9b-21 / trainer_state.json
furproxy's picture
Upload folder using huggingface_hub
3e8e6ee 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": 0.2425529807806015,
"learning_rate": 7.633587786259542e-08,
"loss": 1.4360580444335938,
"step": 2
},
{
"epoch": 0.004581901489117984,
"grad_norm": 0.3157375752925873,
"learning_rate": 2.2900763358778629e-07,
"loss": 1.9720795154571533,
"step": 4
},
{
"epoch": 0.006872852233676976,
"grad_norm": 0.3054039776325226,
"learning_rate": 3.8167938931297716e-07,
"loss": 1.8266024589538574,
"step": 6
},
{
"epoch": 0.009163802978235968,
"grad_norm": 0.7713625431060791,
"learning_rate": 5.34351145038168e-07,
"loss": 2.4644742012023926,
"step": 8
},
{
"epoch": 0.011454753722794959,
"grad_norm": 0.17603641748428345,
"learning_rate": 6.870229007633589e-07,
"loss": 1.273919939994812,
"step": 10
},
{
"epoch": 0.013745704467353952,
"grad_norm": 0.12217849493026733,
"learning_rate": 8.396946564885497e-07,
"loss": 1.6275849342346191,
"step": 12
},
{
"epoch": 0.016036655211912942,
"grad_norm": 0.4887905418872833,
"learning_rate": 9.923664122137404e-07,
"loss": 1.9397943019866943,
"step": 14
},
{
"epoch": 0.018327605956471937,
"grad_norm": 0.14639756083488464,
"learning_rate": 1.1450381679389313e-06,
"loss": 1.5786923170089722,
"step": 16
},
{
"epoch": 0.020618556701030927,
"grad_norm": 1.0165174007415771,
"learning_rate": 1.297709923664122e-06,
"loss": 2.2075228691101074,
"step": 18
},
{
"epoch": 0.022909507445589918,
"grad_norm": 0.6907394528388977,
"learning_rate": 1.450381679389313e-06,
"loss": 2.0885233879089355,
"step": 20
},
{
"epoch": 0.025200458190148912,
"grad_norm": 0.5898762941360474,
"learning_rate": 1.603053435114504e-06,
"loss": 1.5722987651824951,
"step": 22
},
{
"epoch": 0.027491408934707903,
"grad_norm": 0.28665709495544434,
"learning_rate": 1.7557251908396948e-06,
"loss": 1.920871376991272,
"step": 24
},
{
"epoch": 0.029782359679266894,
"grad_norm": 0.2708616256713867,
"learning_rate": 1.908396946564886e-06,
"loss": 1.9028164148330688,
"step": 26
},
{
"epoch": 0.032073310423825885,
"grad_norm": 0.12096082419157028,
"learning_rate": 2.0610687022900764e-06,
"loss": 1.724308967590332,
"step": 28
},
{
"epoch": 0.03436426116838488,
"grad_norm": 0.3193468749523163,
"learning_rate": 2.2137404580152674e-06,
"loss": 2.3334269523620605,
"step": 30
},
{
"epoch": 0.03665521191294387,
"grad_norm": 0.25921180844306946,
"learning_rate": 2.3664122137404585e-06,
"loss": 1.7049498558044434,
"step": 32
},
{
"epoch": 0.038946162657502864,
"grad_norm": 0.2873923182487488,
"learning_rate": 2.5190839694656487e-06,
"loss": 1.8035858869552612,
"step": 34
},
{
"epoch": 0.041237113402061855,
"grad_norm": 0.2893233895301819,
"learning_rate": 2.67175572519084e-06,
"loss": 1.8687759637832642,
"step": 36
},
{
"epoch": 0.043528064146620846,
"grad_norm": 0.3628486394882202,
"learning_rate": 2.824427480916031e-06,
"loss": 1.5172038078308105,
"step": 38
},
{
"epoch": 0.045819014891179836,
"grad_norm": 0.14311300218105316,
"learning_rate": 2.9770992366412218e-06,
"loss": 1.3586844205856323,
"step": 40
},
{
"epoch": 0.048109965635738834,
"grad_norm": 0.2167440503835678,
"learning_rate": 3.129770992366413e-06,
"loss": 1.8630679845809937,
"step": 42
},
{
"epoch": 0.050400916380297825,
"grad_norm": 0.28841111063957214,
"learning_rate": 3.2824427480916034e-06,
"loss": 1.8234974145889282,
"step": 44
},
{
"epoch": 0.052691867124856816,
"grad_norm": 1.6244624853134155,
"learning_rate": 3.4351145038167944e-06,
"loss": 2.474839210510254,
"step": 46
},
{
"epoch": 0.054982817869415807,
"grad_norm": 0.7337155938148499,
"learning_rate": 3.587786259541985e-06,
"loss": 1.9647748470306396,
"step": 48
},
{
"epoch": 0.0572737686139748,
"grad_norm": 0.21595202386379242,
"learning_rate": 3.740458015267176e-06,
"loss": 1.7915287017822266,
"step": 50
},
{
"epoch": 0.05956471935853379,
"grad_norm": 0.12472744286060333,
"learning_rate": 3.893129770992366e-06,
"loss": 1.57389497756958,
"step": 52
},
{
"epoch": 0.061855670103092786,
"grad_norm": 0.2959010601043701,
"learning_rate": 4.045801526717557e-06,
"loss": 1.9410815238952637,
"step": 54
},
{
"epoch": 0.06414662084765177,
"grad_norm": 0.5962556600570679,
"learning_rate": 4.198473282442748e-06,
"loss": 1.5538616180419922,
"step": 56
},
{
"epoch": 0.06643757159221077,
"grad_norm": 0.223605677485466,
"learning_rate": 4.351145038167939e-06,
"loss": 2.3718690872192383,
"step": 58
},
{
"epoch": 0.06872852233676977,
"grad_norm": 2.0641846656799316,
"learning_rate": 4.5038167938931296e-06,
"loss": 2.0545084476470947,
"step": 60
},
{
"epoch": 0.07101947308132875,
"grad_norm": 0.11926179379224777,
"learning_rate": 4.656488549618321e-06,
"loss": 1.518899917602539,
"step": 62
},
{
"epoch": 0.07331042382588775,
"grad_norm": 0.5377867221832275,
"learning_rate": 4.8091603053435125e-06,
"loss": 2.0711913108825684,
"step": 64
},
{
"epoch": 0.07560137457044673,
"grad_norm": 0.20864170789718628,
"learning_rate": 4.961832061068703e-06,
"loss": 1.5393109321594238,
"step": 66
},
{
"epoch": 0.07789232531500573,
"grad_norm": 0.17572522163391113,
"learning_rate": 5.114503816793893e-06,
"loss": 1.7614434957504272,
"step": 68
},
{
"epoch": 0.08018327605956473,
"grad_norm": 0.4484013020992279,
"learning_rate": 5.267175572519084e-06,
"loss": 1.6294645071029663,
"step": 70
},
{
"epoch": 0.08247422680412371,
"grad_norm": 0.1890186220407486,
"learning_rate": 5.419847328244276e-06,
"loss": 1.6021493673324585,
"step": 72
},
{
"epoch": 0.08476517754868271,
"grad_norm": 0.19444267451763153,
"learning_rate": 5.572519083969467e-06,
"loss": 1.861649990081787,
"step": 74
},
{
"epoch": 0.08705612829324169,
"grad_norm": 0.0960668995976448,
"learning_rate": 5.725190839694656e-06,
"loss": 1.5107250213623047,
"step": 76
},
{
"epoch": 0.08934707903780069,
"grad_norm": 0.5166394114494324,
"learning_rate": 5.877862595419848e-06,
"loss": 1.977331280708313,
"step": 78
},
{
"epoch": 0.09163802978235967,
"grad_norm": 0.19248326122760773,
"learning_rate": 6.030534351145039e-06,
"loss": 1.7162344455718994,
"step": 80
},
{
"epoch": 0.09392898052691867,
"grad_norm": 0.14979946613311768,
"learning_rate": 6.18320610687023e-06,
"loss": 1.4223518371582031,
"step": 82
},
{
"epoch": 0.09621993127147767,
"grad_norm": 0.19589261710643768,
"learning_rate": 6.335877862595419e-06,
"loss": 1.7889583110809326,
"step": 84
},
{
"epoch": 0.09851088201603665,
"grad_norm": 0.10083930939435959,
"learning_rate": 6.488549618320611e-06,
"loss": 1.309556484222412,
"step": 86
},
{
"epoch": 0.10080183276059565,
"grad_norm": 0.1550842672586441,
"learning_rate": 6.641221374045802e-06,
"loss": 1.6037399768829346,
"step": 88
},
{
"epoch": 0.10309278350515463,
"grad_norm": 0.1855807900428772,
"learning_rate": 6.793893129770993e-06,
"loss": 1.73233962059021,
"step": 90
},
{
"epoch": 0.10538373424971363,
"grad_norm": 0.07532945275306702,
"learning_rate": 6.946564885496184e-06,
"loss": 1.5398590564727783,
"step": 92
},
{
"epoch": 0.10767468499427263,
"grad_norm": 0.15782028436660767,
"learning_rate": 7.0992366412213746e-06,
"loss": 1.7258206605911255,
"step": 94
},
{
"epoch": 0.10996563573883161,
"grad_norm": 0.09372120350599289,
"learning_rate": 7.251908396946566e-06,
"loss": 1.3687584400177002,
"step": 96
},
{
"epoch": 0.11225658648339061,
"grad_norm": 0.21025855839252472,
"learning_rate": 7.404580152671757e-06,
"loss": 1.646334171295166,
"step": 98
},
{
"epoch": 0.1145475372279496,
"grad_norm": 1.09811532497406,
"learning_rate": 7.557251908396948e-06,
"loss": 1.6427111625671387,
"step": 100
},
{
"epoch": 0.11683848797250859,
"grad_norm": 0.2197725772857666,
"learning_rate": 7.709923664122137e-06,
"loss": 1.7769626379013062,
"step": 102
},
{
"epoch": 0.11912943871706758,
"grad_norm": 0.06639471650123596,
"learning_rate": 7.862595419847328e-06,
"loss": 1.4597866535186768,
"step": 104
},
{
"epoch": 0.12142038946162657,
"grad_norm": 0.08875848352909088,
"learning_rate": 8.015267175572519e-06,
"loss": 1.3246572017669678,
"step": 106
},
{
"epoch": 0.12371134020618557,
"grad_norm": 0.07279494404792786,
"learning_rate": 8.16793893129771e-06,
"loss": 1.551082730293274,
"step": 108
},
{
"epoch": 0.12600229095074456,
"grad_norm": 0.06380033493041992,
"learning_rate": 8.320610687022901e-06,
"loss": 1.481245517730713,
"step": 110
},
{
"epoch": 0.12829324169530354,
"grad_norm": 0.16885097324848175,
"learning_rate": 8.473282442748092e-06,
"loss": 1.7407457828521729,
"step": 112
},
{
"epoch": 0.13058419243986255,
"grad_norm": 0.20184065401554108,
"learning_rate": 8.625954198473283e-06,
"loss": 1.5867414474487305,
"step": 114
},
{
"epoch": 0.13287514318442153,
"grad_norm": 0.20837143063545227,
"learning_rate": 8.778625954198474e-06,
"loss": 1.6812255382537842,
"step": 116
},
{
"epoch": 0.13516609392898052,
"grad_norm": 0.20289385318756104,
"learning_rate": 8.931297709923665e-06,
"loss": 1.5645579099655151,
"step": 118
},
{
"epoch": 0.13745704467353953,
"grad_norm": 0.32700014114379883,
"learning_rate": 9.083969465648855e-06,
"loss": 1.3761099576950073,
"step": 120
},
{
"epoch": 0.13974799541809851,
"grad_norm": 0.04891065135598183,
"learning_rate": 9.236641221374046e-06,
"loss": 1.300285816192627,
"step": 122
},
{
"epoch": 0.1420389461626575,
"grad_norm": 0.0589461550116539,
"learning_rate": 9.389312977099237e-06,
"loss": 1.210198163986206,
"step": 124
},
{
"epoch": 0.14432989690721648,
"grad_norm": 0.1429443210363388,
"learning_rate": 9.54198473282443e-06,
"loss": 1.5391454696655273,
"step": 126
},
{
"epoch": 0.1466208476517755,
"grad_norm": 0.16855759918689728,
"learning_rate": 9.694656488549619e-06,
"loss": 1.4557760953903198,
"step": 128
},
{
"epoch": 0.14891179839633448,
"grad_norm": 0.16834264993667603,
"learning_rate": 9.84732824427481e-06,
"loss": 1.6899962425231934,
"step": 130
},
{
"epoch": 0.15120274914089346,
"grad_norm": 0.1502954512834549,
"learning_rate": 1e-05,
"loss": 1.5119028091430664,
"step": 132
},
{
"epoch": 0.15349369988545247,
"grad_norm": 0.048700153827667236,
"learning_rate": 9.999985650351204e-06,
"loss": 1.1397861242294312,
"step": 134
},
{
"epoch": 0.15578465063001146,
"grad_norm": 0.049325715750455856,
"learning_rate": 9.999942601496331e-06,
"loss": 1.2158293724060059,
"step": 136
},
{
"epoch": 0.15807560137457044,
"grad_norm": 0.05145062878727913,
"learning_rate": 9.999870853709929e-06,
"loss": 1.4632039070129395,
"step": 138
},
{
"epoch": 0.16036655211912945,
"grad_norm": 0.040577393025159836,
"learning_rate": 9.999770407449582e-06,
"loss": 1.1962344646453857,
"step": 140
},
{
"epoch": 0.16265750286368844,
"grad_norm": 0.1325235515832901,
"learning_rate": 9.999641263355893e-06,
"loss": 1.342354416847229,
"step": 142
},
{
"epoch": 0.16494845360824742,
"grad_norm": 0.11594824492931366,
"learning_rate": 9.999483422252499e-06,
"loss": 1.4513139724731445,
"step": 144
},
{
"epoch": 0.1672394043528064,
"grad_norm": 0.05856647342443466,
"learning_rate": 9.999296885146047e-06,
"loss": 1.2211672067642212,
"step": 146
},
{
"epoch": 0.16953035509736541,
"grad_norm": 0.046867020428180695,
"learning_rate": 9.999081653226205e-06,
"loss": 1.316023588180542,
"step": 148
},
{
"epoch": 0.1718213058419244,
"grad_norm": 0.24781149625778198,
"learning_rate": 9.998837727865636e-06,
"loss": 1.1912282705307007,
"step": 150
},
{
"epoch": 0.17411225658648338,
"grad_norm": 0.04234715923666954,
"learning_rate": 9.998565110620006e-06,
"loss": 1.0865659713745117,
"step": 152
},
{
"epoch": 0.1764032073310424,
"grad_norm": 0.11648119986057281,
"learning_rate": 9.998263803227965e-06,
"loss": 1.4158616065979004,
"step": 154
},
{
"epoch": 0.17869415807560138,
"grad_norm": 0.21850669384002686,
"learning_rate": 9.997933807611133e-06,
"loss": 1.7990024089813232,
"step": 156
},
{
"epoch": 0.18098510882016036,
"grad_norm": 0.12010746449232101,
"learning_rate": 9.997575125874104e-06,
"loss": 1.472090244293213,
"step": 158
},
{
"epoch": 0.18327605956471935,
"grad_norm": 0.1091763824224472,
"learning_rate": 9.997187760304411e-06,
"loss": 1.0730650424957275,
"step": 160
},
{
"epoch": 0.18556701030927836,
"grad_norm": 0.14433470368385315,
"learning_rate": 9.996771713372525e-06,
"loss": 1.29958176612854,
"step": 162
},
{
"epoch": 0.18785796105383734,
"grad_norm": 0.12404698878526688,
"learning_rate": 9.996326987731836e-06,
"loss": 1.0624301433563232,
"step": 164
},
{
"epoch": 0.19014891179839633,
"grad_norm": 0.12703843414783478,
"learning_rate": 9.995853586218636e-06,
"loss": 1.5410776138305664,
"step": 166
},
{
"epoch": 0.19243986254295534,
"grad_norm": 2.6877777576446533,
"learning_rate": 9.995351511852102e-06,
"loss": 0.6895647048950195,
"step": 168
},
{
"epoch": 0.19473081328751432,
"grad_norm": 0.12183137238025665,
"learning_rate": 9.994820767834273e-06,
"loss": 1.4578661918640137,
"step": 170
},
{
"epoch": 0.1970217640320733,
"grad_norm": 0.17509934306144714,
"learning_rate": 9.994261357550034e-06,
"loss": 1.5483499765396118,
"step": 172
},
{
"epoch": 0.19931271477663232,
"grad_norm": 0.1465608775615692,
"learning_rate": 9.993673284567092e-06,
"loss": 1.6950434446334839,
"step": 174
},
{
"epoch": 0.2016036655211913,
"grad_norm": 0.11729174107313156,
"learning_rate": 9.993056552635954e-06,
"loss": 1.4780206680297852,
"step": 176
},
{
"epoch": 0.20389461626575028,
"grad_norm": 0.11391811072826385,
"learning_rate": 9.992411165689902e-06,
"loss": 1.4607470035552979,
"step": 178
},
{
"epoch": 0.20618556701030927,
"grad_norm": 0.24819950759410858,
"learning_rate": 9.99173712784497e-06,
"loss": 1.5854477882385254,
"step": 180
},
{
"epoch": 0.20847651775486828,
"grad_norm": 0.12101314961910248,
"learning_rate": 9.99103444339992e-06,
"loss": 1.3363440036773682,
"step": 182
},
{
"epoch": 0.21076746849942726,
"grad_norm": 0.1651735156774521,
"learning_rate": 9.990303116836204e-06,
"loss": 1.3371784687042236,
"step": 184
},
{
"epoch": 0.21305841924398625,
"grad_norm": 0.10060884058475494,
"learning_rate": 9.989543152817945e-06,
"loss": 1.1477423906326294,
"step": 186
},
{
"epoch": 0.21534936998854526,
"grad_norm": 0.12473434954881668,
"learning_rate": 9.98875455619191e-06,
"loss": 1.3203130960464478,
"step": 188
},
{
"epoch": 0.21764032073310424,
"grad_norm": 0.03830314055085182,
"learning_rate": 9.987937331987466e-06,
"loss": 1.29048752784729,
"step": 190
},
{
"epoch": 0.21993127147766323,
"grad_norm": 0.0400603786110878,
"learning_rate": 9.987091485416564e-06,
"loss": 1.1765823364257812,
"step": 192
},
{
"epoch": 0.2222222222222222,
"grad_norm": 0.1159832775592804,
"learning_rate": 9.986217021873688e-06,
"loss": 1.2849760055541992,
"step": 194
},
{
"epoch": 0.22451317296678122,
"grad_norm": 0.036686670035123825,
"learning_rate": 9.985313946935841e-06,
"loss": 1.1865625381469727,
"step": 196
},
{
"epoch": 0.2268041237113402,
"grad_norm": 0.10680826008319855,
"learning_rate": 9.98438226636249e-06,
"loss": 1.4698808193206787,
"step": 198
},
{
"epoch": 0.2290950744558992,
"grad_norm": 0.19631391763687134,
"learning_rate": 9.983421986095543e-06,
"loss": 1.5152738094329834,
"step": 200
},
{
"epoch": 0.2313860252004582,
"grad_norm": 0.2143646776676178,
"learning_rate": 9.982433112259304e-06,
"loss": 1.1555585861206055,
"step": 202
},
{
"epoch": 0.23367697594501718,
"grad_norm": 0.11695901304483414,
"learning_rate": 9.981415651160434e-06,
"loss": 1.3724945783615112,
"step": 204
},
{
"epoch": 0.23596792668957617,
"grad_norm": 0.0444357693195343,
"learning_rate": 9.980369609287918e-06,
"loss": 1.28928804397583,
"step": 206
},
{
"epoch": 0.23825887743413515,
"grad_norm": 0.2176477164030075,
"learning_rate": 9.979294993313013e-06,
"loss": 1.3569140434265137,
"step": 208
},
{
"epoch": 0.24054982817869416,
"grad_norm": 0.09549252688884735,
"learning_rate": 9.978191810089213e-06,
"loss": 1.2554432153701782,
"step": 210
},
{
"epoch": 0.24284077892325315,
"grad_norm": 0.035298705101013184,
"learning_rate": 9.977060066652208e-06,
"loss": 1.19698166847229,
"step": 212
},
{
"epoch": 0.24513172966781213,
"grad_norm": 0.03924322500824928,
"learning_rate": 9.975899770219823e-06,
"loss": 1.253554105758667,
"step": 214
},
{
"epoch": 0.24742268041237114,
"grad_norm": 0.14062990248203278,
"learning_rate": 9.974710928191994e-06,
"loss": 1.505861520767212,
"step": 216
},
{
