learning_rate": 1.2294684330925773e-06,
"loss": 1.0731245279312134,
"step": 2366
},
{
"epoch": 2.7124856815578466,
"grad_norm": 0.23822948336601257,
"learning_rate": 1.225899408590418e-06,
"loss": 1.0538825988769531,
"step": 2368
},
{
"epoch": 2.7147766323024056,
"grad_norm": 0.5003798007965088,
"learning_rate": 1.222357642684886e-06,
"loss": 0.9127865433692932,
"step": 2370
},
{
"epoch": 2.7170675830469646,
"grad_norm": 0.18163853883743286,
"learning_rate": 1.2188431579640229e-06,
"loss": 0.5616147518157959,
"step": 2372
},
{
"epoch": 2.7193585337915236,
"grad_norm": 1.1338001489639282,
"learning_rate": 1.2153559768418826e-06,
"loss": 1.188947081565857,
"step": 2374
},
{
"epoch": 2.7216494845360826,
"grad_norm": 0.4888297915458679,
"learning_rate": 1.2118961215583876e-06,
"loss": 0.907243013381958,
"step": 2376
},
{
"epoch": 2.7239404352806416,
"grad_norm": 0.5065628886222839,
"learning_rate": 1.2084636141791853e-06,
"loss": 1.178124189376831,
"step": 2378
},
{
"epoch": 2.7262313860252005,
"grad_norm": 0.6853643655776978,
"learning_rate": 1.2050584765955106e-06,
"loss": 0.5477138757705688,
"step": 2380
},
{
"epoch": 2.7285223367697595,
"grad_norm": 1.9078017473220825,
"learning_rate": 1.2016807305240418e-06,
"loss": 1.0968685150146484,
"step": 2382
},
{
"epoch": 2.7308132875143185,
"grad_norm": 0.4655286967754364,
"learning_rate": 1.198330397506765e-06,
"loss": 1.0445671081542969,
"step": 2384
},
{
"epoch": 2.7331042382588775,
"grad_norm": 0.6571381688117981,
"learning_rate": 1.1950074989108369e-06,
"loss": 1.2365548610687256,
"step": 2386
},
{
"epoch": 2.7353951890034365,
"grad_norm": 0.1842108517885208,
"learning_rate": 1.1917120559284484e-06,
"loss": 1.0173747539520264,
"step": 2388
},
{
"epoch": 2.7376861397479955,
"grad_norm": 0.7447629570960999,
"learning_rate": 1.188444089576689e-06,
"loss": 1.1226527690887451,
"step": 2390
},
{
"epoch": 2.7399770904925544,
"grad_norm": 1.3908346891403198,
"learning_rate": 1.1852036206974093e-06,
"loss": 1.0542950630187988,
"step": 2392
},
{
"epoch": 2.7422680412371134,
"grad_norm": 0.5131960511207581,
"learning_rate": 1.1819906699570951e-06,
"loss": 0.9882605075836182,
"step": 2394
},
{
"epoch": 2.7445589919816724,
"grad_norm": 1.6375041007995605,
"learning_rate": 1.178805257846731e-06,
"loss": 0.8442957401275635,
"step": 2396
},
{
"epoch": 2.7468499427262314,
"grad_norm": 0.6786810159683228,
"learning_rate": 1.1756474046816705e-06,
"loss": 1.0617409944534302,
"step": 2398
},
{
"epoch": 2.7491408934707904,
"grad_norm": 0.3984721899032593,
"learning_rate": 1.172517130601506e-06,
"loss": 0.9120774269104004,
"step": 2400
},
{
"epoch": 2.7514318442153494,
"grad_norm": 0.8128576874732971,
"learning_rate": 1.169414455569941e-06,
"loss": 1.0682992935180664,
"step": 2402
},
{
"epoch": 2.7537227949599083,
"grad_norm": 0.6351785063743591,
"learning_rate": 1.1663393993746639e-06,
"loss": 0.6914603114128113,
"step": 2404
},
{
"epoch": 2.7560137457044673,
"grad_norm": 0.9976605176925659,
"learning_rate": 1.1632919816272185e-06,
"loss": 1.171576976776123,
"step": 2406
},
{
"epoch": 2.7583046964490263,
"grad_norm": 0.303443044424057,
"learning_rate": 1.1602722217628847e-06,
"loss": 0.5294689536094666,
"step": 2408
},
{
"epoch": 2.7605956471935853,
"grad_norm": 0.26282933354377747,
"learning_rate": 1.157280139040546e-06,
"loss": 1.1015565395355225,
"step": 2410
},
{
"epoch": 2.7628865979381443,
"grad_norm": 0.6320168375968933,
"learning_rate": 1.1543157525425753e-06,
"loss": 1.0779060125350952,
"step": 2412
},
{
"epoch": 2.7651775486827033,
"grad_norm": 0.8445232510566711,
"learning_rate": 1.1513790811747077e-06,
"loss": 1.0145206451416016,
"step": 2414
},
{
"epoch": 2.7674684994272623,
"grad_norm": 1.2781503200531006,
"learning_rate": 1.1484701436659228e-06,
"loss": 0.9530794620513916,
"step": 2416
},
{
"epoch": 2.7697594501718212,
"grad_norm": 0.4669745862483978,
"learning_rate": 1.1455889585683234e-06,
"loss": 1.1770665645599365,
"step": 2418
},
{
"epoch": 2.77205040091638,
"grad_norm": 0.1694551408290863,
"learning_rate": 1.1427355442570174e-06,
"loss": 0.8900678157806396,
"step": 2420
},
{
"epoch": 2.774341351660939,
"grad_norm": 0.6487224698066711,
"learning_rate": 1.1399099189300022e-06,
"loss": 1.1467430591583252,
"step": 2422
},
{
"epoch": 2.776632302405498,
"grad_norm": 0.4884292483329773,
"learning_rate": 1.1371121006080476e-06,
"loss": 0.6432940363883972,
"step": 2424
},
{
"epoch": 2.778923253150057,
"grad_norm": 0.5796436071395874,
"learning_rate": 1.1343421071345807e-06,
"loss": 0.8554515838623047,
"step": 2426
},
{
"epoch": 2.781214203894616,
"grad_norm": 0.6898307800292969,
"learning_rate": 1.1315999561755704e-06,
"loss": 1.0714622735977173,
"step": 2428
},
{
"epoch": 2.783505154639175,
"grad_norm": 0.8374386429786682,
"learning_rate": 1.1288856652194187e-06,
"loss": 1.2249865531921387,
"step": 2430
},
{
"epoch": 2.785796105383734,
"grad_norm": 0.5728365778923035,
"learning_rate": 1.1261992515768475e-06,
"loss": 1.0746301412582397,
"step": 2432
},
{
"epoch": 2.788087056128293,
"grad_norm": 0.5857115387916565,
"learning_rate": 1.1235407323807863e-06,
"loss": 0.9600643515586853,
"step": 2434
},
{
"epoch": 2.790378006872852,
"grad_norm": 0.43586304783821106,
"learning_rate": 1.1209101245862641e-06,
"loss": 1.0450245141983032,
"step": 2436
},
{
"epoch": 2.792668957617411,
"grad_norm": 0.9094945192337036,
"learning_rate": 1.1183074449703026e-06,
"loss": 1.141032099723816,
"step": 2438
},
{
"epoch": 2.79495990836197,
"grad_norm": 0.5066821575164795,
"learning_rate": 1.1157327101318079e-06,
"loss": 1.046962022781372,
"step": 2440
},
{
"epoch": 2.797250859106529,
"grad_norm": 0.302134245634079,
"learning_rate": 1.113185936491464e-06,
"loss": 0.7253270149230957,
"step": 2442
},
{
"epoch": 2.799541809851088,
"grad_norm": 0.46006661653518677,
"learning_rate": 1.1106671402916311e-06,
"loss": 1.059772253036499,
"step": 2444
},
{
"epoch": 2.801832760595647,
"grad_norm": 0.4403242766857147,
"learning_rate": 1.1081763375962376e-06,
"loss": 1.0585718154907227,
"step": 2446
},
{
"epoch": 2.804123711340206,
"grad_norm": 0.6102672219276428,
"learning_rate": 1.1057135442906802e-06,
"loss": 1.0364320278167725,
"step": 2448
},
{
"epoch": 2.806414662084765,
"grad_norm": 0.2114000916481018,
"learning_rate": 1.1032787760817245e-06,
"loss": 1.0323476791381836,
"step": 2450
},
{
"epoch": 2.808705612829324,
"grad_norm": 0.44233590364456177,