"epoch": 0.24971363115693013,
"grad_norm": 0.10934654623270035,
"learning_rate": 9.973493548150705e-06,
"loss": 1.2586157321929932,
"step": 218
},
{
"epoch": 0.2520045819014891,
"grad_norm": 0.16258153319358826,
"learning_rate": 9.972247637859942e-06,
"loss": 1.5800942182540894,
"step": 220
},
{
"epoch": 0.2542955326460481,
"grad_norm": 0.04441208764910698,
"learning_rate": 9.970973205265654e-06,
"loss": 1.2052358388900757,
"step": 222
},
{
"epoch": 0.2565864833906071,
"grad_norm": 0.04366284981369972,
"learning_rate": 9.969670258495689e-06,
"loss": 1.4153897762298584,
"step": 224
},
{
"epoch": 0.2588774341351661,
"grad_norm": 0.10389053076505661,
"learning_rate": 9.968338805859746e-06,
"loss": 1.1192145347595215,
"step": 226
},
{
"epoch": 0.2611683848797251,
"grad_norm": 0.03827785328030586,
"learning_rate": 9.966978855849328e-06,
"loss": 1.233184814453125,
"step": 228
},
{
"epoch": 0.2634593356242841,
"grad_norm": 0.03582112118601799,
"learning_rate": 9.965590417137683e-06,
"loss": 1.444356918334961,
"step": 230
},
{
"epoch": 0.26575028636884307,
"grad_norm": 0.16782893240451813,
"learning_rate": 9.964173498579744e-06,
"loss": 1.3132915496826172,
"step": 232
},
{
"epoch": 0.26804123711340205,
"grad_norm": 0.10886494070291519,
"learning_rate": 9.962728109212087e-06,
"loss": 1.603651762008667,
"step": 234
},
{
"epoch": 0.27033218785796104,
"grad_norm": 0.32741063833236694,
"learning_rate": 9.961254258252853e-06,
"loss": 1.722959041595459,
"step": 236
},
{
"epoch": 0.27262313860252,
"grad_norm": 0.038513097912073135,
"learning_rate": 9.95975195510171e-06,
"loss": 1.270897626876831,
"step": 238
},
{
"epoch": 0.27491408934707906,
"grad_norm": 0.11721174418926239,
"learning_rate": 9.958221209339776e-06,
"loss": 1.3808534145355225,
"step": 240
},
{
"epoch": 0.27720504009163804,
"grad_norm": 0.11607187986373901,
"learning_rate": 9.956662030729571e-06,
"loss": 1.4212572574615479,
"step": 242
},
{
"epoch": 0.27949599083619703,
"grad_norm": 0.05183471366763115,
"learning_rate": 9.955074429214945e-06,
"loss": 1.2837097644805908,
"step": 244
},
{
"epoch": 0.281786941580756,
"grad_norm": 0.048340827226638794,
"learning_rate": 9.95345841492102e-06,
"loss": 1.3732128143310547,
"step": 246
},
{
"epoch": 0.284077892325315,
"grad_norm": 0.03368377685546875,
"learning_rate": 9.951813998154122e-06,
"loss": 1.150505542755127,
"step": 248
},
{
"epoch": 0.286368843069874,
"grad_norm": 0.16787323355674744,
"learning_rate": 9.950141189401722e-06,
"loss": 1.4632198810577393,
"step": 250
},
{
"epoch": 0.28865979381443296,
"grad_norm": 0.14481912553310394,
"learning_rate": 9.948439999332362e-06,
"loss": 1.4926843643188477,
"step": 252
},
{
"epoch": 0.290950744558992,
"grad_norm": 0.15263405442237854,
"learning_rate": 9.946710438795586e-06,
"loss": 1.4954164028167725,
"step": 254
},
{
"epoch": 0.293241695303551,
"grad_norm": 0.11114485561847687,
"learning_rate": 9.944952518821877e-06,
"loss": 0.9889236688613892,
"step": 256
},
{
"epoch": 0.29553264604810997,
"grad_norm": 0.2147989124059677,
"learning_rate": 9.943166250622585e-06,
"loss": 1.2724560499191284,
"step": 258
},
{
"epoch": 0.29782359679266895,
"grad_norm": 0.21461904048919678,
"learning_rate": 9.941351645589853e-06,
"loss": 0.9281865358352661,
"step": 260
},
{
"epoch": 0.30011454753722794,
"grad_norm": 0.03682840242981911,
"learning_rate": 9.939508715296543e-06,
"loss": 1.3381445407867432,
"step": 262
},
{
"epoch": 0.3024054982817869,
"grad_norm": 0.2321019023656845,
"learning_rate": 9.93763747149617e-06,
"loss": 1.2490191459655762,
"step": 264
},
{
"epoch": 0.30469644902634596,
"grad_norm": 0.11454650014638901,
"learning_rate": 9.935737926122816e-06,
"loss": 1.2711057662963867,
"step": 266
},
{
"epoch": 0.30698739977090495,
"grad_norm": 0.10006029158830643,
"learning_rate": 9.933810091291065e-06,
"loss": 1.2933646440505981,
"step": 268
},
{
"epoch": 0.30927835051546393,
"grad_norm": 0.1007399782538414,
"learning_rate": 9.93185397929592e-06,
"loss": 1.2037980556488037,
"step": 270
},
{
"epoch": 0.3115693012600229,
"grad_norm": 0.03689350187778473,
"learning_rate": 9.929869602612718e-06,
"loss": 1.1460704803466797,
"step": 272
},
{
"epoch": 0.3138602520045819,
"grad_norm": 0.14365845918655396,
"learning_rate": 9.927856973897068e-06,
"loss": 0.5115751624107361,
"step": 274
},
{
"epoch": 0.3161512027491409,
"grad_norm": 0.15885332226753235,
"learning_rate": 9.925816105984751e-06,
"loss": 1.3992141485214233,
"step": 276
},
{
"epoch": 0.31844215349369986,
"grad_norm": 0.2972363829612732,
"learning_rate": 9.923747011891653e-06,
"loss": 0.6298142671585083,
"step": 278
},
{
"epoch": 0.3207331042382589,
"grad_norm": 0.10289888083934784,
"learning_rate": 9.92164970481367e-06,
"loss": 1.3492999076843262,
"step": 280
},
{
"epoch": 0.3230240549828179,
"grad_norm": 0.30722421407699585,
"learning_rate": 9.919524198126637e-06,
"loss": 1.1495556831359863,
"step": 282
},
{
"epoch": 0.32531500572737687,
"grad_norm": 0.1234731450676918,
"learning_rate": 9.91737050538623e-06,
"loss": 1.625375747680664,
"step": 284
},
{
"epoch": 0.32760595647193586,
"grad_norm": 0.11733370274305344,
"learning_rate": 9.915188640327887e-06,
"loss": 0.7891396284103394,
"step": 286
},
{
"epoch": 0.32989690721649484,
"grad_norm": 0.11923382431268692,
"learning_rate": 9.912978616866716e-06,
"loss": 1.469787836074829,
"step": 288
},
{
"epoch": 0.3321878579610538,
"grad_norm": 0.17214825749397278,
"learning_rate": 9.910740449097412e-06,
"loss": 1.4425544738769531,
"step": 290
},
{
"epoch": 0.3344788087056128,
"grad_norm": 0.04059191793203354,
"learning_rate": 9.908474151294161e-06,
"loss": 1.1002707481384277,
"step": 292
},
{
"epoch": 0.33676975945017185,
"grad_norm": 0.03392655774950981,
"learning_rate": 9.906179737910554e-06,
"loss": 1.1838493347167969,
"step": 294
},
{
"epoch": 0.33906071019473083,
"grad_norm": 0.032716911286115646,
"learning_rate": 9.903857223579496e-06,
"loss": 1.0073325634002686,
"step": 296
},
{
"epoch": 0.3413516609392898,
"grad_norm": 0.030965758487582207,
"learning_rate": 9.901506623113098e-06,
"loss": 0.969083845615387,
"step": 298
},
{
"epoch": 0.3436426116838488,
"grad_norm": 0.12395308911800385,
"learning_rate": 9.899127951502601e-06,
"loss": 1.2470171451568604,
"step": 300
},
{
"epoch": 0.3459335624284078,
"grad_norm": 0.3706231713294983,
"learning_rate": 9.896721223918276e-06,
"loss": 1.5252048969268799,
"step": 302
},
{
"epoch": 0.34822451317296677,
"grad_norm": 0.044303592294454575,
"learning_rate": 9.89428645570932e-06,
"loss": 1.1785590648651123,
"step": 304
},
{
"epoch": 0.35051546391752575,
"grad_norm": 0.22373101115226746,
"learning_rate": 9.891823662403763e-06,
"loss": 1.0534144639968872,
"step": 306
},
{
"epoch": 0.3528064146620848,
"grad_norm": 0.0998837873339653,
"learning_rate": 9.88933285970837e-06,
"loss": 1.358349084854126,
"step": 308
},
{
"epoch": 0.3550973654066438,
"grad_norm": 0.10202368348836899,
"learning_rate": 9.886814063508536e-06,
"loss": 1.2032561302185059,
"step": 310
},
{
"epoch": 0.35738831615120276,
"grad_norm": 0.12603235244750977,
"learning_rate": 9.884267289868194e-06,
"loss": 1.270133376121521,
"step": 312
},
{
"epoch": 0.35967926689576174,
"grad_norm": 0.13588766753673553,
"learning_rate": 9.8816925550297e-06,
"loss": 1.3731555938720703,
"step": 314
},
{
"epoch": 0.3619702176403207,
"grad_norm": 0.2927129566669464,
"learning_rate": 9.879089875413736e-06,
"loss": 1.2470248937606812,
"step": 316
},
{
"epoch": 0.3642611683848797,
"grad_norm": 0.03428524360060692,
"learning_rate": 9.876459267619215e-06,
"loss": 1.2745685577392578,
"step": 318
},
{
"epoch": 0.3665521191294387,
"grad_norm": 0.3623892068862915,
"learning_rate": 9.873800748423152e-06,
"loss": 1.3339478969573975,
"step": 320
},
{
"epoch": 0.36884306987399773,
"grad_norm": 0.11408170312643051,
"learning_rate": 9.871114334780583e-06,
"loss": 1.3286080360412598,
"step": 322
},
{
"epoch": 0.3711340206185567,
"grad_norm": 0.10345927625894547,
"learning_rate": 9.868400043824431e-06,
"loss": 1.332814335823059,
"step": 324
},
{
"epoch": 0.3734249713631157,
"grad_norm": 0.451455682516098,
"learning_rate": 9.86565789286542e-06,
"loss": 1.435546875,
"step": 326
},
{
"epoch": 0.3757159221076747,
"grad_norm": 0.03447026014328003,
"learning_rate": 9.862887899391953e-06,
"loss": 1.2484092712402344,
"step": 328
},
{
"epoch": 0.37800687285223367,
"grad_norm": 0.10319351404905319,
"learning_rate": 9.860090081069998e-06,
"loss": 1.1266441345214844,
"step": 330
},
{
"epoch": 0.38029782359679265,
"grad_norm": 0.10190501809120178,
"learning_rate": 9.857264455742983e-06,
"loss": 1.2466825246810913,
"step": 332
},
{
"epoch": 0.38258877434135163,
"grad_norm": 0.04389253631234169,
"learning_rate": 9.854411041431678e-06,
"loss": 1.1545830965042114,
"step": 334
},
{
"epoch": 0.3848797250859107,
"grad_norm": 0.1142214685678482,
"learning_rate": 9.851529856334079e-06,
"loss": 1.2927944660186768,
"step": 336
},
{
"epoch": 0.38717067583046966,
"grad_norm": 0.03781864792108536,
"learning_rate": 9.848620918825294e-06,
"loss": 1.2539844512939453,
"step": 338
},
{
"epoch": 0.38946162657502864,
"grad_norm": 0.12804462015628815,
"learning_rate": 9.845684247457425e-06,
"loss": 1.3808146715164185,
"step": 340
},
{
"epoch": 0.3917525773195876,
"grad_norm": 0.037370480597019196,
"learning_rate": 9.842719860959455e-06,
"loss": 1.2456262111663818,
"step": 342
},
{
"epoch": 0.3940435280641466,
"grad_norm": 0.042404066771268845,
"learning_rate": 9.839727778237116e-06,
"loss": 1.3200464248657227,
"step": 344
},
{
"epoch": 0.3963344788087056,
"grad_norm": 0.04442799836397171,
"learning_rate": 9.836708018372782e-06,
"loss": 1.3215160369873047,
"step": 346
},
{
"epoch": 0.39862542955326463,
"grad_norm": 0.08433306217193604,
"learning_rate": 9.833660600625338e-06,
"loss": 1.2241692543029785,
"step": 348
},
{
"epoch": 0.4009163802978236,
"grad_norm": 0.14272065460681915,
"learning_rate": 9.83058554443006e-06,
"loss": 1.2168128490447998,
"step": 350
},
{
"epoch": 0.4032073310423826,
"grad_norm": 0.10155618190765381,
"learning_rate": 9.827482869398496e-06,
"loss": 1.3269870281219482,
"step": 352
},
{
"epoch": 0.4054982817869416,
"grad_norm": 0.345684289932251,
"learning_rate": 9.82435259531833e-06,
"loss": 1.048119068145752,
"step": 354
},
{
"epoch": 0.40778923253150057,
"grad_norm": 0.03348676860332489,
"learning_rate": 9.82119474215327e-06,
"loss": 0.8612366914749146,
"step": 356
},
{
"epoch": 0.41008018327605955,
"grad_norm": 0.10538852959871292,
"learning_rate": 9.818009330042906e-06,
"loss": 1.4399499893188477,
"step": 358
},
{
"epoch": 0.41237113402061853,
"grad_norm": 0.12840358912944794,
"learning_rate": 9.814796379302592e-06,
"loss": 1.3075119256973267,
"step": 360
},
{
"epoch": 0.4146620847651776,
"grad_norm": 0.03234279155731201,
"learning_rate": 9.811555910423312e-06,
"loss": 1.195176601409912,
"step": 362
},
{
"epoch": 0.41695303550973656,
"grad_norm": 0.2873457968235016,
"learning_rate": 9.808287944071552e-06,
"loss": 1.3421657085418701,
"step": 364
},
{
"epoch": 0.41924398625429554,
"grad_norm": 0.03667648881673813,
"learning_rate": 9.804992501089164e-06,
"loss": 1.4639015197753906,
"step": 366
},
{
"epoch": 0.4215349369988545,
"grad_norm": 0.1039833277463913,
"learning_rate": 9.801669602493236e-06,
"loss": 1.2831900119781494,
"step": 368
},
{
"epoch": 0.4238258877434135,
"grad_norm": 0.1346283257007599,
"learning_rate": 9.798319269475959e-06,
"loss": 1.1615617275238037,
"step": 370
},
{
"epoch": 0.4261168384879725,
"grad_norm": 0.11276187002658844,
"learning_rate": 9.794941523404491e-06,
"loss": 1.3308217525482178,
"step": 372
},
{
"epoch": 0.4284077892325315,
"grad_norm": 0.10832475125789642,
"learning_rate": 9.791536385820815e-06,
"loss": 1.3403054475784302,
"step": 374
},
{
"epoch": 0.4306987399770905,
"grad_norm": 0.07494664937257767,
"learning_rate": 9.788103878441614e-06,
"loss": 0.9897347688674927,
"step": 376
},
{
"epoch": 0.4329896907216495,
"grad_norm": 0.13242274522781372,
"learning_rate": 9.784644023158118e-06,
"loss": 1.2988442182540894,
"step": 378
},
{
"epoch": 0.4352806414662085,
"grad_norm": 0.04049122706055641,
"learning_rate": 9.781156842035978e-06,
"loss": 1.2798575162887573,
"step": 380
},
{
"epoch": 0.43757159221076747,
"grad_norm": 0.10166372358798981,
"learning_rate": 9.777642357315115e-06,
"loss": 1.4548561573028564,
"step": 382
},
{
"epoch": 0.43986254295532645,
"grad_norm": 0.09827661514282227,
"learning_rate": 9.774100591409583e-06,
"loss": 0.8798496127128601,
"step": 384
},
{
"epoch": 0.44215349369988544,
"grad_norm": 0.039044998586177826,
"learning_rate": 9.770531566907424e-06,
"loss": 1.1994266510009766,
"step": 386
},
{
"epoch": 0.4444444444444444,
"grad_norm": 0.03467918932437897,
"learning_rate": 9.766935306570528e-06,
"loss": 1.0867178440093994,
"step": 388
},
{
"epoch": 0.44673539518900346,
"grad_norm": 0.09605854749679565,
"learning_rate": 9.763311833334482e-06,
"loss": 1.2179234027862549,
"step": 390
},
{
"epoch": 0.44902634593356244,
"grad_norm": 0.04672601819038391,
"learning_rate": 9.759661170308426e-06,
"loss": 1.0760804414749146,
"step": 392
},
{
"epoch": 0.4513172966781214,
"grad_norm": 0.14849010109901428,
"learning_rate": 9.75598334077491e-06,
"loss": 1.4439059495925903,
"step": 394
},
{
"epoch": 0.4536082474226804,
"grad_norm": 0.10495574027299881,
"learning_rate": 9.752278368189738e-06,
"loss": 1.2911549806594849,
"step": 396
},
{
"epoch": 0.4558991981672394,
"grad_norm": 0.10133286565542221,
"learning_rate": 9.748546276181824e-06,
"loss": 1.3363006114959717,
"step": 398
},
{
"epoch": 0.4581901489117984,
"grad_norm": 0.10403479635715485,
"learning_rate": 9.74478708855304e-06,
"loss": 1.1435672044754028,
"step": 400
},
{
"epoch": 0.46048109965635736,
"grad_norm": 0.23274336755275726,
"learning_rate": 9.74100082927806e-06,
"loss": 1.1191637516021729,
"step": 402
},
{
"epoch": 0.4627720504009164,
"grad_norm": 0.035993631929159164,
"learning_rate": 9.737187522504215e-06,
"loss": 1.2965900897979736,
"step": 404
},
{
"epoch": 0.4650630011454754,
"grad_norm": 0.11053793877363205,
"learning_rate": 9.733347192551333e-06,
"loss": 1.3513212203979492,
"step": 406
},
{
"epoch": 0.46735395189003437,
"grad_norm": 0.38711977005004883,
"learning_rate": 9.729479863911585e-06,
"loss": 0.9576661586761475,
"step": 408
},
{
"epoch": 0.46964490263459335,
"grad_norm": 0.10174008458852768,
"learning_rate": 9.725585561249331e-06,
"loss": 1.3105225563049316,
"step": 410
},
{
"epoch": 0.47193585337915234,
"grad_norm": 0.5640531182289124,
"learning_rate": 9.72166430940096e-06,
"loss": 1.5542221069335938,
"step": 412
},
{
"epoch": 0.4742268041237113,
"grad_norm": 0.043333012610673904,
"learning_rate": 9.71771613337473e-06,
"loss": 1.212013602256775,
"step": 414
},
{
"epoch": 0.4765177548682703,
"grad_norm": 0.12072140723466873,
"learning_rate": 9.713741058350618e-06,
"loss": 1.3732361793518066,
"step": 416
},
{
"epoch": 0.47880870561282934,
"grad_norm": 0.039414238184690475,
"learning_rate": 9.709739109680146e-06,
"loss": 1.0561060905456543,
"step": 418
},
{
"epoch": 0.48109965635738833,
"grad_norm": 0.11718718707561493,
"learning_rate": 9.70571031288623e-06,
"loss": 1.4380638599395752,
"step": 420
},
{
"epoch": 0.4833906071019473,
"grad_norm": 0.03386823460459709,
"learning_rate": 9.701654693663012e-06,
"loss": 1.223938226699829,
"step": 422
},
{
"epoch": 0.4856815578465063,
"grad_norm": 0.08439936488866806,
"learning_rate": 9.697572277875696e-06,
"loss": 1.2271449565887451,
"step": 424
},
{
"epoch": 0.4879725085910653,
"grad_norm": 0.03942372649908066,
"learning_rate": 9.693463091560387e-06,
"loss": 1.0300081968307495,
"step": 426
},
{
"epoch": 0.49026345933562426,
"grad_norm": 0.2659618556499481,
"learning_rate": 9.689327160923918e-06,
"loss": 1.275801420211792,
"step": 428
},
{
"epoch": 0.4925544100801833,
"grad_norm": 0.11298934370279312,
"learning_rate": 9.685164512343694e-06,
"loss": 1.2419630289077759,
"step": 430
},
{
"epoch": 0.4948453608247423,
"grad_norm": 0.04946233332157135,
"learning_rate": 9.680975172367508e-06,
"loss": 1.4032783508300781,
"step": 432
},
{
"epoch": 0.49713631156930127,
"grad_norm": 0.0772048756480217,
"learning_rate": 9.67675916771339e-06,