"learning_rate": 1.1008720484973997e-06,
"loss": 1.0494945049285889,
"step": 2452
},
{
"epoch": 2.810996563573883,
"grad_norm": 0.5234547853469849,
"learning_rate": 1.0984933768869045e-06,
"loss": 0.8871585130691528,
"step": 2454
},
{
"epoch": 2.813287514318442,
"grad_norm": 2.157639741897583,
"learning_rate": 1.0961427764205054e-06,
"loss": 0.9773741960525513,
"step": 2456
},
{
"epoch": 2.815578465063001,
"grad_norm": 0.7899772524833679,
"learning_rate": 1.0938202620894444e-06,
"loss": 1.0361276865005493,
"step": 2458
},
{
"epoch": 2.81786941580756,
"grad_norm": 0.3699988126754761,
"learning_rate": 1.0915258487058393e-06,
"loss": 0.9854030609130859,
"step": 2460
},
{
"epoch": 2.820160366552119,
"grad_norm": 0.5396907329559326,
"learning_rate": 1.0892595509025897e-06,
"loss": 1.0175352096557617,
"step": 2462
},
{
"epoch": 2.822451317296678,
"grad_norm": 0.5068468451499939,
"learning_rate": 1.0870213831332856e-06,
"loss": 1.0485467910766602,
"step": 2464
},
{
"epoch": 2.824742268041237,
"grad_norm": 0.43589094281196594,
"learning_rate": 1.0848113596721147e-06,
"loss": 0.9287646412849426,
"step": 2466
},
{
"epoch": 2.827033218785796,
"grad_norm": 0.4347958266735077,
"learning_rate": 1.0826294946137705e-06,
"loss": 1.1217800378799438,
"step": 2468
},
{
"epoch": 2.8293241695303553,
"grad_norm": 0.5516931414604187,
"learning_rate": 1.0804758018733635e-06,
"loss": 1.1510246992111206,
"step": 2470
},
{
"epoch": 2.8316151202749142,
"grad_norm": 1.0004255771636963,
"learning_rate": 1.0783502951863302e-06,
"loss": 0.7378605604171753,
"step": 2472
},
{
"epoch": 2.8339060710194732,
"grad_norm": 0.43937617540359497,
"learning_rate": 1.076252988108349e-06,
"loss": 1.0908252000808716,
"step": 2474
},
{
"epoch": 2.836197021764032,
"grad_norm": 0.5650505423545837,
"learning_rate": 1.0741838940152498e-06,
"loss": 1.0471761226654053,
"step": 2476
},
{
"epoch": 2.838487972508591,
"grad_norm": 0.5975066423416138,
"learning_rate": 1.0721430261029334e-06,
"loss": 1.0682255029678345,
"step": 2478
},
{
"epoch": 2.84077892325315,
"grad_norm": 0.3658738136291504,
"learning_rate": 1.0701303973872825e-06,
"loss": 1.1062527894973755,
"step": 2480
},
{
"epoch": 2.843069873997709,
"grad_norm": 0.29034459590911865,
"learning_rate": 1.0681460207040816e-06,
"loss": 0.9529769420623779,
"step": 2482
},
{
"epoch": 2.845360824742268,
"grad_norm": 1.0845448970794678,
"learning_rate": 1.066189908708935e-06,
"loss": 1.1356077194213867,
"step": 2484
},
{
"epoch": 2.847651775486827,
"grad_norm": 0.6470354199409485,
"learning_rate": 1.0642620738771844e-06,
"loss": 1.1769475936889648,
"step": 2486
},
{
"epoch": 2.849942726231386,
"grad_norm": 0.30026912689208984,
"learning_rate": 1.0623625285038317e-06,
"loss": 1.2107105255126953,
"step": 2488
},
{
"epoch": 2.852233676975945,
"grad_norm": 0.4359968900680542,
"learning_rate": 1.0604912847034579e-06,
"loss": 1.1430937051773071,
"step": 2490
},
{
"epoch": 2.854524627720504,
"grad_norm": 0.9392463564872742,
"learning_rate": 1.0586483544101484e-06,
"loss": 0.9589693546295166,
"step": 2492
},
{
"epoch": 2.856815578465063,
"grad_norm": 0.5891343951225281,
"learning_rate": 1.0568337493774153e-06,
"loss": 1.0557307004928589,
"step": 2494
},
{