"loss": 1.1696724891662598,
"step": 434
},
{
"epoch": 0.49942726231386025,
"grad_norm": 0.13884860277175903,
"learning_rate": 9.67251652526942e-06,
"loss": 1.1072807312011719,
"step": 436
},
{
"epoch": 0.5017182130584192,
"grad_norm": 0.1219138652086258,
"learning_rate": 9.668247272093568e-06,
"loss": 1.2660953998565674,
"step": 438
},
{
"epoch": 0.5040091638029782,
"grad_norm": 0.11671170592308044,
"learning_rate": 9.663951435413512e-06,
"loss": 1.4654450416564941,
"step": 440
},
{
"epoch": 0.5063001145475372,
"grad_norm": 0.10897082090377808,
"learning_rate": 9.659629042626478e-06,
"loss": 1.2332985401153564,
"step": 442
},
{
"epoch": 0.5085910652920962,
"grad_norm": 0.2291656881570816,
"learning_rate": 9.655280121299049e-06,
"loss": 1.2597649097442627,
"step": 444
},
{
"epoch": 0.5108820160366552,
"grad_norm": 0.18447436392307281,
"learning_rate": 9.650904699167002e-06,
"loss": 1.2398802042007446,
"step": 446
},
{
"epoch": 0.5131729667812142,
"grad_norm": 0.09861113131046295,
"learning_rate": 9.646502804135125e-06,
"loss": 1.157179832458496,
"step": 448
},
{
"epoch": 0.5154639175257731,
"grad_norm": 0.04085477441549301,
"learning_rate": 9.642074464277035e-06,
"loss": 1.0565853118896484,
"step": 450
},
{
"epoch": 0.5177548682703322,
"grad_norm": 0.03253213316202164,
"learning_rate": 9.637619707835011e-06,
"loss": 1.115795373916626,
"step": 452
},
{
"epoch": 0.5200458190148912,
"grad_norm": 0.10820797830820084,
"learning_rate": 9.633138563219805e-06,
"loss": 1.2870464324951172,
"step": 454
},
{
"epoch": 0.5223367697594502,
"grad_norm": 0.08356233686208725,
"learning_rate": 9.628631059010459e-06,
"loss": 1.161600112915039,
"step": 456
},
{
"epoch": 0.5246277205040092,
"grad_norm": 0.10461345314979553,
"learning_rate": 9.624097223954132e-06,
"loss": 1.2577989101409912,
"step": 458
},
{
"epoch": 0.5269186712485682,
"grad_norm": 0.2117365151643753,
"learning_rate": 9.619537086965909e-06,
"loss": 0.9334989190101624,
"step": 460
},
{
"epoch": 0.5292096219931272,
"grad_norm": 0.03358057513833046,
"learning_rate": 9.614950677128618e-06,
"loss": 1.0164105892181396,
"step": 462
},
{
"epoch": 0.5315005727376861,
"grad_norm": 0.04315713420510292,
"learning_rate": 9.610338023692644e-06,
"loss": 1.2515099048614502,
"step": 464
},
{
"epoch": 0.5337915234822451,
"grad_norm": 0.11090853065252304,
"learning_rate": 9.60569915607575e-06,
"loss": 0.9969978928565979,
"step": 466
},
{
"epoch": 0.5360824742268041,
"grad_norm": 0.06928592920303345,
"learning_rate": 9.601034103862875e-06,
"loss": 1.1942644119262695,
"step": 468
},
{
"epoch": 0.5383734249713631,
"grad_norm": 0.11059997975826263,
"learning_rate": 9.596342896805958e-06,
"loss": 1.0854039192199707,
"step": 470
},
{
"epoch": 0.5406643757159221,
"grad_norm": 0.2203303724527359,
"learning_rate": 9.591625564823743e-06,
"loss": 0.6206300258636475,
"step": 472
},
{
"epoch": 0.5429553264604811,
"grad_norm": 0.03575604781508446,
"learning_rate": 9.58688213800159e-06,
"loss": 1.1505436897277832,
"step": 474
},
{
"epoch": 0.54524627720504,
"grad_norm": 0.11237749457359314,
"learning_rate": 9.58211264659128e-06,
"loss": 1.2954638004302979,
"step": 476
},
{
"epoch": 0.5475372279495991,
"grad_norm": 0.03931301459670067,
"learning_rate": 9.577317121010822e-06,
"loss": 1.1090443134307861,
"step": 478
},
{
"epoch": 0.5498281786941581,
"grad_norm": 0.10506019741296768,
"learning_rate": 9.572495591844268e-06,
"loss": 1.2766085863113403,
"step": 480
},
{
"epoch": 0.5521191294387171,
"grad_norm": 0.03373529762029648,
"learning_rate": 9.567648089841504e-06,
"loss": 1.1036125421524048,
"step": 482
},
{
"epoch": 0.5544100801832761,
"grad_norm": 0.03757813945412636,
"learning_rate": 9.562774645918067e-06,
"loss": 1.1739498376846313,
"step": 484
},
{
"epoch": 0.5567010309278351,
"grad_norm": 0.033927951008081436,
"learning_rate": 9.557875291154937e-06,
"loss": 1.0839054584503174,
"step": 486
},
{
"epoch": 0.5589919816723941,
"grad_norm": 0.11530556529760361,
"learning_rate": 9.552950056798345e-06,
"loss": 1.401555061340332,
"step": 488
},
{
"epoch": 0.561282932416953,
"grad_norm": 0.1037631705403328,
"learning_rate": 9.547998974259573e-06,
"loss": 1.266376256942749,
"step": 490
},
{
"epoch": 0.563573883161512,
"grad_norm": 0.03408586606383324,
"learning_rate": 9.543022075114751e-06,
"loss": 1.1710147857666016,
"step": 492
},
{
"epoch": 0.565864833906071,
"grad_norm": 0.09721101075410843,
"learning_rate": 9.538019391104659e-06,
"loss": 1.1715192794799805,
"step": 494
},
{
"epoch": 0.56815578465063,
"grad_norm": 0.13440069556236267,
"learning_rate": 9.532990954134527e-06,
"loss": 1.2787623405456543,
"step": 496
},
{
"epoch": 0.570446735395189,
"grad_norm": 0.14032506942749023,
"learning_rate": 9.527936796273818e-06,
"loss": 1.352327823638916,
"step": 498
},
{
"epoch": 0.572737686139748,
"grad_norm": 0.03532100096344948,
"learning_rate": 9.522856949756042e-06,
"loss": 1.396005630493164,
"step": 500
},
{
"epoch": 0.5750286368843069,
"grad_norm": 0.03295751288533211,
"learning_rate": 9.517751446978537e-06,
"loss": 1.2074257135391235,
"step": 502
},
{
"epoch": 0.5773195876288659,
"grad_norm": 0.03744492679834366,
"learning_rate": 9.51262032050227e-06,
"loss": 1.1247986555099487,
"step": 504
},
{
"epoch": 0.579610538373425,
"grad_norm": 0.10635876655578613,
"learning_rate": 9.507463603051624e-06,
"loss": 1.279697060585022,
"step": 506
},
{
"epoch": 0.581901489117984,
"grad_norm": 0.1893984079360962,
"learning_rate": 9.502281327514192e-06,
"loss": 1.3793418407440186,
"step": 508
},
{
"epoch": 0.584192439862543,
"grad_norm": 0.03022020496428013,
"learning_rate": 9.497073526940564e-06,
"loss": 1.1290874481201172,
"step": 510
},
{
"epoch": 0.586483390607102,
"grad_norm": 0.08362241834402084,
"learning_rate": 9.491840234544127e-06,
"loss": 1.1272192001342773,
"step": 512
},
{
"epoch": 0.588774341351661,
"grad_norm": 0.04655982553958893,
"learning_rate": 9.486581483700836e-06,
"loss": 1.0976135730743408,
"step": 514
},
{
"epoch": 0.5910652920962199,
"grad_norm": 0.09682846069335938,
"learning_rate": 9.481297307949016e-06,
"loss": 1.361159324645996,
"step": 516
},
{
"epoch": 0.5933562428407789,
"grad_norm": 0.10151924192905426,
"learning_rate": 9.47598774098914e-06,
"loss": 1.1374366283416748,
"step": 518
},
{
"epoch": 0.5956471935853379,
"grad_norm": 0.3335345685482025,
"learning_rate": 9.470652816683619e-06,
"loss": 1.183037519454956,
"step": 520
},
{
"epoch": 0.5979381443298969,
"grad_norm": 0.11204201728105545,
"learning_rate": 9.46529256905658e-06,
"loss": 1.3378653526306152,
"step": 522
},
{
"epoch": 0.6002290950744559,
"grad_norm": 0.03878810256719589,
"learning_rate": 9.459907032293654e-06,
"loss": 1.2830768823623657,
"step": 524
},
{
"epoch": 0.6025200458190149,
"grad_norm": 0.03078841231763363,
"learning_rate": 9.454496240741761e-06,
"loss": 0.988889753818512,
"step": 526
},
{
"epoch": 0.6048109965635738,
"grad_norm": 0.04741325229406357,
"learning_rate": 9.44906022890888e-06,
"loss": 1.2009971141815186,
"step": 528
},
{
"epoch": 0.6071019473081328,
"grad_norm": 0.09710617363452911,
"learning_rate": 9.443599031463838e-06,
"loss": 1.4264302253723145,
"step": 530
},
{
"epoch": 0.6093928980526919,
"grad_norm": 0.12086698412895203,
"learning_rate": 9.438112683236086e-06,
"loss": 0.7716737985610962,
"step": 532
},
{
"epoch": 0.6116838487972509,
"grad_norm": 0.028488747775554657,
"learning_rate": 9.432601219215479e-06,
"loss": 1.1653621196746826,
"step": 534
},
{
"epoch": 0.6139747995418099,
"grad_norm": 0.030125625431537628,
"learning_rate": 9.427064674552046e-06,
"loss": 0.9844313859939575,
"step": 536
},
{
"epoch": 0.6162657502863689,
"grad_norm": 1.246670126914978,
"learning_rate": 9.421503084555778e-06,
"loss": 1.1310303211212158,
"step": 538
},
{
"epoch": 0.6185567010309279,
"grad_norm": 0.047299824655056,
"learning_rate": 9.41591648469639e-06,
"loss": 1.1721080541610718,
"step": 540
},
{
"epoch": 0.6208476517754868,
"grad_norm": 0.07911659777164459,
"learning_rate": 9.410304910603105e-06,
"loss": 0.9692875146865845,
"step": 542
},
{
"epoch": 0.6231386025200458,
"grad_norm": 0.041665468364953995,
"learning_rate": 9.404668398064415e-06,
"loss": 1.225880742073059,
"step": 544
},
{
"epoch": 0.6254295532646048,
"grad_norm": 0.09727603197097778,
"learning_rate": 9.399006983027869e-06,
"loss": 1.2111965417861938,
"step": 546
},
{
"epoch": 0.6277205040091638,
"grad_norm": 0.2791145443916321,
"learning_rate": 9.393320701599826e-06,
"loss": 1.3773910999298096,
"step": 548
},
{
"epoch": 0.6300114547537228,
"grad_norm": 0.2922375202178955,
"learning_rate": 9.387609590045243e-06,
"loss": 1.5269827842712402,
"step": 550
},
{
"epoch": 0.6323024054982818,
"grad_norm": 0.034879785031080246,
"learning_rate": 9.381873684787424e-06,
"loss": 1.096055269241333,
"step": 552
},
{
"epoch": 0.6345933562428407,
"grad_norm": 0.21075953543186188,
"learning_rate": 9.376113022407806e-06,
"loss": 1.3973267078399658,
"step": 554
},
{
"epoch": 0.6368843069873997,
"grad_norm": 0.04559031128883362,
"learning_rate": 9.370327639645715e-06,
"loss": 1.3103339672088623,
"step": 556
},
{
"epoch": 0.6391752577319587,
"grad_norm": 0.09411913901567459,
"learning_rate": 9.364517573398128e-06,
"loss": 1.1426537036895752,
"step": 558
},
{
"epoch": 0.6414662084765178,
"grad_norm": 0.11268170177936554,
"learning_rate": 9.358682860719456e-06,
"loss": 1.2462685108184814,
"step": 560
},
{
"epoch": 0.6437571592210768,
"grad_norm": 0.15371939539909363,
"learning_rate": 9.352823538821286e-06,
"loss": 1.350957989692688,
"step": 562
},
{
"epoch": 0.6460481099656358,
"grad_norm": 0.10083743184804916,
"learning_rate": 9.346939645072158e-06,
"loss": 1.2344059944152832,
"step": 564
},
{
"epoch": 0.6483390607101948,
"grad_norm": 0.12374784797430038,
"learning_rate": 9.341031216997318e-06,
"loss": 1.332352876663208,
"step": 566
},
{
"epoch": 0.6506300114547537,
"grad_norm": 0.09682541340589523,
"learning_rate": 9.335098292278487e-06,
"loss": 1.0717089176177979,
"step": 568
},
{
"epoch": 0.6529209621993127,
"grad_norm": 0.04859980568289757,
"learning_rate": 9.329140908753612e-06,
"loss": 1.2429944276809692,
"step": 570
},
{
"epoch": 0.6552119129438717,
"grad_norm": 0.03579086810350418,
"learning_rate": 9.323159104416637e-06,
"loss": 1.4028278589248657,
"step": 572
},
{
"epoch": 0.6575028636884307,
"grad_norm": 0.11958760768175125,
"learning_rate": 9.31715291741724e-06,
"loss": 1.2857556343078613,
"step": 574
},
{
"epoch": 0.6597938144329897,
"grad_norm": 0.03217066451907158,
"learning_rate": 9.311122386060612e-06,
"loss": 1.1168699264526367,
"step": 576
},
{
"epoch": 0.6620847651775487,
"grad_norm": 0.12310899794101715,
"learning_rate": 9.305067548807202e-06,
"loss": 1.1735236644744873,
"step": 578
},
{
"epoch": 0.6643757159221076,
"grad_norm": 0.11337030678987503,
"learning_rate": 9.29898844427247e-06,
"loss": 1.301731824874878,
"step": 580
},
{
"epoch": 0.6666666666666666,
"grad_norm": 0.03304918482899666,
"learning_rate": 9.292885111226647e-06,
"loss": 1.1898317337036133,
"step": 582
},
{
"epoch": 0.6689576174112256,
"grad_norm": 0.035968322306871414,
"learning_rate": 9.286757588594479e-06,
"loss": 1.3450303077697754,
"step": 584
},
{
"epoch": 0.6712485681557846,
"grad_norm": 0.232716366648674,
"learning_rate": 9.28060591545499e-06,
"loss": 0.5568069219589233,
"step": 586
},
{
"epoch": 0.6735395189003437,
"grad_norm": 0.18080194294452667,
"learning_rate": 9.274430131041224e-06,
"loss": 1.2939491271972656,
"step": 588
},
{
"epoch": 0.6758304696449027,
"grad_norm": 0.1996665745973587,
"learning_rate": 9.268230274739993e-06,
"loss": 1.680454134941101,
"step": 590
},
{
"epoch": 0.6781214203894617,
"grad_norm": 0.10235057026147842,
"learning_rate": 9.262006386091643e-06,
"loss": 1.300153136253357,
"step": 592
},
{
"epoch": 0.6804123711340206,
"grad_norm": 0.1189015805721283,
"learning_rate": 9.255758504789773e-06,
"loss": 1.4123055934906006,
"step": 594
},
{
"epoch": 0.6827033218785796,
"grad_norm": 0.11398884654045105,
"learning_rate": 9.249486670681011e-06,
"loss": 1.3474668264389038,
"step": 596
},
{
"epoch": 0.6849942726231386,
"grad_norm": 0.33024370670318604,
"learning_rate": 9.243190923764743e-06,
"loss": 0.9317454695701599,
"step": 598
},
{
"epoch": 0.6872852233676976,
"grad_norm": 0.18342778086662292,
"learning_rate": 9.236871304192857e-06,
"loss": 1.3693324327468872,
"step": 600
},
{
"epoch": 0.6895761741122566,
"grad_norm": 0.09077564626932144,
"learning_rate": 9.2305278522695e-06,
"loss": 1.2000863552093506,
"step": 602
},
{
"epoch": 0.6918671248568156,
"grad_norm": 0.1056990921497345,
"learning_rate": 9.224160608450806e-06,
"loss": 1.2265493869781494,
"step": 604
},
{
"epoch": 0.6941580756013745,
"grad_norm": 0.039485201239585876,
"learning_rate": 9.217769613344647e-06,
"loss": 1.1047804355621338,
"step": 606
},
{
"epoch": 0.6964490263459335,
"grad_norm": 0.08278033137321472,
"learning_rate": 9.211354907710373e-06,
"loss": 1.2274484634399414,
"step": 608
},
{
"epoch": 0.6987399770904925,
"grad_norm": 0.04109884425997734,
"learning_rate": 9.204916532458552e-06,
"loss": 1.237788438796997,
"step": 610
},
{
"epoch": 0.7010309278350515,
"grad_norm": 0.11578814685344696,
"learning_rate": 9.198454528650702e-06,
"loss": 1.2939280271530151,
"step": 612
},
{
"epoch": 0.7033218785796106,
"grad_norm": 0.10241305083036423,
"learning_rate": 9.191968937499041e-06,
"loss": 1.1080031394958496,
"step": 614
},
{
"epoch": 0.7056128293241696,
"grad_norm": 0.08389988541603088,
"learning_rate": 9.185459800366212e-06,
"loss": 1.1695685386657715,
"step": 616
},
{
"epoch": 0.7079037800687286,
"grad_norm": 0.18472033739089966,
"learning_rate": 9.178927158765037e-06,
"loss": 1.1244120597839355,
"step": 618
},
{
"epoch": 0.7101947308132875,
"grad_norm": 0.13071289658546448,
"learning_rate": 9.172371054358224e-06,
"loss": 1.2239925861358643,
"step": 620
},
{
"epoch": 0.7124856815578465,
"grad_norm": 0.039994560182094574,
"learning_rate": 9.16579152895813e-06,
"loss": 1.1075098514556885,
"step": 622
},
{
"epoch": 0.7147766323024055,
"grad_norm": 0.12017904967069626,
"learning_rate": 9.15918862452648e-06,
"loss": 1.0645837783813477,
"step": 624
},
{
"epoch": 0.7170675830469645,
"grad_norm": 0.19609086215496063,
"learning_rate": 9.152562383174102e-06,
"loss": 0.6706198453903198,
"step": 626
},
{
"epoch": 0.7193585337915235,
"grad_norm": 0.13229237496852875,
"learning_rate": 9.145912847160652e-06,
"loss": 1.3926395177841187,
"step": 628
},
{
"epoch": 0.7216494845360825,
"grad_norm": 0.09191222488880157,
"learning_rate": 9.139240058894358e-06,
"loss": 1.2411160469055176,
"step": 630
},
{
"epoch": 0.7239404352806414,
"grad_norm": 0.13508380949497223,
"learning_rate": 9.132544060931738e-06,
"loss": 1.4096720218658447,
"step": 632
},
{
"epoch": 0.7262313860252004,
"grad_norm": 0.11203691363334656,
"learning_rate": 9.125824895977334e-06,
"loss": 0.7960629463195801,
"step": 634
},
{
"epoch": 0.7285223367697594,
"grad_norm": 0.17772053182125092,
"learning_rate": 9.11908260688344e-06,
"loss": 1.3206441402435303,
"step": 636
},
{
"epoch": 0.7308132875143184,
"grad_norm": 0.09526784718036652,
"learning_rate": 9.112317236649822e-06,
"loss": 1.159492015838623,
"step": 638
},
{
"epoch": 0.7331042382588774,
"grad_norm": 0.11944857239723206,
"learning_rate": 9.105528828423455e-06,
"loss": 1.3975093364715576,
"step": 640
},
{
"epoch": 0.7353951890034365,
"grad_norm": 0.034794408828020096,
"learning_rate": 9.098717425498237e-06,
"loss": 1.132660150527954,
"step": 642
},
{
"epoch": 0.7376861397479955,
"grad_norm": 0.13092458248138428,
"learning_rate": 9.09188307131472e-06,
"loss": 1.3938100337982178,
"step": 644
},
{
"epoch": 0.7399770904925544,
"grad_norm": 0.033246614038944244,
"learning_rate": 9.085025809459826e-06,
"loss": 1.1749005317687988,
"step": 646
},
{
"epoch": 0.7422680412371134,
"grad_norm": 0.036673709750175476,
"learning_rate": 9.078145683666582e-06,
"loss": 1.0377411842346191,
"step": 648
},
{
"epoch": 0.7445589919816724,
"grad_norm": 0.1842445731163025,
"learning_rate": 9.071242737813824e-06,
"loss": 1.0705593824386597,
"step": 650
},
{
"epoch": 0.7468499427262314,
"grad_norm": 0.04267792031168938,
"learning_rate": 9.06431701592593e-06,