"epoch": 2.859106529209622,
"grad_norm": 0.40429064631462097,
"learning_rate": 1.055047481178123e-06,
"loss": 1.2271441221237183,
"step": 2496
},
{
"epoch": 2.861397479954181,
"grad_norm": 0.5770910382270813,
"learning_rate": 1.0532895612044148e-06,
"loss": 1.2372446060180664,
"step": 2498
},
{
"epoch": 2.86368843069874,
"grad_norm": 0.7154263854026794,
"learning_rate": 1.0515600006676388e-06,
"loss": 1.0392417907714844,
"step": 2500
},
{
"epoch": 2.865979381443299,
"grad_norm": 0.37973713874816895,
"learning_rate": 1.0498588105982773e-06,
"loss": 1.0606379508972168,
"step": 2502
},
{
"epoch": 2.868270332187858,
"grad_norm": 0.41982316970825195,
"learning_rate": 1.0481860018458777e-06,
"loss": 0.9914207458496094,
"step": 2504
},
{
"epoch": 2.870561282932417,
"grad_norm": 0.6768814325332642,
"learning_rate": 1.0465415850789818e-06,
"loss": 1.011329174041748,
"step": 2506
},
{
"epoch": 2.872852233676976,
"grad_norm": 0.5299584269523621,
"learning_rate": 1.044925570785057e-06,
"loss": 1.0899381637573242,
"step": 2508
},
{
"epoch": 2.875143184421535,
"grad_norm": 0.6537067294120789,
"learning_rate": 1.0433379692704303e-06,
"loss": 0.8742640018463135,
"step": 2510
},
{
"epoch": 2.877434135166094,
"grad_norm": 0.7677899599075317,
"learning_rate": 1.0417787906602254e-06,
"loss": 1.1713974475860596,
"step": 2512
},
{
"epoch": 2.879725085910653,
"grad_norm": 1.137424349784851,
"learning_rate": 1.0402480448982916e-06,
"loss": 1.0399460792541504,
"step": 2514
},
{
"epoch": 2.882016036655212,
"grad_norm": 5.416769027709961,
"learning_rate": 1.0387457417471484e-06,
"loss": 0.7440576553344727,
"step": 2516
},
{
"epoch": 2.884306987399771,
"grad_norm": 0.3989621698856354,
"learning_rate": 1.0372718907879148e-06,
"loss": 0.6085208058357239,
"step": 2518
},
{
"epoch": 2.88659793814433,
"grad_norm": 0.39145538210868835,
"learning_rate": 1.0358265014202562e-06,
"loss": 1.0786077976226807,
"step": 2520
},
{
"epoch": 2.888888888888889,
"grad_norm": 0.6507622003555298,
"learning_rate": 1.0344095828623187e-06,
"loss": 1.1602535247802734,
"step": 2522
},
{
"epoch": 2.891179839633448,
"grad_norm": 0.5639922022819519,
"learning_rate": 1.0330211441506725e-06,
"loss": 1.2351245880126953,
"step": 2524
},
{
"epoch": 2.893470790378007,
"grad_norm": 2.3547637462615967,
"learning_rate": 1.0316611941402554e-06,
"loss": 1.2815968990325928,
"step": 2526
},
{
"epoch": 2.895761741122566,
"grad_norm": 0.5312323570251465,
"learning_rate": 1.0303297415043125e-06,
"loss": 1.054031491279602,
"step": 2528
},
{
"epoch": 2.8980526918671248,
"grad_norm": 0.10760916769504547,
"learning_rate": 1.0290267947343467e-06,
"loss": 0.5422964692115784,
"step": 2530
},
{
"epoch": 2.9003436426116838,
"grad_norm": 0.5096286535263062,
"learning_rate": 1.027752362140058e-06,
"loss": 0.909081220626831,
"step": 2532
},
{
"epoch": 2.9026345933562427,
"grad_norm": 0.9880173802375793,
"learning_rate": 1.0265064518492971e-06,
"loss": 0.6273051500320435,
"step": 2534
},
{
"epoch": 2.9049255441008017,
"grad_norm": 0.2785288393497467,
"learning_rate": 1.0252890718080065e-06,
"loss": 0.9630579948425293,
"step": 2536
},
{
"epoch": 2.9072164948453607,
"grad_norm": 0.9245594143867493,
"learning_rate": 1.024100229780177e-06,