"loss": 1.1197433471679688,
"step": 652
},
{
"epoch": 0.7491408934707904,
"grad_norm": 0.11130751669406891,
"learning_rate": 9.057368562172535e-06,
"loss": 1.0569872856140137,
"step": 654
},
{
"epoch": 0.7514318442153494,
"grad_norm": 0.11254757642745972,
"learning_rate": 9.050397420868246e-06,
"loss": 1.1947691440582275,
"step": 656
},
{
"epoch": 0.7537227949599083,
"grad_norm": 0.11397068947553635,
"learning_rate": 9.043403636472368e-06,
"loss": 0.9924128651618958,
"step": 658
},
{
"epoch": 0.7560137457044673,
"grad_norm": 0.16056719422340393,
"learning_rate": 9.036387253588611e-06,
"loss": 1.4780884981155396,
"step": 660
},
{
"epoch": 0.7583046964490263,
"grad_norm": 0.1511063575744629,
"learning_rate": 9.02934831696481e-06,
"loss": 0.638008713722229,
"step": 662
},
{
"epoch": 0.7605956471935853,
"grad_norm": 0.04247143492102623,
"learning_rate": 9.022286871492641e-06,
"loss": 1.166100025177002,
"step": 664
},
{
"epoch": 0.7628865979381443,
"grad_norm": 0.10111839324235916,
"learning_rate": 9.015202962207329e-06,
"loss": 1.248875617980957,
"step": 666
},
{
"epoch": 0.7651775486827033,
"grad_norm": 0.10688889771699905,
"learning_rate": 9.008096634287372e-06,
"loss": 1.218861699104309,
"step": 668
},
{
"epoch": 0.7674684994272624,
"grad_norm": 0.14122603833675385,
"learning_rate": 9.000967933054236e-06,
"loss": 1.2160475254058838,
"step": 670
},
{
"epoch": 0.7697594501718213,
"grad_norm": 0.09510156512260437,
"learning_rate": 8.993816903972083e-06,
"loss": 1.3408292531967163,
"step": 672
},
{
"epoch": 0.7720504009163803,
"grad_norm": 0.032357197254896164,
"learning_rate": 8.986643592647473e-06,
"loss": 0.9297527074813843,
"step": 674
},
{
"epoch": 0.7743413516609393,
"grad_norm": 0.10768794268369675,
"learning_rate": 8.979448044829068e-06,
"loss": 1.2648060321807861,
"step": 676
},
{
"epoch": 0.7766323024054983,
"grad_norm": 0.1046016663312912,
"learning_rate": 8.972230306407354e-06,
"loss": 0.9922665357589722,
"step": 678
},
{
"epoch": 0.7789232531500573,
"grad_norm": 0.10708943009376526,
"learning_rate": 8.964990423414334e-06,
"loss": 1.2871710062026978,
"step": 680
},
{
"epoch": 0.7812142038946163,
"grad_norm": 0.11082033067941666,
"learning_rate": 8.957728442023243e-06,
"loss": 1.2588818073272705,
"step": 682
},
{
"epoch": 0.7835051546391752,
"grad_norm": 0.15941235423088074,
"learning_rate": 8.95044440854825e-06,
"loss": 1.4140057563781738,
"step": 684
},
{
"epoch": 0.7857961053837342,
"grad_norm": 0.10488958656787872,
"learning_rate": 8.943138369444165e-06,
"loss": 1.216564655303955,
"step": 686
},
{
"epoch": 0.7880870561282932,
"grad_norm": 0.09411117434501648,
"learning_rate": 8.935810371306143e-06,
"loss": 1.0623620748519897,
"step": 688
},
{
"epoch": 0.7903780068728522,
"grad_norm": 0.1014682799577713,
"learning_rate": 8.928460460869383e-06,
"loss": 1.3300588130950928,
"step": 690
},
{
"epoch": 0.7926689576174112,
"grad_norm": 0.1440027803182602,
"learning_rate": 8.921088685008833e-06,
"loss": 1.4568748474121094,
"step": 692
},
{
"epoch": 0.7949599083619702,
"grad_norm": 0.05546777695417404,
"learning_rate": 8.913695090738891e-06,
"loss": 1.1724140644073486,
"step": 694
},
{
"epoch": 0.7972508591065293,
"grad_norm": 0.0671909973025322,
"learning_rate": 8.906279725213105e-06,
"loss": 1.0626245737075806,
"step": 696
},
{
"epoch": 0.7995418098510882,
"grad_norm": 0.10349666327238083,
"learning_rate": 8.898842635723868e-06,
"loss": 1.2186851501464844,
"step": 698
},
{
"epoch": 0.8018327605956472,
"grad_norm": 0.09945754706859589,
"learning_rate": 8.891383869702127e-06,
"loss": 1.2207574844360352,
"step": 700
},
{
"epoch": 0.8041237113402062,
"grad_norm": 0.11375889927148819,
"learning_rate": 8.883903474717067e-06,
"loss": 1.1474676132202148,
"step": 702
},
{
"epoch": 0.8064146620847652,
"grad_norm": 0.033355362713336945,
"learning_rate": 8.876401498475818e-06,
"loss": 1.218152403831482,
"step": 704
},
{
"epoch": 0.8087056128293242,
"grad_norm": 0.0988926887512207,
"learning_rate": 8.868877988823148e-06,
"loss": 1.217521071434021,
"step": 706
},
{
"epoch": 0.8109965635738832,
"grad_norm": 0.10163167864084244,
"learning_rate": 8.861332993741155e-06,
"loss": 1.3256537914276123,
"step": 708
},
{
"epoch": 0.8132875143184422,
"grad_norm": 0.15732599794864655,
"learning_rate": 8.85376656134896e-06,
"loss": 1.1980226039886475,
"step": 710
},
{
"epoch": 0.8155784650630011,
"grad_norm": 0.12234638631343842,
"learning_rate": 8.846178739902409e-06,
"loss": 1.5141435861587524,
"step": 712
},
{
"epoch": 0.8178694158075601,
"grad_norm": 0.03240432217717171,
"learning_rate": 8.838569577793756e-06,
"loss": 1.032310962677002,
"step": 714
},
{
"epoch": 0.8201603665521191,
"grad_norm": 0.09315624833106995,
"learning_rate": 8.83093912355136e-06,
"loss": 1.1437958478927612,
"step": 716
},
{
"epoch": 0.8224513172966781,
"grad_norm": 0.11046700924634933,
"learning_rate": 8.82328742583937e-06,
"loss": 1.1637240648269653,
"step": 718
},
{
"epoch": 0.8247422680412371,
"grad_norm": 0.0841628909111023,
"learning_rate": 8.815614533457419e-06,
"loss": 1.0901081562042236,
"step": 720
},
{
"epoch": 0.827033218785796,
"grad_norm": 0.10208731889724731,
"learning_rate": 8.807920495340313e-06,
"loss": 1.2723631858825684,
"step": 722
},
{
"epoch": 0.8293241695303551,
"grad_norm": 0.10736406594514847,
"learning_rate": 8.800205360557714e-06,
"loss": 1.3155536651611328,
"step": 724
},
{
"epoch": 0.8316151202749141,
"grad_norm": 0.24670641124248505,
"learning_rate": 8.792469178313835e-06,
"loss": 1.1443843841552734,
"step": 726
},
{
"epoch": 0.8339060710194731,
"grad_norm": 0.11407999694347382,
"learning_rate": 8.784711997947121e-06,
"loss": 1.2705860137939453,
"step": 728
},
{
"epoch": 0.8361970217640321,
"grad_norm": 0.1088922917842865,
"learning_rate": 8.776933868929929e-06,
"loss": 1.2330048084259033,
"step": 730
},
{
"epoch": 0.8384879725085911,
"grad_norm": 0.04192059114575386,
"learning_rate": 8.769134840868228e-06,
"loss": 1.17930006980896,
"step": 732
},
{
"epoch": 0.8407789232531501,
"grad_norm": 0.10472884774208069,
"learning_rate": 8.761314963501265e-06,
"loss": 1.2617244720458984,
"step": 734
},
{
"epoch": 0.843069873997709,
"grad_norm": 0.036843329668045044,
"learning_rate": 8.753474286701263e-06,
"loss": 1.0243253707885742,
"step": 736
},
{
"epoch": 0.845360824742268,
"grad_norm": 0.14724741876125336,
"learning_rate": 8.74561286047309e-06,
"loss": 1.342656135559082,
"step": 738
},
{
"epoch": 0.847651775486827,
"grad_norm": 0.10858327895402908,
"learning_rate": 8.737730734953949e-06,
"loss": 1.3815677165985107,
"step": 740
},
{
"epoch": 0.849942726231386,
"grad_norm": 0.04007134586572647,
"learning_rate": 8.729827960413054e-06,
"loss": 1.385345458984375,
"step": 742
},
{
"epoch": 0.852233676975945,
"grad_norm": 0.1044420599937439,
"learning_rate": 8.721904587251315e-06,
"loss": 1.3040602207183838,
"step": 744
},
{
"epoch": 0.854524627720504,
"grad_norm": 0.034813616424798965,
"learning_rate": 8.713960666001e-06,
"loss": 1.0024397373199463,
"step": 746
},
{
"epoch": 0.856815578465063,
"grad_norm": 0.03408174216747284,
"learning_rate": 8.705996247325443e-06,
"loss": 1.1536363363265991,
"step": 748
},
{
"epoch": 0.8591065292096219,
"grad_norm": 0.10098287463188171,
"learning_rate": 8.698011382018687e-06,
"loss": 1.4317657947540283,
"step": 750
},
{
"epoch": 0.861397479954181,
"grad_norm": 0.12679803371429443,
"learning_rate": 8.690006121005187e-06,
"loss": 1.4823131561279297,
"step": 752
},
{
"epoch": 0.86368843069874,
"grad_norm": 0.11496943235397339,
"learning_rate": 8.681980515339464e-06,
"loss": 1.172654151916504,
"step": 754
},
{
"epoch": 0.865979381443299,
"grad_norm": 0.03390732407569885,
"learning_rate": 8.6739346162058e-06,
"loss": 1.1092259883880615,
"step": 756
},
{
"epoch": 0.868270332187858,
"grad_norm": 0.09868176281452179,
"learning_rate": 8.66586847491789e-06,
"loss": 1.1055936813354492,
"step": 758
},
{
"epoch": 0.870561282932417,
"grad_norm": 0.1280670017004013,
"learning_rate": 8.657782142918537e-06,
"loss": 1.2542710304260254,
"step": 760
},
{
"epoch": 0.872852233676976,
"grad_norm": 0.101657435297966,
"learning_rate": 8.649675671779304e-06,
"loss": 1.2047970294952393,
"step": 762
},
{
"epoch": 0.8751431844215349,
"grad_norm": 0.16795100271701813,
"learning_rate": 8.641549113200198e-06,
"loss": 1.1056723594665527,
"step": 764
},
{
"epoch": 0.8774341351660939,
"grad_norm": 0.045386865735054016,
"learning_rate": 8.633402519009337e-06,
"loss": 1.307566523551941,
"step": 766
},
{
"epoch": 0.8797250859106529,
"grad_norm": 0.09542455524206161,
"learning_rate": 8.625235941162615e-06,
"loss": 1.1384682655334473,
"step": 768
},
{
"epoch": 0.8820160366552119,
"grad_norm": 0.3451470136642456,
"learning_rate": 8.617049431743376e-06,
"loss": 1.0609101057052612,
"step": 770
},
{
"epoch": 0.8843069873997709,
"grad_norm": 0.043013185262680054,
"learning_rate": 8.60884304296208e-06,
"loss": 0.8049713373184204,
"step": 772
},
{
"epoch": 0.8865979381443299,
"grad_norm": 0.0997074693441391,
"learning_rate": 8.600616827155968e-06,
"loss": 1.2299315929412842,
"step": 774
},
{
"epoch": 0.8888888888888888,
"grad_norm": 0.11632440239191055,
"learning_rate": 8.592370836788738e-06,
"loss": 1.325179100036621,
"step": 776
},
{
"epoch": 0.8911798396334479,
"grad_norm": 0.11443230509757996,
"learning_rate": 8.584105124450192e-06,
"loss": 1.4285557270050049,
"step": 778
},
{
"epoch": 0.8934707903780069,
"grad_norm": 0.21292933821678162,
"learning_rate": 8.575819742855918e-06,
"loss": 1.6177531480789185,
"step": 780
},
{
"epoch": 0.8957617411225659,
"grad_norm": 0.03320442512631416,
"learning_rate": 8.567514744846947e-06,
"loss": 1.1647987365722656,
"step": 782
},
{
"epoch": 0.8980526918671249,
"grad_norm": 0.21047484874725342,
"learning_rate": 8.559190183389411e-06,
"loss": 0.6821029186248779,
"step": 784
},
{
"epoch": 0.9003436426116839,
"grad_norm": 0.08516217768192291,
"learning_rate": 8.550846111574216e-06,
"loss": 0.9978764057159424,
"step": 786
},
{
"epoch": 0.9026345933562429,
"grad_norm": 0.20089198648929596,
"learning_rate": 8.542482582616694e-06,
"loss": 1.1663529872894287,
"step": 788
},
{
"epoch": 0.9049255441008018,
"grad_norm": 0.08953817188739777,
"learning_rate": 8.53409964985627e-06,
"loss": 1.1072628498077393,
"step": 790
},
{
"epoch": 0.9072164948453608,
"grad_norm": 0.20335598289966583,
"learning_rate": 8.525697366756117e-06,
"loss": 1.1753069162368774,
"step": 792
},
{
"epoch": 0.9095074455899198,
"grad_norm": 0.10008153319358826,
"learning_rate": 8.51727578690282e-06,
"loss": 1.2463250160217285,
"step": 794
},
{
"epoch": 0.9117983963344788,
"grad_norm": 0.03641178086400032,
"learning_rate": 8.508834964006026e-06,
"loss": 1.1079061031341553,
"step": 796
},
{
"epoch": 0.9140893470790378,
"grad_norm": 0.11625296622514725,
"learning_rate": 8.500374951898111e-06,
"loss": 1.2706644535064697,
"step": 798
},
{
"epoch": 0.9163802978235968,
"grad_norm": 0.11276042461395264,
"learning_rate": 8.491895804533834e-06,
"loss": 1.3342492580413818,
"step": 800
},
{
"epoch": 0.9186712485681557,
"grad_norm": 0.10194030404090881,
"learning_rate": 8.483397575989984e-06,
"loss": 1.307550311088562,
"step": 802
},
{
"epoch": 0.9209621993127147,
"grad_norm": 0.13294701278209686,
"learning_rate": 8.474880320465054e-06,
"loss": 1.3148338794708252,
"step": 804
},
{
"epoch": 0.9232531500572738,
"grad_norm": 0.13636107742786407,
"learning_rate": 8.466344092278874e-06,
"loss": 1.4288030862808228,
"step": 806
},
{
"epoch": 0.9255441008018328,
"grad_norm": 0.17754265666007996,
"learning_rate": 8.457788945872278e-06,
"loss": 1.3068530559539795,
"step": 808
},
{
"epoch": 0.9278350515463918,
"grad_norm": 0.16769464313983917,
"learning_rate": 8.449214935806754e-06,
"loss": 1.2810285091400146,
"step": 810
},
{
"epoch": 0.9301260022909508,
"grad_norm": 0.17232482135295868,
"learning_rate": 8.440622116764095e-06,
"loss": 1.3921419382095337,
"step": 812
},
{
"epoch": 0.9324169530355098,
"grad_norm": 0.03388385474681854,
"learning_rate": 8.43201054354605e-06,
"loss": 1.0712862014770508,
"step": 814
},
{
"epoch": 0.9347079037800687,
"grad_norm": 0.24018289148807526,
"learning_rate": 8.423380271073975e-06,
"loss": 1.0584461688995361,
"step": 816
},
{
"epoch": 0.9369988545246277,
"grad_norm": 0.17711585760116577,
"learning_rate": 8.41473135438848e-06,
"loss": 1.2261204719543457,
"step": 818
},
{
"epoch": 0.9392898052691867,
"grad_norm": 0.036880411207675934,
"learning_rate": 8.406063848649089e-06,
"loss": 1.1226576566696167,
"step": 820
},
{
"epoch": 0.9415807560137457,
"grad_norm": 0.03658941015601158,
"learning_rate": 8.397377809133872e-06,
"loss": 1.0902087688446045,
"step": 822
},
{
"epoch": 0.9438717067583047,
"grad_norm": 0.10014797747135162,
"learning_rate": 8.388673291239098e-06,
"loss": 1.2248773574829102,
"step": 824
},
{
"epoch": 0.9461626575028637,
"grad_norm": 0.10653302818536758,
"learning_rate": 8.379950350478899e-06,
"loss": 1.1442151069641113,
"step": 826
},
{
"epoch": 0.9484536082474226,
"grad_norm": 0.044977616518735886,
"learning_rate": 8.371209042484884e-06,
"loss": 1.190359354019165,
"step": 828
},
{
"epoch": 0.9507445589919816,
"grad_norm": 0.11633345484733582,
"learning_rate": 8.362449423005811e-06,
"loss": 0.9401731491088867,
"step": 830
},
{
"epoch": 0.9530355097365406,
"grad_norm": 0.10338146239519119,
"learning_rate": 8.353671547907218e-06,
"loss": 1.2098982334136963,
"step": 832
},
{
"epoch": 0.9553264604810997,
"grad_norm": 0.037584755569696426,
"learning_rate": 8.344875473171072e-06,
"loss": 1.0929856300354004,
"step": 834
},
{
"epoch": 0.9576174112256587,
"grad_norm": 0.037858281284570694,
"learning_rate": 8.33606125489541e-06,
"loss": 1.1181831359863281,
"step": 836
},
{
"epoch": 0.9599083619702177,
"grad_norm": 0.033685605973005295,
"learning_rate": 8.327228949293983e-06,
"loss": 1.1918667554855347,
"step": 838
},
{
"epoch": 0.9621993127147767,
"grad_norm": 0.10927028208971024,
"learning_rate": 8.318378612695893e-06,
"loss": 1.1812386512756348,
"step": 840
},
{
"epoch": 0.9644902634593356,
"grad_norm": 0.09622111916542053,
"learning_rate": 8.30951030154524e-06,
"loss": 1.2083207368850708,
"step": 842
},
{
"epoch": 0.9667812142038946,
"grad_norm": 0.09567157924175262,
"learning_rate": 8.300624072400757e-06,
"loss": 1.1253591775894165,
"step": 844
},
{
"epoch": 0.9690721649484536,
"grad_norm": 0.1568741798400879,
"learning_rate": 8.29171998193545e-06,
"loss": 1.2886474132537842,
"step": 846
},
{
"epoch": 0.9713631156930126,
"grad_norm": 0.21366208791732788,
"learning_rate": 8.28279808693624e-06,
"loss": 1.2641284465789795,
"step": 848
},
{
"epoch": 0.9736540664375716,
"grad_norm": 0.11669494956731796,
"learning_rate": 8.273858444303601e-06,
"loss": 1.2815788984298706,
"step": 850
},
{
"epoch": 0.9759450171821306,
"grad_norm": 0.12636975944042206,
"learning_rate": 8.26490111105119e-06,
"loss": 1.280737042427063,
"step": 852
},
{
"epoch": 0.9782359679266895,
"grad_norm": 0.09644594043493271,
"learning_rate": 8.25592614430549e-06,
"loss": 1.4965810775756836,
"step": 854
},
{
"epoch": 0.9805269186712485,
"grad_norm": 0.10086604207754135,
"learning_rate": 8.246933601305441e-06,
"loss": 1.2899303436279297,
"step": 856
},
{
"epoch": 0.9828178694158075,
"grad_norm": 0.03850960731506348,
"learning_rate": 8.237923539402083e-06,
"loss": 1.1815292835235596,
"step": 858
},
{
"epoch": 0.9851088201603666,
"grad_norm": 0.03153403848409653,
"learning_rate": 8.228896016058182e-06,
"loss": 1.0716750621795654,
"step": 860
},
{
"epoch": 0.9873997709049256,
"grad_norm": 0.5610406398773193,
"learning_rate": 8.219851088847866e-06,
"loss": 0.8392553329467773,
"step": 862
},
{
"epoch": 0.9896907216494846,
"grad_norm": 0.029171844944357872,
"learning_rate": 8.210788815456259e-06,
"loss": 0.5268185138702393,
"step": 864
},
{
"epoch": 0.9919816723940436,
"grad_norm": 0.048108626157045364,
"learning_rate": 8.201709253679113e-06,
"loss": 1.1043140888214111,
"step": 866
},
{
"epoch": 0.9942726231386025,
"grad_norm": 0.03910600021481514,
"learning_rate": 8.192612461422436e-06,
"loss": 1.1640534400939941,
"step": 868
},
{
"epoch": 0.9965635738831615,
"grad_norm": 0.14607083797454834,
"learning_rate": 8.18349849670213e-06,
"loss": 1.191579818725586,
"step": 870
},
{
"epoch": 0.9988545246277205,
"grad_norm": 0.1132192313671112,
"learning_rate": 8.174367417643614e-06,
"loss": 1.300398349761963,
"step": 872
},
{
"epoch": 1.0011454753722795,
"grad_norm": 0.03182588517665863,
"learning_rate": 8.165219282481454e-06,
"loss": 1.0970737934112549,
"step": 874
},
{
"epoch": 1.0034364261168385,
"grad_norm": 0.09465012699365616,
"learning_rate": 8.156054149558997e-06,
"loss": 1.2283321619033813,
"step": 876
},
{
"epoch": 1.0057273768613975,
"grad_norm": 0.09852684289216995,
"learning_rate": 8.146872077327992e-06,
"loss": 1.2641890048980713,
"step": 878
},
{
"epoch": 1.0080183276059564,
"grad_norm": 0.09122022986412048,
"learning_rate": 8.137673124348224e-06,
"loss": 1.2633693218231201,
"step": 880
},
{
"epoch": 1.0103092783505154,
"grad_norm": 0.0346677266061306,
"learning_rate": 8.128457349287134e-06,
"loss": 1.2950749397277832,
"step": 882
},
{
"epoch": 1.0126002290950744,
"grad_norm": 0.10290050506591797,
"learning_rate": 8.119224810919446e-06,
"loss": 1.155712604522705,
"step": 884
},
{
"epoch": 1.0148911798396334,
"grad_norm": 0.11368633806705475,
"learning_rate": 8.1099755681268e-06,
"loss": 1.2856930494308472,
"step": 886
},
{
"epoch": 1.0171821305841924,
"grad_norm": 0.09924730658531189,
"learning_rate": 8.10070967989737e-06,
"loss": 1.3079783916473389,
"step": 888
},
{
"epoch": 1.0194730813287514,
"grad_norm": 0.03917548060417175,
"learning_rate": 8.091427205325481e-06,
"loss": 1.1257250308990479,
"step": 890
},
{
"epoch": 1.0217640320733103,
"grad_norm": 0.10367637127637863,
"learning_rate": 8.082128203611245e-06,
"loss": 1.1606723070144653,
"step": 892
},
{
"epoch": 1.0240549828178693,
"grad_norm": 0.04987473413348198,
"learning_rate": 8.07281273406018e-06,
"loss": 1.2995059490203857,
"step": 894
},
{
"epoch": 1.0263459335624283,
"grad_norm": 0.1082967147231102,
"learning_rate": 8.063480856082822e-06,
"loss": 1.3261157274246216,
"step": 896
},
{
"epoch": 1.0286368843069873,
"grad_norm": 0.10131801664829254,
"learning_rate": 8.054132629194363e-06,
"loss": 1.2500728368759155,
"step": 898
},
{
"epoch": 1.0309278350515463,
"grad_norm": 0.20225836336612701,
"learning_rate": 8.044768113014253e-06,
"loss": 1.2365930080413818,
"step": 900
},
{
"epoch": 1.0332187857961055,
"grad_norm": 0.3037818372249603,
"learning_rate": 8.03538736726584e-06,
"loss": 0.9093201756477356,
"step": 902
},
{
"epoch": 1.0355097365406645,
"grad_norm": 0.09588132798671722,
"learning_rate": 8.025990451775963e-06,
"loss": 1.203194499015808,
"step": 904
},
{
"epoch": 1.0378006872852235,
"grad_norm": 0.09174283593893051,
"learning_rate": 8.016577426474602e-06,
"loss": 1.1838289499282837,
"step": 906
},
{
"epoch": 1.0400916380297824,
"grad_norm": 0.09558971971273422,
"learning_rate": 8.007148351394465e-06,
"loss": 1.2214634418487549,
"step": 908
},
{
"epoch": 1.0423825887743414,
"grad_norm": 0.06839102506637573,
"learning_rate": 7.99770328667063e-06,
"loss": 1.0787734985351562,
"step": 910
},
{
"epoch": 1.0446735395189004,
"grad_norm": 0.040618084371089935,
"learning_rate": 7.988242292540144e-06,
"loss": 1.246540904045105,
"step": 912
},
{
"epoch": 1.0469644902634594,
"grad_norm": 0.12011495232582092,
"learning_rate": 7.978765429341651e-06,
"loss": 1.1854981184005737,
"step": 914
},
{
"epoch": 1.0492554410080184,
"grad_norm": 0.12019508332014084,
"learning_rate": 7.969272757514997e-06,
"loss": 1.1858718395233154,
"step": 916
},
{
"epoch": 1.0515463917525774,
"grad_norm": 0.09292206168174744,
"learning_rate": 7.959764337600852e-06,
"loss": 1.2531483173370361,
"step": 918
},
{
"epoch": 1.0538373424971363,
"grad_norm": 0.030973954126238823,
"learning_rate": 7.950240230240323e-06,
"loss": 0.6899155974388123,
"step": 920
},
{
"epoch": 1.0561282932416953,
"grad_norm": 0.17956514656543732,
"learning_rate": 7.94070049617456e-06,
"loss": 1.1831200122833252,
"step": 922
},
{
"epoch": 1.0584192439862543,
"grad_norm": 0.10381026566028595,
"learning_rate": 7.93114519624438e-06,
"loss": 1.2370057106018066,
"step": 924
},
{
"epoch": 1.0607101947308133,
"grad_norm": 0.1092928797006607,
"learning_rate": 7.921574391389874e-06,
"loss": 1.1756248474121094,
"step": 926
},
{
"epoch": 1.0630011454753723,
"grad_norm": 0.0376867949962616,
"learning_rate": 7.911988142650008e-06,
"loss": 1.072977066040039,
"step": 928
},
{
"epoch": 1.0652920962199313,
"grad_norm": 0.32605865597724915,
"learning_rate": 7.902386511162257e-06,
"loss": 1.0307834148406982,
"step": 930
},
{
"epoch": 1.0675830469644902,
"grad_norm": 0.10231613367795944,
"learning_rate": 7.892769558162188e-06,
"loss": 1.206171989440918,
"step": 932
},
{
"epoch": 1.0698739977090492,
"grad_norm": 0.10850352048873901,
"learning_rate": 7.883137344983094e-06,
"loss": 0.7696674466133118,
"step": 934
},
{
"epoch": 1.0721649484536082,
"grad_norm": 0.10014284402132034,
"learning_rate": 7.873489933055586e-06,
"loss": 1.127623200416565,
"step": 936
},
{
"epoch": 1.0744558991981672,
"grad_norm": 0.06162659451365471,
"learning_rate": 7.863827383907202e-06,
"loss": 1.3174957036972046,
"step": 938
},
{
"epoch": 1.0767468499427262,
"grad_norm": 0.12214836478233337,
"learning_rate": 7.85414975916203e-06,
"loss": 1.235177993774414,
"step": 940
},
{
"epoch": 1.0790378006872852,
"grad_norm": 0.03565968945622444,
"learning_rate": 7.8444571205403e-06,
"loss": 1.1172802448272705,
"step": 942
},
{
"epoch": 1.0813287514318441,
"grad_norm": 0.04510068520903587,
"learning_rate": 7.834749529857991e-06,
"loss": 1.1934261322021484,
"step": 944
},
{
"epoch": 1.0836197021764031,
"grad_norm": 0.0956355631351471,
"learning_rate": 7.825027049026448e-06,
"loss": 1.2597328424453735,
"step": 946
},
{
"epoch": 1.0859106529209621,
"grad_norm": 0.09545722603797913,
"learning_rate": 7.81528974005197e-06,
"loss": 1.2638235092163086,
"step": 948
},
{
"epoch": 1.088201603665521,
"grad_norm": 0.1045144647359848,
"learning_rate": 7.805537665035435e-06,
"loss": 1.1498044729232788,
"step": 950
},
{
"epoch": 1.09049255441008,
"grad_norm": 0.09164479374885559,
"learning_rate": 7.795770886171885e-06,
"loss": 1.224025011062622,
"step": 952
},
{
"epoch": 1.0927835051546393,
"grad_norm": 0.03645243123173714,
"learning_rate": 7.785989465750144e-06,
"loss": 1.1202518939971924,
"step": 954
},
{
"epoch": 1.0950744558991983,
"grad_norm": 0.09394631534814835,
"learning_rate": 7.776193466152408e-06,
"loss": 1.231307864189148,
"step": 956
},
{
"epoch": 1.0973654066437573,
"grad_norm": 0.08598010987043381,
"learning_rate": 7.766382949853856e-06,
"loss": 1.1217191219329834,
"step": 958
},
{
"epoch": 1.0996563573883162,
"grad_norm": 0.04581398144364357,
"learning_rate": 7.756557979422254e-06,
"loss": 1.2090941667556763,
"step": 960
},
{
"epoch": 1.1019473081328752,
"grad_norm": 0.09387822449207306,
"learning_rate": 7.746718617517541e-06,
"loss": 1.245224952697754,
"step": 962
},
{
"epoch": 1.1042382588774342,
"grad_norm": 0.14695440232753754,
"learning_rate": 7.73686492689145e-06,
"loss": 1.2612382173538208,
"step": 964
},
{
"epoch": 1.1065292096219932,
"grad_norm": 0.09694456309080124,
"learning_rate": 7.726996970387087e-06,
"loss": 1.185286283493042,
"step": 966
},
{
"epoch": 1.1088201603665522,
"grad_norm": 0.1064264178276062,
"learning_rate": 7.717114810938548e-06,
"loss": 1.2336416244506836,
"step": 968
},
{
"epoch": 1.1111111111111112,
"grad_norm": 0.09071429073810577,
"learning_rate": 7.707218511570506e-06,
"loss": 0.9867799282073975,
"step": 970
},
{
"epoch": 1.1134020618556701,
"grad_norm": 0.0929940938949585,
"learning_rate": 7.697308135397819e-06,
"loss": 1.303302526473999,
"step": 972
},
{
"epoch": 1.1156930126002291,
"grad_norm": 0.09926512837409973,
"learning_rate": 7.687383745625113e-06,
"loss": 0.7288157939910889,
"step": 974
},
{
"epoch": 1.1179839633447881,
"grad_norm": 0.12962137162685394,
"learning_rate": 7.67744540554639e-06,
"loss": 1.3191977739334106,
"step": 976
},
{
"epoch": 1.120274914089347,
"grad_norm": 0.14595907926559448,
"learning_rate": 7.667493178544626e-06,
"loss": 1.0432918071746826,
"step": 978
},
{
"epoch": 1.122565864833906,
"grad_norm": 0.11306449770927429,
"learning_rate": 7.65752712809136e-06,
"loss": 1.110289454460144,
"step": 980
},
{
"epoch": 1.124856815578465,
"grad_norm": 0.13710643351078033,
"learning_rate": 7.64754731774629e-06,
"loss": 1.3338663578033447,
"step": 982
},
{
"epoch": 1.127147766323024,
"grad_norm": 0.14222948253154755,
"learning_rate": 7.637553811156871e-06,
"loss": 1.2077035903930664,
"step": 984
},
{
"epoch": 1.129438717067583,
"grad_norm": 0.1056067943572998,
"learning_rate": 7.627546672057908e-06,
"loss": 1.1796958446502686,
"step": 986
},
{
"epoch": 1.131729667812142,
"grad_norm": 0.12675486505031586,
"learning_rate": 7.617525964271149e-06,
"loss": 1.2140767574310303,
"step": 988
},
{
"epoch": 1.134020618556701,
"grad_norm": 0.09778056293725967,
"learning_rate": 7.607491751704876e-06,
"loss": 1.1973233222961426,
"step": 990
},
{
"epoch": 1.13631156930126,
"grad_norm": 0.043336447328329086,
"learning_rate": 7.5974440983535015e-06,
"loss": 1.0020596981048584,
"step": 992
},
{
"epoch": 1.138602520045819,
"grad_norm": 0.10499600321054459,
"learning_rate": 7.587383068297157e-06,
"loss": 1.1112275123596191,
"step": 994
},
{
"epoch": 1.140893470790378,
"grad_norm": 0.0355721153318882,
"learning_rate": 7.577308725701285e-06,
"loss": 0.9966940879821777,
"step": 996
},
{
"epoch": 1.143184421534937,
"grad_norm": 0.0992012470960617,
"learning_rate": 7.567221134816235e-06,
"loss": 1.1800178289413452,
"step": 998
},
{
"epoch": 1.145475372279496,
"grad_norm": 0.10058074444532394,
"learning_rate": 7.557120359976843e-06,
"loss": 1.1242153644561768,
"step": 1000
},
{
"epoch": 1.147766323024055,
"grad_norm": 0.19073814153671265,
"learning_rate": 7.547006465602026e-06,
"loss": 1.0217759609222412,
"step": 1002
},
{
"epoch": 1.1500572737686139,
"grad_norm": 0.09970526397228241,
"learning_rate": 7.5368795161943835e-06,
"loss": 1.3643629550933838,
"step": 1004
},
{
"epoch": 1.1523482245131729,
"grad_norm": 0.06471025198698044,
"learning_rate": 7.526739576339761e-06,
"loss": 1.0752243995666504,
"step": 1006
},
{
"epoch": 1.1546391752577319,
"grad_norm": 0.042841047048568726,
"learning_rate": 7.516586710706862e-06,
"loss": 1.0762394666671753,
"step": 1008
},
{
"epoch": 1.1569301260022908,
"grad_norm": 0.09638847410678864,
"learning_rate": 7.506420984046823e-06,
"loss": 1.1684939861297607,
"step": 1010
},
{
"epoch": 1.1592210767468498,
"grad_norm": 0.1367417573928833,
"learning_rate": 7.496242461192801e-06,
"loss": 1.1517741680145264,
"step": 1012
},
{
"epoch": 1.161512027491409,
"grad_norm": 0.03387519717216492,
"learning_rate": 7.486051207059567e-06,
"loss": 0.9555083513259888,
"step": 1014
},
{
"epoch": 1.163802978235968,
"grad_norm": 0.1254238784313202,
"learning_rate": 7.475847286643081e-06,
"loss": 1.1272577047348022,
"step": 1016
},
{
"epoch": 1.166093928980527,
"grad_norm": 0.0902896374464035,
"learning_rate": 7.46563076502009e-06,
"loss": 1.1893162727355957,
"step": 1018
},
{
"epoch": 1.168384879725086,
"grad_norm": 0.09928741306066513,
"learning_rate": 7.4554017073477e-06,
"loss": 1.0811066627502441,
"step": 1020
},
{
"epoch": 1.170675830469645,
"grad_norm": 0.03147672861814499,
"learning_rate": 7.445160178862977e-06,
"loss": 0.9947164058685303,
"step": 1022
},
{
"epoch": 1.172966781214204,
"grad_norm": 0.04409791901707649,
"learning_rate": 7.434906244882508e-06,
"loss": 0.9831681847572327,
"step": 1024
},
{
"epoch": 1.175257731958763,
"grad_norm": 0.20074661076068878,
"learning_rate": 7.42463997080201e-06,
"loss": 0.9452756643295288,
"step": 1026
},
{
"epoch": 1.177548682703322,
"grad_norm": 0.09279465675354004,
"learning_rate": 7.414361422095894e-06,
"loss": 1.1467375755310059,
"step": 1028
},
{
"epoch": 1.179839633447881,
"grad_norm": 0.09862932562828064,
"learning_rate": 7.404070664316855e-06,
"loss": 1.4994152784347534,
"step": 1030
},
{
"epoch": 1.1821305841924399,
"grad_norm": 0.15487772226333618,
"learning_rate": 7.393767763095452e-06,
"loss": 0.6926403045654297,
"step": 1032
},
{
"epoch": 1.1844215349369989,
"grad_norm": 0.03889283537864685,
"learning_rate": 7.383452784139694e-06,
"loss": 1.1457732915878296,
"step": 1034
},
{
"epoch": 1.1867124856815578,
"grad_norm": 0.20745863020420074,
"learning_rate": 7.37312579323461e-06,
"loss": 0.6812342405319214,
"step": 1036
},
{
"epoch": 1.1890034364261168,
"grad_norm": 0.15700620412826538,
"learning_rate": 7.362786856241845e-06,
"loss": 1.1620197296142578,
"step": 1038
},
{
"epoch": 1.1912943871706758,
"grad_norm": 0.031710829585790634,
"learning_rate": 7.3524360390992275e-06,
"loss": 1.0484931468963623,
"step": 1040
},
{
"epoch": 1.1935853379152348,
"grad_norm": 0.10261533409357071,
"learning_rate": 7.342073407820351e-06,
"loss": 0.7066391706466675,
"step": 1042
},
{
"epoch": 1.1958762886597938,
"grad_norm": 0.18190829455852509,
"learning_rate": 7.331699028494161e-06,
"loss": 1.1231335401535034,
"step": 1044
},
{
"epoch": 1.1981672394043528,
"grad_norm": 0.2540644109249115,
"learning_rate": 7.321312967284518e-06,
"loss": 1.2889418601989746,
"step": 1046
},
{
"epoch": 1.2004581901489118,
"grad_norm": 0.22046895325183868,
"learning_rate": 7.310915290429799e-06,
"loss": 1.2246109247207642,
"step": 1048
},
{
"epoch": 1.2027491408934707,
"grad_norm": 0.10056769847869873,
"learning_rate": 7.300506064242448e-06,
"loss": 1.1946299076080322,
"step": 1050
},
{
"epoch": 1.2050400916380297,
"grad_norm": 0.09971389919519424,
"learning_rate": 7.290085355108573e-06,
"loss": 1.191572666168213,
"step": 1052
},
{
"epoch": 1.2073310423825887,
"grad_norm": 0.047876957803964615,
"learning_rate": 7.279653229487517e-06,
"loss": 1.1152572631835938,
"step": 1054
},
{
"epoch": 1.2096219931271477,
"grad_norm": 0.038673873990774155,
"learning_rate": 7.269209753911426e-06,
"loss": 1.1568899154663086,
"step": 1056
},
{
"epoch": 1.2119129438717067,
"grad_norm": 0.23321399092674255,
"learning_rate": 7.258754994984839e-06,
"loss": 0.8744759559631348,
"step": 1058
},
{
"epoch": 1.2142038946162657,
"grad_norm": 0.036044131964445114,
"learning_rate": 7.248289019384255e-06,
"loss": 1.0827986001968384,
"step": 1060
},
{
"epoch": 1.2164948453608249,
"grad_norm": 0.10701718181371689,
"learning_rate": 7.237811893857703e-06,
"loss": 1.2203986644744873,
"step": 1062
},
{
"epoch": 1.2187857961053838,
"grad_norm": 0.04341509938240051,
"learning_rate": 7.227323685224329e-06,
"loss": 1.0838563442230225,
"step": 1064
},
{
"epoch": 1.2210767468499428,
"grad_norm": 0.12016027420759201,
"learning_rate": 7.216824460373959e-06,
"loss": 1.111936092376709,
"step": 1066
},
{
"epoch": 1.2233676975945018,
"grad_norm": 0.22775952517986298,
"learning_rate": 7.206314286266676e-06,
"loss": 0.9151073694229126,
"step": 1068
},
{
"epoch": 1.2256586483390608,
"grad_norm": 0.10555245727300644,
"learning_rate": 7.195793229932397e-06,
"loss": 1.1633076667785645,
"step": 1070
},
{
"epoch": 1.2279495990836198,
"grad_norm": 0.1947338879108429,
"learning_rate": 7.185261358470436e-06,
"loss": 1.0789947509765625,
"step": 1072
},
{
"epoch": 1.2302405498281788,
"grad_norm": 0.032104719430208206,
"learning_rate": 7.174718739049087e-06,
"loss": 1.0364956855773926,
"step": 1074
},
{
"epoch": 1.2325315005727377,
"grad_norm": 0.09648443758487701,
"learning_rate": 7.164165438905186e-06,
"loss": 1.002371072769165,
"step": 1076
},
{
"epoch": 1.2348224513172967,
"grad_norm": 0.0989924743771553,
"learning_rate": 7.153601525343692e-06,
"loss": 1.1737957000732422,
"step": 1078
},
{
"epoch": 1.2371134020618557,
"grad_norm": 0.23395802080631256,
"learning_rate": 7.143027065737247e-06,
"loss": 1.070110559463501,
"step": 1080
},
{
"epoch": 1.2394043528064147,
"grad_norm": 0.05930604040622711,
"learning_rate": 7.132442127525754e-06,
"loss": 0.9888740181922913,
"step": 1082
},
{
"epoch": 1.2416953035509737,
"grad_norm": 0.10642720013856888,
"learning_rate": 7.121846778215946e-06,
"loss": 1.1234185695648193,
"step": 1084
},
{
"epoch": 1.2439862542955327,
"grad_norm": 0.09997859597206116,
"learning_rate": 7.111241085380951e-06,
"loss": 1.0492407083511353,
"step": 1086
},
{
"epoch": 1.2462772050400917,
"grad_norm": 0.09804773330688477,