"loss": 0.8256567716598511,
"step": 2538
},
{
"epoch": 2.9095074455899197,
"grad_norm": 0.6933486461639404,
"learning_rate": 1.0229399333477933e-06,
"loss": 1.098947286605835,
"step": 2540
},
{
"epoch": 2.9117983963344787,
"grad_norm": 0.5825735330581665,
"learning_rate": 1.0218081899107866e-06,
"loss": 1.055496335029602,
"step": 2542
},
{
"epoch": 2.9140893470790377,
"grad_norm": 0.6609728336334229,
"learning_rate": 1.0207050066869884e-06,
"loss": 1.1041927337646484,
"step": 2544
},
{
"epoch": 2.9163802978235966,
"grad_norm": 0.6976462602615356,
"learning_rate": 1.0196303907120835e-06,
"loss": 1.2125544548034668,
"step": 2546
},
{
"epoch": 2.9186712485681556,
"grad_norm": 0.5264245867729187,
"learning_rate": 1.0185843488395675e-06,
"loss": 1.159142017364502,
"step": 2548
},
{
"epoch": 2.9209621993127146,
"grad_norm": 0.5094210505485535,
"learning_rate": 1.017566887740698e-06,
"loss": 1.1290173530578613,
"step": 2550
},
{
"epoch": 2.923253150057274,
"grad_norm": 1.0814720392227173,
"learning_rate": 1.0165780139044579e-06,
"loss": 1.0745420455932617,
"step": 2552
},
{
"epoch": 2.925544100801833,
"grad_norm": 0.922420084476471,
"learning_rate": 1.0156177336375104e-06,
"loss": 1.032870888710022,
"step": 2554
},
{
"epoch": 2.927835051546392,
"grad_norm": 2.324207067489624,
"learning_rate": 1.0146860530641597e-06,
"loss": 1.0902037620544434,
"step": 2556
},
{
"epoch": 2.930126002290951,
"grad_norm": 0.6379292607307434,
"learning_rate": 1.013782978126312e-06,
"loss": 1.1656874418258667,
"step": 2558
},
{
"epoch": 2.93241695303551,
"grad_norm": 0.3427787721157074,
"learning_rate": 1.0129085145834374e-06,
"loss": 0.8560712337493896,
"step": 2560
},
{
"epoch": 2.934707903780069,
"grad_norm": 1.8992148637771606,
"learning_rate": 1.012062668012534e-06,
"loss": 0.7727330327033997,
"step": 2562
},
{
"epoch": 2.936998854524628,
"grad_norm": 0.9722429513931274,
"learning_rate": 1.0112454438080912e-06,
"loss": 0.9743561744689941,
"step": 2564
},
{
"epoch": 2.939289805269187,
"grad_norm": 0.20309951901435852,
"learning_rate": 1.010456847182055e-06,
"loss": 1.0358002185821533,
"step": 2566
},
{
"epoch": 2.941580756013746,
"grad_norm": 0.3222387731075287,
"learning_rate": 1.0096968831637976e-06,
"loss": 1.0403361320495605,
"step": 2568
},
{
"epoch": 2.943871706758305,
"grad_norm": 0.5108765363693237,
"learning_rate": 1.008965556600081e-06,
"loss": 1.0720305442810059,
"step": 2570
},
{
"epoch": 2.946162657502864,
"grad_norm": 0.6678656935691833,
"learning_rate": 1.0082628721550296e-06,
"loss": 1.0465478897094727,
"step": 2572
},
{
"epoch": 2.948453608247423,
"grad_norm": 0.264837384223938,
"learning_rate": 1.0075888343100988e-06,
"loss": 1.0957000255584717,
"step": 2574
},
{
"epoch": 2.950744558991982,
"grad_norm": 0.42729780077934265,
"learning_rate": 1.0069434473640478e-06,
"loss": 0.6375606060028076,
"step": 2576
},
{
"epoch": 2.953035509736541,
"grad_norm": 0.5195022225379944,
"learning_rate": 1.0063267154329093e-06,
"loss": 1.1245808601379395,
"step": 2578
},
{
"epoch": 2.9553264604811,
"grad_norm": 0.3333146572113037,
"learning_rate": 1.0057386424499674e-06,
"loss": 1.0144271850585938,
"step": 2580
},
{
"epoch": 2.957617411225659,
"grad_norm": 0.1994045525789261,