"learning_rate": 7.100625116659867e-06,
"loss": 1.124328851699829,
"step": 1088
},
{
"epoch": 1.2485681557846506,
"grad_norm": 0.034081630408763885,
"learning_rate": 7.089998939757323e-06,
"loss": 1.098294734954834,
"step": 1090
},
{
"epoch": 1.2508591065292096,
"grad_norm": 0.13619168102741241,
"learning_rate": 7.07936262244306e-06,
"loss": 1.3204905986785889,
"step": 1092
},
{
"epoch": 1.2531500572737686,
"grad_norm": 0.04348979890346527,
"learning_rate": 7.068716232551484e-06,
"loss": 1.1737202405929565,
"step": 1094
},
{
"epoch": 1.2554410080183276,
"grad_norm": 0.1019207090139389,
"learning_rate": 7.05805983798124e-06,
"loss": 1.240083932876587,
"step": 1096
},
{
"epoch": 1.2577319587628866,
"grad_norm": 0.033528897911310196,
"learning_rate": 7.047393506694784e-06,
"loss": 1.1311392784118652,
"step": 1098
},
{
"epoch": 1.2600229095074456,
"grad_norm": 0.11534976214170456,
"learning_rate": 7.036717306717941e-06,
"loss": 1.096567153930664,
"step": 1100
},
{
"epoch": 1.2623138602520045,
"grad_norm": 0.17929257452487946,
"learning_rate": 7.026031306139476e-06,
"loss": 1.2111990451812744,
"step": 1102
},
{
"epoch": 1.2646048109965635,
"grad_norm": 0.06933283805847168,
"learning_rate": 7.015335573110655e-06,
"loss": 1.0563737154006958,
"step": 1104
},
{
"epoch": 1.2668957617411225,
"grad_norm": 0.2785737216472626,
"learning_rate": 7.004630175844821e-06,
"loss": 1.2497079372406006,
"step": 1106
},
{
"epoch": 1.2691867124856815,
"grad_norm": 0.10546611249446869,
"learning_rate": 6.9939151826169435e-06,
"loss": 1.2129267454147339,
"step": 1108
},
{
"epoch": 1.2714776632302405,
"grad_norm": 0.11540545523166656,
"learning_rate": 6.9831906617632015e-06,
"loss": 1.0614213943481445,
"step": 1110
},
{
"epoch": 1.2737686139747995,
"grad_norm": 0.10855554044246674,
"learning_rate": 6.972456681680526e-06,
"loss": 1.1559650897979736,
"step": 1112
},
{
"epoch": 1.2760595647193584,
"grad_norm": 0.10004860907793045,
"learning_rate": 6.961713310826186e-06,
"loss": 1.2221918106079102,
"step": 1114
},
{
"epoch": 1.2783505154639174,
"grad_norm": 0.11250676959753036,
"learning_rate": 6.950960617717331e-06,
"loss": 1.2137792110443115,
"step": 1116
},
{
"epoch": 1.2806414662084764,
"grad_norm": 0.10565535724163055,
"learning_rate": 6.940198670930575e-06,
"loss": 1.1705491542816162,
"step": 1118
},
{
"epoch": 1.2829324169530354,
"grad_norm": 0.1519957184791565,
"learning_rate": 6.929427539101542e-06,
"loss": 1.167567491531372,
"step": 1120
},
{
"epoch": 1.2852233676975944,
"grad_norm": 0.3252302408218384,
"learning_rate": 6.918647290924431e-06,
"loss": 0.6762139201164246,
"step": 1122
},
{
"epoch": 1.2875143184421534,
"grad_norm": 0.10116905719041824,
"learning_rate": 6.907857995151593e-06,
"loss": 1.3349790573120117,
"step": 1124
},
{
"epoch": 1.2898052691867126,
"grad_norm": 0.26615846157073975,
"learning_rate": 6.897059720593072e-06,
"loss": 1.2615208625793457,
"step": 1126
},
{
"epoch": 1.2920962199312716,
"grad_norm": 0.28528809547424316,
"learning_rate": 6.886252536116178e-06,
"loss": 0.752361536026001,
"step": 1128
},
{
"epoch": 1.2943871706758305,
"grad_norm": 0.10344325751066208,
"learning_rate": 6.875436510645046e-06,
"loss": 1.1990864276885986,
"step": 1130
},
{
"epoch": 1.2966781214203895,
"grad_norm": 0.18242916464805603,
"learning_rate": 6.864611713160195e-06,
"loss": 0.5096596479415894,
"step": 1132
},
{
"epoch": 1.2989690721649485,
"grad_norm": 0.10747134685516357,
"learning_rate": 6.853778212698085e-06,
"loss": 1.1793510913848877,
"step": 1134
},
{
"epoch": 1.3012600229095075,
"grad_norm": 0.10210155695676804,
"learning_rate": 6.842936078350687e-06,
"loss": 1.1578278541564941,
"step": 1136
},
{
"epoch": 1.3035509736540665,
"grad_norm": 0.09255377948284149,
"learning_rate": 6.832085379265032e-06,
"loss": 1.0329861640930176,
"step": 1138
},
{
"epoch": 1.3058419243986255,
"grad_norm": 0.23646067082881927,
"learning_rate": 6.821226184642769e-06,
"loss": 1.1236613988876343,
"step": 1140
},
{
"epoch": 1.3081328751431844,
"grad_norm": 0.17102530598640442,
"learning_rate": 6.810358563739741e-06,
"loss": 0.9725713729858398,
"step": 1142
},
{
"epoch": 1.3104238258877434,
"grad_norm": 0.07726361602544785,
"learning_rate": 6.799482585865517e-06,
"loss": 1.1270265579223633,
"step": 1144
},
{
"epoch": 1.3127147766323024,
"grad_norm": 0.2417464554309845,
"learning_rate": 6.788598320382969e-06,
"loss": 0.6766105890274048,
"step": 1146
},
{
"epoch": 1.3150057273768614,
"grad_norm": 0.12223535776138306,
"learning_rate": 6.777705836707827e-06,
"loss": 0.6427032947540283,
"step": 1148
},
{
"epoch": 1.3172966781214204,
"grad_norm": 0.035287536680698395,
"learning_rate": 6.766805204308232e-06,
"loss": 1.086439847946167,
"step": 1150
},
{
"epoch": 1.3195876288659794,
"grad_norm": 0.15588583052158356,
"learning_rate": 6.755896492704292e-06,
"loss": 0.690952479839325,
"step": 1152
},
{
"epoch": 1.3218785796105383,
"grad_norm": 0.036040663719177246,
"learning_rate": 6.7449797714676446e-06,
"loss": 1.0549356937408447,
"step": 1154
},
{
"epoch": 1.3241695303550973,
"grad_norm": 0.12717777490615845,
"learning_rate": 6.734055110221004e-06,
"loss": 1.0231207609176636,
"step": 1156
},
{
"epoch": 1.3264604810996563,
"grad_norm": 0.1270737648010254,
"learning_rate": 6.72312257863773e-06,
"loss": 0.8079697489738464,
"step": 1158
},
{
"epoch": 1.3287514318442153,
"grad_norm": 0.279462605714798,
"learning_rate": 6.712182246441372e-06,
"loss": 1.142216444015503,
"step": 1160
},
{
"epoch": 1.3310423825887743,
"grad_norm": 0.11068404465913773,
"learning_rate": 6.701234183405228e-06,
"loss": 1.2053807973861694,
"step": 1162
},
{
"epoch": 1.3333333333333333,
"grad_norm": 0.18221408128738403,
"learning_rate": 6.690278459351907e-06,
"loss": 1.0797538757324219,
"step": 1164
},
{
"epoch": 1.3356242840778922,
"grad_norm": 0.10141758620738983,
"learning_rate": 6.679315144152867e-06,
"loss": 1.0947381258010864,
"step": 1166
},
{
"epoch": 1.3379152348224514,
"grad_norm": 0.03468356281518936,
"learning_rate": 6.6683443077279885e-06,
"loss": 0.9148234724998474,
"step": 1168
},
{
"epoch": 1.3402061855670104,
"grad_norm": 0.2539704144001007,
"learning_rate": 6.6573660200451155e-06,
"loss": 0.7626911401748657,
"step": 1170
},
{
"epoch": 1.3424971363115694,
"grad_norm": 0.08057498931884766,
"learning_rate": 6.646380351119612e-06,
"loss": 1.0045770406723022,
"step": 1172
},
{
"epoch": 1.3447880870561284,
"grad_norm": 0.10992975533008575,
"learning_rate": 6.6353873710139185e-06,
"loss": 1.1851505041122437,
"step": 1174
},
{
"epoch": 1.3470790378006874,
"grad_norm": 0.05306171998381615,
"learning_rate": 6.624387149837105e-06,
"loss": 1.1102643013000488,
"step": 1176
},
{
"epoch": 1.3493699885452464,
"grad_norm": 0.03687020018696785,
"learning_rate": 6.613379757744419e-06,
"loss": 1.0992820262908936,
"step": 1178
},
{
"epoch": 1.3516609392898054,
"grad_norm": 0.04827934503555298,
"learning_rate": 6.602365264936843e-06,
"loss": 0.9393289089202881,
"step": 1180
},
{
"epoch": 1.3539518900343643,
"grad_norm": 0.03528865426778793,
"learning_rate": 6.5913437416606485e-06,
"loss": 1.1535475254058838,
"step": 1182
},
{
"epoch": 1.3562428407789233,
"grad_norm": 0.16577307879924774,
"learning_rate": 6.5803152582069365e-06,
"loss": 1.034282922744751,
"step": 1184
},
{
"epoch": 1.3585337915234823,
"grad_norm": 0.10467011481523514,
"learning_rate": 6.569279884911205e-06,
"loss": 1.1791577339172363,
"step": 1186
},
{
"epoch": 1.3608247422680413,
"grad_norm": 0.04116935655474663,
"learning_rate": 6.558237692152889e-06,
"loss": 1.1147451400756836,
"step": 1188
},
{
"epoch": 1.3631156930126003,
"grad_norm": 0.09797222167253494,
"learning_rate": 6.547188750354917e-06,
"loss": 1.1144938468933105,
"step": 1190
},
{
"epoch": 1.3654066437571593,
"grad_norm": 0.07124507427215576,
"learning_rate": 6.536133129983261e-06,
"loss": 1.0669329166412354,
"step": 1192
},
{
"epoch": 1.3676975945017182,
"grad_norm": 0.10105873644351959,
"learning_rate": 6.5250709015464834e-06,
"loss": 1.138416051864624,
"step": 1194
},
{
"epoch": 1.3699885452462772,
"grad_norm": 0.10858606547117233,
"learning_rate": 6.5140021355952935e-06,
"loss": 1.193711757659912,
"step": 1196
},
{
"epoch": 1.3722794959908362,
"grad_norm": 0.20762988924980164,
"learning_rate": 6.502926902722092e-06,
"loss": 1.0726079940795898,
"step": 1198
},
{
"epoch": 1.3745704467353952,
"grad_norm": 0.10816437751054764,
"learning_rate": 6.4918452735605245e-06,
"loss": 1.0476200580596924,
"step": 1200
},
{
"epoch": 1.3768613974799542,
"grad_norm": 0.08065369725227356,
"learning_rate": 6.4807573187850295e-06,
"loss": 0.9677037000656128,
"step": 1202
},
{
"epoch": 1.3791523482245132,
"grad_norm": 0.22005292773246765,
"learning_rate": 6.469663109110389e-06,
"loss": 0.9527295827865601,
"step": 1204
},
{
"epoch": 1.3814432989690721,
"grad_norm": 0.10050750523805618,
"learning_rate": 6.458562715291273e-06,
"loss": 1.1403913497924805,
"step": 1206
},
{
"epoch": 1.3837342497136311,
"grad_norm": 0.034232836216688156,
"learning_rate": 6.4474562081217975e-06,
"loss": 1.007227897644043,
"step": 1208
},
{
"epoch": 1.38602520045819,
"grad_norm": 0.11036860942840576,
"learning_rate": 6.436343658435059e-06,
"loss": 1.2514678239822388,
"step": 1210
},
{
"epoch": 1.388316151202749,
"grad_norm": 0.12463992834091187,
"learning_rate": 6.4252251371026984e-06,
"loss": 1.169065237045288,
"step": 1212
},
{
"epoch": 1.390607101947308,
"grad_norm": 0.10533028095960617,
"learning_rate": 6.414100715034437e-06,
"loss": 1.2318856716156006,
"step": 1214
},
{
"epoch": 1.392898052691867,
"grad_norm": 0.044587064534425735,
"learning_rate": 6.4029704631776334e-06,
"loss": 1.1643192768096924,
"step": 1216
},
{
"epoch": 1.395189003436426,
"grad_norm": 0.10597223788499832,
"learning_rate": 6.3918344525168176e-06,
"loss": 1.227967619895935,
"step": 1218
},
{
"epoch": 1.397479954180985,
"grad_norm": 0.1139107197523117,
"learning_rate": 6.380692754073257e-06,
"loss": 1.2038414478302002,
"step": 1220
},
{
"epoch": 1.399770904925544,
"grad_norm": 0.037755899131298065,
"learning_rate": 6.3695454389044885e-06,
"loss": 1.0631117820739746,
"step": 1222
},
{
"epoch": 1.402061855670103,
"grad_norm": 0.13147905468940735,
"learning_rate": 6.35839257810387e-06,
"loss": 1.1393280029296875,
"step": 1224
},
{
"epoch": 1.404352806414662,
"grad_norm": 0.10851233452558517,
"learning_rate": 6.347234242800131e-06,
"loss": 1.094638705253601,
"step": 1226
},
{
"epoch": 1.406643757159221,
"grad_norm": 0.20785588026046753,
"learning_rate": 6.336070504156907e-06,
"loss": 0.39179080724716187,
"step": 1228
},
{
"epoch": 1.40893470790378,
"grad_norm": 0.1313144713640213,
"learning_rate": 6.324901433372307e-06,
"loss": 1.1032609939575195,
"step": 1230
},
{
"epoch": 1.411225658648339,
"grad_norm": 0.10240298509597778,
"learning_rate": 6.313727101678433e-06,
"loss": 1.151193380355835,
"step": 1232
},
{
"epoch": 1.413516609392898,
"grad_norm": 0.16149231791496277,
"learning_rate": 6.302547580340949e-06,
"loss": 1.1676666736602783,
"step": 1234
},
{
"epoch": 1.4158075601374571,
"grad_norm": 0.033813584595918655,
"learning_rate": 6.291362940658612e-06,
"loss": 0.9767388701438904,
"step": 1236
},
{
"epoch": 1.418098510882016,
"grad_norm": 0.15657632052898407,
"learning_rate": 6.2801732539628205e-06,
"loss": 1.1303942203521729,
"step": 1238
},
{
"epoch": 1.420389461626575,
"grad_norm": 0.05029843747615814,
"learning_rate": 6.2689785916171656e-06,
"loss": 1.1143161058425903,
"step": 1240
},
{
"epoch": 1.422680412371134,
"grad_norm": 0.07945811748504639,
"learning_rate": 6.257779025016967e-06,
"loss": 1.1215400695800781,
"step": 1242
},
{
"epoch": 1.424971363115693,
"grad_norm": 0.12305624037981033,
"learning_rate": 6.246574625588824e-06,
"loss": 1.096776008605957,
"step": 1244
},
{
"epoch": 1.427262313860252,
"grad_norm": 0.10627985745668411,
"learning_rate": 6.235365464790158e-06,
"loss": 1.1911216974258423,
"step": 1246
},
{
"epoch": 1.429553264604811,
"grad_norm": 0.06793335825204849,
"learning_rate": 6.224151614108755e-06,
"loss": 1.1242647171020508,
"step": 1248
},
{
"epoch": 1.43184421534937,
"grad_norm": 0.10330016911029816,
"learning_rate": 6.212933145062313e-06,
"loss": 0.9869295954704285,
"step": 1250
},
{
"epoch": 1.434135166093929,
"grad_norm": 0.09804986417293549,
"learning_rate": 6.201710129197984e-06,
"loss": 1.0983688831329346,
"step": 1252
},
{
"epoch": 1.436426116838488,
"grad_norm": 0.1734234243631363,
"learning_rate": 6.190482638091917e-06,
"loss": 1.254033088684082,
"step": 1254
},
{
"epoch": 1.438717067583047,
"grad_norm": 0.21741870045661926,
"learning_rate": 6.179250743348801e-06,
"loss": 0.6976019144058228,
"step": 1256
},
{
"epoch": 1.441008018327606,
"grad_norm": 0.10098083317279816,
"learning_rate": 6.1680145166014135e-06,
"loss": 1.1904094219207764,
"step": 1258
},
{
"epoch": 1.443298969072165,
"grad_norm": 0.03571165353059769,
"learning_rate": 6.156774029510158e-06,
"loss": 1.0290796756744385,
"step": 1260
},
{
"epoch": 1.445589919816724,
"grad_norm": 0.04223793372511864,
"learning_rate": 6.145529353762608e-06,
"loss": 1.0820773839950562,
"step": 1262
},
{
"epoch": 1.447880870561283,
"grad_norm": 0.08061600476503372,
"learning_rate": 6.1342805610730515e-06,
"loss": 1.0397876501083374,
"step": 1264
},
{
"epoch": 1.4501718213058419,
"grad_norm": 0.1024092435836792,
"learning_rate": 6.123027723182034e-06,
"loss": 1.0905309915542603,
"step": 1266
},
{
"epoch": 1.4524627720504009,
"grad_norm": 0.1202508732676506,
"learning_rate": 6.111770911855895e-06,
"loss": 1.3037893772125244,
"step": 1268
},
{
"epoch": 1.4547537227949598,
"grad_norm": 0.11221358925104141,
"learning_rate": 6.100510198886324e-06,
"loss": 1.166261076927185,
"step": 1270
},
{
"epoch": 1.4570446735395188,
"grad_norm": 0.033267792314291,
"learning_rate": 6.089245656089882e-06,
"loss": 1.0799281597137451,
"step": 1272
},
{
"epoch": 1.4593356242840778,
"grad_norm": 0.20279622077941895,
"learning_rate": 6.077977355307564e-06,
"loss": 0.9322797060012817,
"step": 1274
},
{
"epoch": 1.461626575028637,
"grad_norm": 0.15380460023880005,
"learning_rate": 6.066705368404332e-06,
"loss": 1.0121163129806519,
"step": 1276
},
{
"epoch": 1.463917525773196,
"grad_norm": 0.11669916659593582,
"learning_rate": 6.0554297672686515e-06,
"loss": 1.159261703491211,
"step": 1278
},
{
"epoch": 1.466208476517755,
"grad_norm": 0.13704603910446167,
"learning_rate": 6.044150623812041e-06,
"loss": 1.2306301593780518,
"step": 1280
},
{
"epoch": 1.468499427262314,
"grad_norm": 0.10381774604320526,
"learning_rate": 6.032868009968611e-06,
"loss": 0.6591931581497192,
"step": 1282
},
{
"epoch": 1.470790378006873,
"grad_norm": 0.10854183882474899,
"learning_rate": 6.021581997694604e-06,
"loss": 1.2100105285644531,
"step": 1284
},
{
"epoch": 1.473081328751432,
"grad_norm": 0.05851943418383598,
"learning_rate": 6.010292658967937e-06,
"loss": 0.8533772230148315,
"step": 1286
},
{
"epoch": 1.475372279495991,
"grad_norm": 0.2286427617073059,
"learning_rate": 5.999000065787741e-06,
"loss": 1.1328685283660889,
"step": 1288
},
{
"epoch": 1.47766323024055,
"grad_norm": 0.0334353968501091,
"learning_rate": 5.987704290173904e-06,
"loss": 1.125452995300293,
"step": 1290
},
{
"epoch": 1.479954180985109,
"grad_norm": 0.11280791461467743,
"learning_rate": 5.976405404166609e-06,
"loss": 1.0590157508850098,
"step": 1292
},
{
"epoch": 1.4822451317296679,
"grad_norm": 0.11283966153860092,
"learning_rate": 5.965103479825874e-06,
"loss": 1.32285475730896,
"step": 1294
},
{
"epoch": 1.4845360824742269,
"grad_norm": 0.10280606895685196,
"learning_rate": 5.953798589231102e-06,
"loss": 1.1084046363830566,
"step": 1296
},
{
"epoch": 1.4868270332187858,
"grad_norm": 0.03574017807841301,
"learning_rate": 5.942490804480605e-06,
"loss": 1.0094585418701172,
"step": 1298
},
{
"epoch": 1.4891179839633448,
"grad_norm": 0.18968385457992554,
"learning_rate": 5.931180197691155e-06,
"loss": 1.0455267429351807,
"step": 1300
},
{
"epoch": 1.4914089347079038,
"grad_norm": 0.12325051426887512,
"learning_rate": 5.919866840997528e-06,
"loss": 0.8611164689064026,
"step": 1302
},
{
"epoch": 1.4936998854524628,
"grad_norm": 0.16876213252544403,
"learning_rate": 5.908550806552027e-06,
"loss": 1.2782384157180786,
"step": 1304
},
{
"epoch": 1.4959908361970218,
"grad_norm": 0.034040793776512146,
"learning_rate": 5.897232166524044e-06,
"loss": 1.1155657768249512,
"step": 1306
},
{
"epoch": 1.4982817869415808,
"grad_norm": 0.20665396749973297,
"learning_rate": 5.885910993099581e-06,
"loss": 0.8285434246063232,
"step": 1308
},
{
"epoch": 1.5005727376861397,
"grad_norm": 0.04009568691253662,
"learning_rate": 5.874587358480798e-06,
"loss": 1.116700530052185,
"step": 1310
},
{
"epoch": 1.5028636884306987,
"grad_norm": 0.14039941132068634,
"learning_rate": 5.863261334885553e-06,
"loss": 1.3074442148208618,
"step": 1312
},
{
"epoch": 1.5051546391752577,
"grad_norm": 0.037182215601205826,
"learning_rate": 5.851932994546941e-06,
"loss": 1.1499791145324707,
"step": 1314
},
{
"epoch": 1.5074455899198167,
"grad_norm": 0.2790544033050537,
"learning_rate": 5.840602409712831e-06,
"loss": 1.098710060119629,
"step": 1316
},
{
"epoch": 1.5097365406643757,
"grad_norm": 0.14369939267635345,
"learning_rate": 5.829269652645404e-06,
"loss": 0.9552765488624573,
"step": 1318
},
{
"epoch": 1.5120274914089347,
"grad_norm": 0.07011190801858902,
"learning_rate": 5.817934795620702e-06,
"loss": 0.8580765128135681,
"step": 1320
},
{
"epoch": 1.5143184421534936,
"grad_norm": 0.06823685765266418,
"learning_rate": 5.8065979109281515e-06,
"loss": 1.1291577816009521,
"step": 1322
},
{
"epoch": 1.5166093928980526,
"grad_norm": 0.10182562470436096,
"learning_rate": 5.795259070870114e-06,
"loss": 1.0331019163131714,
"step": 1324
},
{
"epoch": 1.5189003436426116,
"grad_norm": 0.046754829585552216,
"learning_rate": 5.783918347761424e-06,
"loss": 1.0500078201293945,
"step": 1326
},
{
"epoch": 1.5211912943871706,
"grad_norm": 0.27025946974754333,
"learning_rate": 5.77257581392892e-06,
"loss": 0.9796900749206543,
"step": 1328
},
{
"epoch": 1.5234822451317296,
"grad_norm": 0.10360334813594818,
"learning_rate": 5.761231541710994e-06,
"loss": 1.0910930633544922,
"step": 1330
},
{
"epoch": 1.5257731958762886,
"grad_norm": 0.18193475902080536,
"learning_rate": 5.7498856034571235e-06,
"loss": 0.933897852897644,
"step": 1332
},
{
"epoch": 1.5280641466208476,
"grad_norm": 0.04006630554795265,
"learning_rate": 5.7385380715274075e-06,
"loss": 0.7624289989471436,
"step": 1334
},
{
"epoch": 1.5303550973654065,
"grad_norm": 0.10916395485401154,
"learning_rate": 5.727189018292115e-06,
"loss": 1.0581674575805664,
"step": 1336
},
{
"epoch": 1.5326460481099655,
"grad_norm": 0.17847099900245667,
"learning_rate": 5.715838516131212e-06,
"loss": 0.7457289695739746,
"step": 1338
},
{
"epoch": 1.5349369988545245,
"grad_norm": 0.03762081265449524,
"learning_rate": 5.704486637433907e-06,
"loss": 1.13935387134552,
"step": 1340
},
{
"epoch": 1.5372279495990835,
"grad_norm": 0.038652852177619934,
"learning_rate": 5.6931334545981876e-06,
"loss": 1.084958791732788,
"step": 1342
},
{
"epoch": 1.5395189003436425,
"grad_norm": 0.3131856918334961,
"learning_rate": 5.68177904003036e-06,
"loss": 0.5619981288909912,
"step": 1344
},
{
"epoch": 1.5418098510882015,
"grad_norm": 0.23260389268398285,
"learning_rate": 5.670423466144585e-06,
"loss": 0.7182325124740601,
"step": 1346
},
{
"epoch": 1.5441008018327604,
"grad_norm": 0.04350382462143898,
"learning_rate": 5.659066805362416e-06,
"loss": 1.1135191917419434,
"step": 1348
},
{
"epoch": 1.5463917525773194,
"grad_norm": 0.032647889107465744,
"learning_rate": 5.647709130112339e-06,
"loss": 1.0672420263290405,
"step": 1350
},
{
"epoch": 1.5486827033218786,
"grad_norm": 0.20975816249847412,
"learning_rate": 5.6363505128293105e-06,
"loss": 1.0823326110839844,
"step": 1352
},
{
"epoch": 1.5509736540664376,
"grad_norm": 0.10048626363277435,
"learning_rate": 5.624991025954296e-06,
"loss": 1.1277685165405273,
"step": 1354
},
{
"epoch": 1.5532646048109966,
"grad_norm": 0.1214866116642952,
"learning_rate": 5.613630741933801e-06,
"loss": 1.0712693929672241,
"step": 1356
},
{
"epoch": 1.5555555555555556,
"grad_norm": 0.24091294407844543,
"learning_rate": 5.602269733219422e-06,
"loss": 0.8346927165985107,
"step": 1358
},
{
"epoch": 1.5578465063001146,
"grad_norm": 0.14887717366218567,
"learning_rate": 5.590908072267376e-06,
"loss": 1.1978172063827515,
"step": 1360
},
{
"epoch": 1.5601374570446735,
"grad_norm": 0.28521859645843506,
"learning_rate": 5.579545831538036e-06,
"loss": 0.9784339666366577,
"step": 1362
},
{
"epoch": 1.5624284077892325,
"grad_norm": 0.10800915211439133,
"learning_rate": 5.568183083495476e-06,
"loss": 1.2052793502807617,
"step": 1364
},
{
"epoch": 1.5647193585337915,
"grad_norm": 0.11474448442459106,
"learning_rate": 5.556819900607004e-06,
"loss": 1.0182474851608276,
"step": 1366
},
{
"epoch": 1.5670103092783505,
"grad_norm": 0.11097566783428192,
"learning_rate": 5.545456355342702e-06,
"loss": 1.1119379997253418,
"step": 1368
},
{
"epoch": 1.5693012600229095,
"grad_norm": 0.11000394821166992,
"learning_rate": 5.53409252017496e-06,
"loss": 1.1207211017608643,
"step": 1370
},
{
"epoch": 1.5715922107674685,
"grad_norm": 0.13542112708091736,
"learning_rate": 5.522728467578024e-06,
"loss": 1.3491647243499756,
"step": 1372
},
{
"epoch": 1.5738831615120275,
"grad_norm": 0.10330595821142197,
"learning_rate": 5.51136427002752e-06,
"loss": 1.1328983306884766,
"step": 1374
},
{
"epoch": 1.5761741122565864,
"grad_norm": 0.12681104242801666,
"learning_rate": 5.500000000000001e-06,
"loss": 1.0954585075378418,
"step": 1376
},
{
"epoch": 1.5784650630011456,
"grad_norm": 0.13023322820663452,
"learning_rate": 5.488635729972482e-06,
"loss": 1.0727986097335815,
"step": 1378
},
{
"epoch": 1.5807560137457046,
"grad_norm": 0.10406231880187988,
"learning_rate": 5.477271532421978e-06,
"loss": 1.1195957660675049,
"step": 1380
},
{
"epoch": 1.5830469644902636,
"grad_norm": 0.09854082018136978,
"learning_rate": 5.465907479825041e-06,
"loss": 1.2089966535568237,
"step": 1382
},
{
"epoch": 1.5853379152348226,
"grad_norm": 0.05718742311000824,
"learning_rate": 5.454543644657302e-06,
"loss": 1.0215727090835571,
"step": 1384
},
{
"epoch": 1.5876288659793816,
"grad_norm": 0.03251669183373451,
"learning_rate": 5.4431800993929985e-06,
"loss": 0.9556213617324829,
"step": 1386
},
{
"epoch": 1.5899198167239406,
"grad_norm": 0.18236595392227173,
"learning_rate": 5.431816916504526e-06,
"loss": 1.1348545551300049,
"step": 1388
},
{
"epoch": 1.5922107674684995,
"grad_norm": 0.03708245977759361,
"learning_rate": 5.420454168461966e-06,
"loss": 1.1390302181243896,
"step": 1390
},
{
"epoch": 1.5945017182130585,
"grad_norm": 0.18347609043121338,
"learning_rate": 5.4090919277326255e-06,
"loss": 0.8546306490898132,
"step": 1392
},
{
"epoch": 1.5967926689576175,
"grad_norm": 0.05539577454328537,
"learning_rate": 5.39773026678058e-06,
"loss": 0.792382001876831,
"step": 1394
},
{
"epoch": 1.5990836197021765,
"grad_norm": 0.1365845501422882,
"learning_rate": 5.386369258066201e-06,
"loss": 1.3111037015914917,
"step": 1396
},
{
"epoch": 1.6013745704467355,
"grad_norm": 0.03231525421142578,
"learning_rate": 5.3750089740457075e-06,
"loss": 1.0748083591461182,
"step": 1398
},
{
"epoch": 1.6036655211912945,
"grad_norm": 0.10971445590257645,
"learning_rate": 5.363649487170691e-06,
"loss": 1.0106375217437744,
"step": 1400
},
{
"epoch": 1.6059564719358534,
"grad_norm": 0.13350991904735565,
"learning_rate": 5.352290869887663e-06,
"loss": 1.202370524406433,
"step": 1402
},
{
"epoch": 1.6082474226804124,
"grad_norm": 0.09823276847600937,
"learning_rate": 5.340933194637586e-06,
"loss": 0.6368390321731567,
"step": 1404
},
{
"epoch": 1.6105383734249714,
"grad_norm": 0.11566170305013657,
"learning_rate": 5.329576533855415e-06,
"loss": 1.2471826076507568,
"step": 1406
},
{
"epoch": 1.6128293241695304,
"grad_norm": 0.03772154077887535,
"learning_rate": 5.3182209599696415e-06,
"loss": 0.9229879975318909,
"step": 1408
},
{
"epoch": 1.6151202749140894,
"grad_norm": 0.1596955955028534,
"learning_rate": 5.306866545401813e-06,
"loss": 1.1240687370300293,
"step": 1410
},
{
"epoch": 1.6174112256586484,
"grad_norm": 0.03676540032029152,
"learning_rate": 5.295513362566096e-06,
"loss": 0.7489064931869507,
"step": 1412
},
{
"epoch": 1.6197021764032073,
"grad_norm": 0.03176848217844963,
"learning_rate": 5.28416148386879e-06,
"loss": 1.0567030906677246,
"step": 1414
},
{
"epoch": 1.6219931271477663,
"grad_norm": 0.10595987737178802,
"learning_rate": 5.272810981707886e-06,
"loss": 1.0767991542816162,
"step": 1416
},
{
"epoch": 1.6242840778923253,
"grad_norm": 0.25980374217033386,
"learning_rate": 5.261461928472593e-06,
"loss": 0.9103140830993652,
"step": 1418
},
{
"epoch": 1.6265750286368843,
"grad_norm": 0.09718900173902512,
"learning_rate": 5.250114396542877e-06,
"loss": 1.0742920637130737,
"step": 1420
},
{
"epoch": 1.6288659793814433,
"grad_norm": 0.20691142976284027,
"learning_rate": 5.238768458289007e-06,
"loss": 1.2718803882598877,
"step": 1422
},
{
"epoch": 1.6311569301260023,
"grad_norm": 0.07437516003847122,
"learning_rate": 5.227424186071081e-06,
"loss": 0.9761091470718384,
"step": 1424
},
{
"epoch": 1.6334478808705613,
"grad_norm": 0.11155454814434052,
"learning_rate": 5.216081652238579e-06,
"loss": 1.1434038877487183,
"step": 1426
},
{
"epoch": 1.6357388316151202,
"grad_norm": 0.21897216141223907,
"learning_rate": 5.204740929129887e-06,
"loss": 1.1481974124908447,
"step": 1428
},
{
"epoch": 1.6380297823596792,
"grad_norm": 0.04085572436451912,
"learning_rate": 5.193402089071852e-06,
"loss": 1.125546932220459,
"step": 1430
},
{
"epoch": 1.6403207331042382,
"grad_norm": 0.03452634811401367,
"learning_rate": 5.1820652043793e-06,
"loss": 1.0230213403701782,
"step": 1432
},
{
"epoch": 1.6426116838487972,
"grad_norm": 0.10921933501958847,
"learning_rate": 5.1707303473545955e-06,
"loss": 1.2148507833480835,
"step": 1434
},
{
"epoch": 1.6449026345933562,
"grad_norm": 0.10001760721206665,
"learning_rate": 5.1593975902871705e-06,
"loss": 1.1477183103561401,
"step": 1436
},
{
"epoch": 1.6471935853379152,
"grad_norm": 0.0428081750869751,
"learning_rate": 5.1480670054530605e-06,
"loss": 1.2147631645202637,
"step": 1438
},
{
"epoch": 1.6494845360824741,
"grad_norm": 0.042964909225702286,
"learning_rate": 5.136738665114449e-06,
"loss": 1.0643807649612427,
"step": 1440
},
{
"epoch": 1.6517754868270331,
"grad_norm": 0.11255067586898804,
"learning_rate": 5.125412641519204e-06,
"loss": 1.1312243938446045,
"step": 1442
},
{
"epoch": 1.654066437571592,
"grad_norm": 0.13041959702968597,
"learning_rate": 5.114089006900422e-06,
"loss": 1.3391623497009277,
"step": 1444
},
{
"epoch": 1.656357388316151,
"grad_norm": 0.10491371154785156,
"learning_rate": 5.102767833475958e-06,
"loss": 1.1496613025665283,
"step": 1446
},
{
"epoch": 1.65864833906071,
"grad_norm": 0.18518638610839844,
"learning_rate": 5.091449193447974e-06,
"loss": 1.0849189758300781,
"step": 1448
},
{
"epoch": 1.660939289805269,
"grad_norm": 0.11973278224468231,
"learning_rate": 5.080133159002474e-06,
"loss": 1.0063197612762451,
"step": 1450
},
{
"epoch": 1.663230240549828,
"grad_norm": 0.1098853349685669,
"learning_rate": 5.068819802308845e-06,
"loss": 1.132586121559143,
"step": 1452
},
{
"epoch": 1.665521191294387,
"grad_norm": 0.1009354293346405,
"learning_rate": 5.057509195519398e-06,
"loss": 1.1691906452178955,
"step": 1454
},
{
"epoch": 1.667812142038946,
"grad_norm": 0.19018951058387756,
"learning_rate": 5.046201410768899e-06,
"loss": 1.1768360137939453,
"step": 1456
},
{
"epoch": 1.670103092783505,
"grad_norm": 0.2605174779891968,
"learning_rate": 5.034896520174126e-06,
"loss": 0.8101754188537598,
"step": 1458
},
{
"epoch": 1.6723940435280642,
"grad_norm": 0.07710061222314835,
"learning_rate": 5.023594595833393e-06,
"loss": 0.47093313932418823,
"step": 1460
},
{
"epoch": 1.6746849942726232,
"grad_norm": 0.15821929275989532,
"learning_rate": 5.0122957098260975e-06,
"loss": 1.3882811069488525,
"step": 1462
},
{
"epoch": 1.6769759450171822,
"grad_norm": 0.10836261510848999,
"learning_rate": 5.0009999342122606e-06,
"loss": 1.3038394451141357,
"step": 1464
},
{
"epoch": 1.6792668957617412,
"grad_norm": 0.1247217059135437,
"learning_rate": 4.989707341032064e-06,
"loss": 1.1847283840179443,
"step": 1466
},
{
"epoch": 1.6815578465063001,
"grad_norm": 0.1041129007935524,
"learning_rate": 4.978418002305399e-06,
"loss": 1.2749258279800415,
"step": 1468
},
{
"epoch": 1.6838487972508591,
"grad_norm": 0.09283646196126938,
"learning_rate": 4.967131990031391e-06,
"loss": 1.103422999382019,
"step": 1470
},
{
"epoch": 1.686139747995418,
"grad_norm": 0.10496700555086136,
"learning_rate": 4.955849376187961e-06,
"loss": 0.8029290437698364,
"step": 1472
},
{
"epoch": 1.688430698739977,
"grad_norm": 0.110666923224926,
"learning_rate": 4.944570232731351e-06,
"loss": 1.1437358856201172,
"step": 1474
},
{
"epoch": 1.690721649484536,
"grad_norm": 0.18121127784252167,
"learning_rate": 4.9332946315956695e-06,
"loss": 0.9108361005783081,
"step": 1476
},
{
"epoch": 1.693012600229095,
"grad_norm": 0.0387575663626194,
"learning_rate": 4.922022644692438e-06,
"loss": 1.1488621234893799,
"step": 1478
},
{
"epoch": 1.695303550973654,
"grad_norm": 0.1944582760334015,
"learning_rate": 4.910754343910121e-06,
"loss": 1.0412110090255737,
"step": 1480
},
{
"epoch": 1.697594501718213,
"grad_norm": 0.10680762678384781,
"learning_rate": 4.89948980111368e-06,
"loss": 1.1225993633270264,
"step": 1482
},
{
"epoch": 1.699885452462772,
"grad_norm": 0.0894724577665329,
"learning_rate": 4.888229088144106e-06,
"loss": 1.1438815593719482,
"step": 1484
},
{
"epoch": 1.702176403207331,
"grad_norm": 0.07776302099227905,
"learning_rate": 4.8769722768179686e-06,
"loss": 1.077296257019043,
"step": 1486
},
{
"epoch": 1.7044673539518902,
"grad_norm": 0.10468591749668121,
"learning_rate": 4.86571943892695e-06,
"loss": 1.1218129396438599,
"step": 1488
},
{
"epoch": 1.7067583046964492,
"grad_norm": 0.10482863336801529,
"learning_rate": 4.854470646237392e-06,
"loss": 1.1150727272033691,
"step": 1490
},
{
"epoch": 1.7090492554410082,
"grad_norm": 0.03965259715914726,
"learning_rate": 4.843225970489843e-06,
"loss": 0.772418737411499,
"step": 1492
},
{
"epoch": 1.7113402061855671,
"grad_norm": 0.0391482338309288,
"learning_rate": 4.831985483398587e-06,
"loss": 1.2282614707946777,
"step": 1494
},
{
"epoch": 1.7136311569301261,
"grad_norm": 0.03488003835082054,
"learning_rate": 4.820749256651202e-06,
"loss": 0.9615294933319092,
"step": 1496
},
{
"epoch": 1.7159221076746851,
"grad_norm": 0.04074535146355629,
"learning_rate": 4.809517361908086e-06,
"loss": 1.0762970447540283,
"step": 1498
},
{
"epoch": 1.718213058419244,
"grad_norm": 0.043285515159368515,
"learning_rate": 4.798289870802018e-06,
"loss": 0.6143720149993896,
"step": 1500
},
{
"epoch": 1.720504009163803,
"grad_norm": 0.1776183396577835,
"learning_rate": 4.787066854937689e-06,
"loss": 1.129189372062683,
"step": 1502
},
{
"epoch": 1.722794959908362,
"grad_norm": 0.143866166472435,
"learning_rate": 4.775848385891245e-06,
"loss": 1.3030314445495605,
"step": 1504
},
{
"epoch": 1.725085910652921,
"grad_norm": 0.1914648562669754,
"learning_rate": 4.7646345352098445e-06,
"loss": 0.6304170489311218,
"step": 1506
},
{
"epoch": 1.72737686139748,
"grad_norm": 0.04062773287296295,
"learning_rate": 4.753425374411177e-06,
"loss": 1.104649305343628,
"step": 1508
},
{
"epoch": 1.729667812142039,
"grad_norm": 0.037888024002313614,
"learning_rate": 4.742220974983036e-06,
"loss": 1.0669211149215698,
"step": 1510
},
{
"epoch": 1.731958762886598,
"grad_norm": 0.04267639294266701,
"learning_rate": 4.731021408382837e-06,
"loss": 1.1336133480072021,
"step": 1512
},
{
"epoch": 1.734249713631157,
"grad_norm": 0.10452942550182343,
"learning_rate": 4.7198267460371826e-06,
"loss": 1.3840365409851074,
"step": 1514
},
{
"epoch": 1.736540664375716,
"grad_norm": 0.10939494520425797,
"learning_rate": 4.70863705934139e-06,
"loss": 1.0531353950500488,
"step": 1516
},
{
"epoch": 1.738831615120275,
"grad_norm": 0.10143084079027176,
"learning_rate": 4.697452419659052e-06,
"loss": 1.216707468032837,
"step": 1518
},
{
"epoch": 1.741122565864834,
"grad_norm": 0.0358530730009079,
"learning_rate": 4.686272898321568e-06,
"loss": 1.0575618743896484,