"learning_rate": 1.005179232165728e-06,
"loss": 1.0108163356781006,
"step": 2582
},
{
"epoch": 2.959908361970218,
"grad_norm": 0.7289289832115173,
"learning_rate": 1.0046484881478987e-06,
"loss": 1.0791220664978027,
"step": 2584
},
{
"epoch": 2.9621993127147768,
"grad_norm": 0.6395173072814941,
"learning_rate": 1.0041464137813643e-06,
"loss": 1.0681736469268799,
"step": 2586
},
{
"epoch": 2.9644902634593358,
"grad_norm": 0.46986323595046997,
"learning_rate": 1.0036730122681644e-06,
"loss": 1.0714763402938843,
"step": 2588
},
{
"epoch": 2.9667812142038947,
"grad_norm": 0.4828946590423584,
"learning_rate": 1.003228286627476e-06,
"loss": 1.036878228187561,
"step": 2590
},
{
"epoch": 2.9690721649484537,
"grad_norm": 0.661065399646759,
"learning_rate": 1.00281223969559e-06,
"loss": 1.086197853088379,
"step": 2592
},
{
"epoch": 2.9713631156930127,
"grad_norm": 1.195217490196228,
"learning_rate": 1.0024248741258968e-06,
"loss": 0.8736682534217834,
"step": 2594
},
{
"epoch": 2.9736540664375717,
"grad_norm": 0.887296199798584,
"learning_rate": 1.0020661923888674e-06,
"loss": 1.1362636089324951,
"step": 2596
},
{
"epoch": 2.9759450171821307,
"grad_norm": 0.5010428428649902,
"learning_rate": 1.001736196772038e-06,
"loss": 1.0959832668304443,
"step": 2598
},
{
"epoch": 2.9782359679266897,
"grad_norm": 0.4214611351490021,
"learning_rate": 1.0014348893799954e-06,
"loss": 1.2732155323028564,
"step": 2600
},
{
"epoch": 2.9805269186712486,
"grad_norm": 0.6986483335494995,
"learning_rate": 1.0011622721343655e-06,
"loss": 1.0966172218322754,
"step": 2602
},
{
"epoch": 2.9828178694158076,
"grad_norm": 0.5451075434684753,
"learning_rate": 1.0009183467737969e-06,
"loss": 0.8794732093811035,
"step": 2604
},
{
"epoch": 2.9851088201603666,
"grad_norm": 0.4121303856372833,
"learning_rate": 1.0007031148539529e-06,
"loss": 1.0295562744140625,
"step": 2606
},
{
"epoch": 2.9873997709049256,
"grad_norm": 0.9556783437728882,
"learning_rate": 1.0005165777475022e-06,
"loss": 0.5639452338218689,
"step": 2608
},
{
"epoch": 2.9896907216494846,
"grad_norm": 1.1290655136108398,
"learning_rate": 1.000358736644108e-06,
"loss": 0.4423694610595703,
"step": 2610
},
{
"epoch": 2.9919816723940436,
"grad_norm": 0.21035031974315643,
"learning_rate": 1.00022959255042e-06,
"loss": 1.0184959173202515,
"step": 2612
},
{
"epoch": 2.9942726231386025,
"grad_norm": 1.0605756044387817,
"learning_rate": 1.0001291462900717e-06,
"loss": 1.0759494304656982,
"step": 2614
},
{
"epoch": 2.9965635738831615,
"grad_norm": 0.7549898624420166,
"learning_rate": 1.0000573985036708e-06,
"loss": 0.9966956973075867,
"step": 2616
},
{
"epoch": 2.9988545246277205,
"grad_norm": 0.7441288232803345,
"learning_rate": 1.0000143496487968e-06,
"loss": 1.153484582901001,
"step": 2618
},
{
"epoch": 3.0,
"step": 2619,
"total_flos": 4.688730399671583e+18,
"train_loss": 1.122755285885872,
"train_runtime": 13889.3173,
"train_samples_per_second": 3.017,
"train_steps_per_second": 0.189
}
],
"logging_steps": 2,
"max_steps": 2619,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 9999999,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 4.688730399671583e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}