"step": 1520
},
{
"epoch": 1.743413516609393,
"grad_norm": 0.032807547599077225,
"learning_rate": 4.675098566627695e-06,
"loss": 0.9459750652313232,
"step": 1522
},
{
"epoch": 1.745704467353952,
"grad_norm": 0.04335108399391174,
"learning_rate": 4.6639294958430946e-06,
"loss": 1.0062423944473267,
"step": 1524
},
{
"epoch": 1.747995418098511,
"grad_norm": 0.03665841370820999,
"learning_rate": 4.652765757199872e-06,
"loss": 1.0454283952713013,
"step": 1526
},
{
"epoch": 1.7502863688430699,
"grad_norm": 0.039896849542856216,
"learning_rate": 4.641607421896132e-06,
"loss": 0.9870510101318359,
"step": 1528
},
{
"epoch": 1.7525773195876289,
"grad_norm": 0.1942906230688095,
"learning_rate": 4.630454561095513e-06,
"loss": 0.8729791641235352,
"step": 1530
},
{
"epoch": 1.7548682703321878,
"grad_norm": 0.1113077700138092,
"learning_rate": 4.619307245926743e-06,
"loss": 1.1512171030044556,
"step": 1532
},
{
"epoch": 1.7571592210767468,
"grad_norm": 0.03571424260735512,
"learning_rate": 4.608165547483183e-06,
"loss": 1.2152729034423828,
"step": 1534
},
{
"epoch": 1.7594501718213058,
"grad_norm": 0.04555622860789299,
"learning_rate": 4.597029536822368e-06,
"loss": 0.6297133564949036,
"step": 1536
},
{
"epoch": 1.7617411225658648,
"grad_norm": 0.10868772864341736,
"learning_rate": 4.585899284965563e-06,
"loss": 1.1049292087554932,
"step": 1538
},
{
"epoch": 1.7640320733104238,
"grad_norm": 0.11979663372039795,
"learning_rate": 4.574774862897302e-06,
"loss": 1.1207685470581055,
"step": 1540
},
{
"epoch": 1.7663230240549828,
"grad_norm": 0.10041869431734085,
"learning_rate": 4.563656341564941e-06,
"loss": 1.046482801437378,
"step": 1542
},
{
"epoch": 1.7686139747995417,
"grad_norm": 0.13303270936012268,
"learning_rate": 4.552543791878205e-06,
"loss": 1.305598497390747,
"step": 1544
},
{
"epoch": 1.7709049255441007,
"grad_norm": 0.033517755568027496,
"learning_rate": 4.541437284708727e-06,
"loss": 0.9493808746337891,
"step": 1546
},
{
"epoch": 1.7731958762886597,
"grad_norm": 0.05814732611179352,
"learning_rate": 4.530336890889613e-06,
"loss": 1.0616586208343506,
"step": 1548
},
{
"epoch": 1.7754868270332187,
"grad_norm": 0.20835857093334198,
"learning_rate": 4.519242681214971e-06,
"loss": 0.7091290950775146,
"step": 1550
},
{
"epoch": 1.7777777777777777,
"grad_norm": 0.30232834815979004,
"learning_rate": 4.508154726439478e-06,
"loss": 1.036962866783142,
"step": 1552
},
{
"epoch": 1.7800687285223367,
"grad_norm": 0.11091706156730652,
"learning_rate": 4.49707309727791e-06,
"loss": 1.1720504760742188,
"step": 1554
},
{
"epoch": 1.7823596792668956,
"grad_norm": 0.06305071711540222,
"learning_rate": 4.485997864404709e-06,
"loss": 1.1253163814544678,
"step": 1556
},
{
"epoch": 1.7846506300114546,
"grad_norm": 0.07569506764411926,
"learning_rate": 4.474929098453519e-06,
"loss": 1.2266206741333008,
"step": 1558
},
{
"epoch": 1.7869415807560136,
"grad_norm": 0.035198770463466644,
"learning_rate": 4.463866870016741e-06,
"loss": 1.0873854160308838,
"step": 1560
},
{
"epoch": 1.7892325315005726,
"grad_norm": 0.19046899676322937,
"learning_rate": 4.452811249645085e-06,
"loss": 1.1058852672576904,
"step": 1562
},
{
"epoch": 1.7915234822451316,
"grad_norm": 0.1516924798488617,
"learning_rate": 4.441762307847113e-06,
"loss": 1.233043909072876,
"step": 1564
},
{
"epoch": 1.7938144329896906,
"grad_norm": 0.12142589688301086,
"learning_rate": 4.430720115088797e-06,
"loss": 1.1887891292572021,
"step": 1566
},
{
"epoch": 1.7961053837342495,
"grad_norm": 0.2711428701877594,
"learning_rate": 4.419684741793066e-06,
"loss": 0.8487293720245361,
"step": 1568
},
{
"epoch": 1.7983963344788088,
"grad_norm": 0.09922181069850922,
"learning_rate": 4.408656258339355e-06,
"loss": 1.0131475925445557,
"step": 1570
},
{
"epoch": 1.8006872852233677,
"grad_norm": 0.11347451061010361,
"learning_rate": 4.397634735063158e-06,
"loss": 1.1046950817108154,
"step": 1572
},
{
"epoch": 1.8029782359679267,
"grad_norm": 0.03463425114750862,
"learning_rate": 4.386620242255583e-06,
"loss": 1.096078872680664,
"step": 1574
},
{
"epoch": 1.8052691867124857,
"grad_norm": 0.1700165718793869,
"learning_rate": 4.375612850162897e-06,
"loss": 1.2153785228729248,
"step": 1576
},
{
"epoch": 1.8075601374570447,
"grad_norm": 0.10484757274389267,
"learning_rate": 4.364612628986082e-06,
"loss": 1.0014398097991943,
"step": 1578
},
{
"epoch": 1.8098510882016037,
"grad_norm": 0.23074306547641754,
"learning_rate": 4.35361964888039e-06,
"loss": 0.98294597864151,
"step": 1580
},
{
"epoch": 1.8121420389461627,
"grad_norm": 0.1004788875579834,
"learning_rate": 4.342633979954887e-06,
"loss": 1.168994665145874,
"step": 1582
},
{
"epoch": 1.8144329896907216,
"grad_norm": 0.2612786293029785,
"learning_rate": 4.331655692272011e-06,
"loss": 0.9925752282142639,
"step": 1584
},
{
"epoch": 1.8167239404352806,
"grad_norm": 0.037493567913770676,
"learning_rate": 4.320684855847135e-06,
"loss": 1.2383928298950195,
"step": 1586
},
{
"epoch": 1.8190148911798396,
"grad_norm": 0.08211533725261688,
"learning_rate": 4.309721540648094e-06,
"loss": 0.9583084583282471,
"step": 1588
},
{
"epoch": 1.8213058419243986,
"grad_norm": 0.034089405089616776,
"learning_rate": 4.298765816594773e-06,
"loss": 1.1073875427246094,
"step": 1590
},
{
"epoch": 1.8235967926689576,
"grad_norm": 0.11309397965669632,
"learning_rate": 4.2878177535586294e-06,
"loss": 1.0401699542999268,
"step": 1592
},
{
"epoch": 1.8258877434135166,
"grad_norm": 0.10397134721279144,
"learning_rate": 4.276877421362273e-06,
"loss": 1.1131930351257324,
"step": 1594
},
{
"epoch": 1.8281786941580758,
"grad_norm": 0.10276107490062714,
"learning_rate": 4.265944889778998e-06,
"loss": 1.2132148742675781,
"step": 1596
},
{
"epoch": 1.8304696449026348,
"grad_norm": 0.10183647274971008,
"learning_rate": 4.255020228532357e-06,
"loss": 1.179892659187317,
"step": 1598
},
{
"epoch": 1.8327605956471937,
"grad_norm": 0.11640535295009613,
"learning_rate": 4.244103507295709e-06,
"loss": 0.8675186634063721,
"step": 1600
},
{
"epoch": 1.8350515463917527,
"grad_norm": 0.1039697676897049,
"learning_rate": 4.233194795691767e-06,
"loss": 1.1238439083099365,
"step": 1602
},
{
"epoch": 1.8373424971363117,
"grad_norm": 0.11134682595729828,
"learning_rate": 4.222294163292173e-06,
"loss": 1.1501944065093994,
"step": 1604
},
{
"epoch": 1.8396334478808707,
"grad_norm": 0.09264901280403137,
"learning_rate": 4.211401679617031e-06,
"loss": 1.0822536945343018,
"step": 1606
},
{
"epoch": 1.8419243986254297,
"grad_norm": 0.09437301009893417,
"learning_rate": 4.200517414134487e-06,
"loss": 1.1331408023834229,
"step": 1608
},
{
"epoch": 1.8442153493699887,
"grad_norm": 0.042578209191560745,
"learning_rate": 4.1896414362602615e-06,
"loss": 0.9823397994041443,
"step": 1610
},
{
"epoch": 1.8465063001145476,
"grad_norm": 0.1340235471725464,
"learning_rate": 4.178773815357231e-06,
"loss": 1.3153901100158691,
"step": 1612
},
{
"epoch": 1.8487972508591066,
"grad_norm": 0.15407730638980865,
"learning_rate": 4.167914620734971e-06,
"loss": 1.372065782546997,
"step": 1614
},
{
"epoch": 1.8510882016036656,
"grad_norm": 0.09786330908536911,
"learning_rate": 4.157063921649314e-06,
"loss": 1.1111478805541992,
"step": 1616
},
{
"epoch": 1.8533791523482246,
"grad_norm": 0.03377949446439743,
"learning_rate": 4.146221787301916e-06,
"loss": 1.1279263496398926,
"step": 1618
},
{
"epoch": 1.8556701030927836,
"grad_norm": 0.10334134101867676,
"learning_rate": 4.135388286839806e-06,
"loss": 1.0680410861968994,
"step": 1620
},
{
"epoch": 1.8579610538373426,
"grad_norm": 0.13414083421230316,
"learning_rate": 4.1245634893549545e-06,
"loss": 1.1650376319885254,
"step": 1622
},
{
"epoch": 1.8602520045819015,
"grad_norm": 0.14877556264400482,
"learning_rate": 4.113747463883823e-06,
"loss": 1.367631196975708,
"step": 1624
},
{
"epoch": 1.8625429553264605,
"grad_norm": 0.031263113021850586,
"learning_rate": 4.102940279406928e-06,
"loss": 1.0940788984298706,
"step": 1626
},
{
"epoch": 1.8648339060710195,
"grad_norm": 0.04384366795420647,
"learning_rate": 4.092142004848408e-06,
"loss": 1.1474003791809082,
"step": 1628
},
{
"epoch": 1.8671248568155785,
"grad_norm": 0.03476424515247345,
"learning_rate": 4.08135270907557e-06,
"loss": 1.0565738677978516,
"step": 1630
},
{
"epoch": 1.8694158075601375,
"grad_norm": 0.14033450186252594,
"learning_rate": 4.070572460898462e-06,
"loss": 1.0951378345489502,
"step": 1632
},
{
"epoch": 1.8717067583046965,
"grad_norm": 0.03384094685316086,
"learning_rate": 4.059801329069426e-06,
"loss": 1.029362678527832,
"step": 1634
},
{
"epoch": 1.8739977090492554,
"grad_norm": 0.03602081537246704,
"learning_rate": 4.04903938228267e-06,
"loss": 1.1431338787078857,
"step": 1636
},
{
"epoch": 1.8762886597938144,
"grad_norm": 0.1386619508266449,
"learning_rate": 4.038286689173817e-06,
"loss": 1.1247670650482178,
"step": 1638
},
{
"epoch": 1.8785796105383734,
"grad_norm": 0.047482799738645554,
"learning_rate": 4.027543318319474e-06,
"loss": 1.0399794578552246,
"step": 1640
},
{
"epoch": 1.8808705612829324,
"grad_norm": 0.10432790964841843,
"learning_rate": 4.0168093382368e-06,
"loss": 1.1941301822662354,
"step": 1642
},
{
"epoch": 1.8831615120274914,
"grad_norm": 0.21906164288520813,
"learning_rate": 4.006084817383056e-06,
"loss": 0.24486073851585388,
"step": 1644
},
{
"epoch": 1.8854524627720504,
"grad_norm": 0.11396703869104385,
"learning_rate": 3.995369824155181e-06,
"loss": 1.13923978805542,
"step": 1646
},
{
"epoch": 1.8877434135166093,
"grad_norm": 0.10515320301055908,
"learning_rate": 3.984664426889346e-06,
"loss": 1.1944549083709717,
"step": 1648
},
{
"epoch": 1.8900343642611683,
"grad_norm": 0.12443188577890396,
"learning_rate": 3.973968693860527e-06,
"loss": 1.2203004360198975,
"step": 1650
},
{
"epoch": 1.8923253150057273,
"grad_norm": 0.1255001723766327,
"learning_rate": 3.96328269328206e-06,
"loss": 1.3052341938018799,
"step": 1652
},
{
"epoch": 1.8946162657502863,
"grad_norm": 0.12038350105285645,
"learning_rate": 3.952606493305216e-06,
"loss": 1.4456214904785156,
"step": 1654
},
{
"epoch": 1.8969072164948453,
"grad_norm": 0.04385162517428398,
"learning_rate": 3.941940162018761e-06,
"loss": 1.0220682621002197,
"step": 1656
},
{
"epoch": 1.8991981672394043,
"grad_norm": 0.0347161740064621,
"learning_rate": 3.931283767448518e-06,
"loss": 0.5256936550140381,
"step": 1658
},
{
"epoch": 1.9014891179839633,
"grad_norm": 0.3636147975921631,
"learning_rate": 3.920637377556942e-06,
"loss": 0.8584147691726685,
"step": 1660
},
{
"epoch": 1.9037800687285222,
"grad_norm": 0.11466902494430542,
"learning_rate": 3.910001060242677e-06,
"loss": 0.9281732439994812,
"step": 1662
},
{
"epoch": 1.9060710194730812,
"grad_norm": 0.1715666800737381,
"learning_rate": 3.899374883340137e-06,
"loss": 1.1350767612457275,
"step": 1664
},
{
"epoch": 1.9083619702176402,
"grad_norm": 0.10768167674541473,
"learning_rate": 3.88875891461905e-06,
"loss": 0.8779376745223999,
"step": 1666
},
{
"epoch": 1.9106529209621992,
"grad_norm": 0.05779425799846649,
"learning_rate": 3.878153221784054e-06,
"loss": 1.1153831481933594,
"step": 1668
},
{
"epoch": 1.9129438717067582,
"grad_norm": 0.11072687059640884,
"learning_rate": 3.867557872474248e-06,
"loss": 1.098388433456421,
"step": 1670
},
{
"epoch": 1.9152348224513172,
"grad_norm": 0.11078870296478271,
"learning_rate": 3.856972934262755e-06,
"loss": 1.206322193145752,
"step": 1672
},
{
"epoch": 1.9175257731958761,
"grad_norm": 0.11533273011445999,
"learning_rate": 3.84639847465631e-06,
"loss": 1.2478669881820679,
"step": 1674
},
{
"epoch": 1.9198167239404351,
"grad_norm": 0.1029491275548935,
"learning_rate": 3.835834561094815e-06,
"loss": 1.144696831703186,
"step": 1676
},
{
"epoch": 1.9221076746849943,
"grad_norm": 0.2026187777519226,
"learning_rate": 3.825281260950916e-06,
"loss": 1.330293893814087,
"step": 1678
},
{
"epoch": 1.9243986254295533,
"grad_norm": 0.16172434389591217,
"learning_rate": 3.814738641529566e-06,
"loss": 1.2509064674377441,
"step": 1680
},
{
"epoch": 1.9266895761741123,
"grad_norm": 0.03767630457878113,
"learning_rate": 3.8042067700676044e-06,
"loss": 0.9259130954742432,
"step": 1682
},
{
"epoch": 1.9289805269186713,
"grad_norm": 0.10923905670642853,
"learning_rate": 3.793685713733326e-06,
"loss": 1.2091310024261475,
"step": 1684
},
{
"epoch": 1.9312714776632303,
"grad_norm": 0.17867109179496765,
"learning_rate": 3.7831755396260416e-06,
"loss": 1.122981071472168,
"step": 1686
},
{
"epoch": 1.9335624284077892,
"grad_norm": 0.04535387083888054,
"learning_rate": 3.772676314775674e-06,
"loss": 0.9912348985671997,
"step": 1688
},
{
"epoch": 1.9358533791523482,
"grad_norm": 0.09147147089242935,
"learning_rate": 3.762188106142298e-06,
"loss": 0.8392770290374756,
"step": 1690
},
{
"epoch": 1.9381443298969072,
"grad_norm": 0.10301242023706436,
"learning_rate": 3.751710980615748e-06,
"loss": 1.138921856880188,
"step": 1692
},
{
"epoch": 1.9404352806414662,
"grad_norm": 0.04164798557758331,
"learning_rate": 3.741245005015162e-06,
"loss": 0.9997767210006714,
"step": 1694
},
{
"epoch": 1.9427262313860252,
"grad_norm": 0.10936956107616425,
"learning_rate": 3.7307902460885746e-06,
"loss": 1.1422123908996582,
"step": 1696
},
{
"epoch": 1.9450171821305842,
"grad_norm": 0.10265643894672394,
"learning_rate": 3.720346770512486e-06,
"loss": 1.039827585220337,
"step": 1698
},
{
"epoch": 1.9473081328751431,
"grad_norm": 0.1073760837316513,
"learning_rate": 3.709914644891427e-06,
"loss": 1.201385259628296,
"step": 1700
},
{
"epoch": 1.9495990836197021,
"grad_norm": 0.37369316816329956,
"learning_rate": 3.6994939357575533e-06,
"loss": 0.6540091633796692,
"step": 1702
},
{
"epoch": 1.9518900343642611,
"grad_norm": 0.0465695895254612,
"learning_rate": 3.689084709570202e-06,
"loss": 1.1277987957000732,
"step": 1704
},
{
"epoch": 1.9541809851088203,
"grad_norm": 0.09042780101299286,
"learning_rate": 3.6786870327154832e-06,
"loss": 1.1528184413909912,
"step": 1706
},
{
"epoch": 1.9564719358533793,
"grad_norm": 0.12911838293075562,
"learning_rate": 3.6683009715058416e-06,
"loss": 1.0916279554367065,
"step": 1708
},
{
"epoch": 1.9587628865979383,
"grad_norm": 0.12108128517866135,
"learning_rate": 3.657926592179649e-06,
"loss": 1.0954606533050537,
"step": 1710
},
{
"epoch": 1.9610538373424973,
"grad_norm": 0.053682778030633926,
"learning_rate": 3.647563960900774e-06,
"loss": 1.096421480178833,
"step": 1712
},
{
"epoch": 1.9633447880870563,
"grad_norm": 0.08973834663629532,
"learning_rate": 3.637213143758156e-06,
"loss": 1.0274837017059326,
"step": 1714
},
{
"epoch": 1.9656357388316152,
"grad_norm": 0.038498666137456894,
"learning_rate": 3.626874206765392e-06,
"loss": 1.1263710260391235,
"step": 1716
},
{
"epoch": 1.9679266895761742,
"grad_norm": 0.10485488921403885,
"learning_rate": 3.616547215860309e-06,
"loss": 1.0969138145446777,
"step": 1718
},
{
"epoch": 1.9702176403207332,
"grad_norm": 0.1671239733695984,
"learning_rate": 3.606232236904549e-06,
"loss": 1.2424817085266113,
"step": 1720
},
{
"epoch": 1.9725085910652922,
"grad_norm": 0.10209770500659943,
"learning_rate": 3.595929335683146e-06,
"loss": 0.9946843385696411,
"step": 1722
},
{
"epoch": 1.9747995418098512,
"grad_norm": 0.10356201231479645,
"learning_rate": 3.585638577904107e-06,
"loss": 1.1440200805664062,
"step": 1724
},
{
"epoch": 1.9770904925544102,
"grad_norm": 0.16762997210025787,
"learning_rate": 3.5753600291979917e-06,
"loss": 1.3794069290161133,
"step": 1726
},
{
"epoch": 1.9793814432989691,
"grad_norm": 0.16176775097846985,
"learning_rate": 3.5650937551174914e-06,
"loss": 1.1582558155059814,
"step": 1728
},
{
"epoch": 1.9816723940435281,
"grad_norm": 0.24708229303359985,
"learning_rate": 3.554839821137026e-06,
"loss": 0.991188108921051,
"step": 1730
},
{
"epoch": 1.9839633447880871,
"grad_norm": 0.03801441565155983,
"learning_rate": 3.5445982926523006e-06,
"loss": 1.1123301982879639,
"step": 1732
},
{
"epoch": 1.986254295532646,
"grad_norm": 0.11292216181755066,
"learning_rate": 3.534369234979914e-06,
"loss": 1.094356894493103,
"step": 1734
},
{
"epoch": 1.988545246277205,
"grad_norm": 0.24388711154460907,
"learning_rate": 3.5241527133569197e-06,
"loss": 0.09452319145202637,
"step": 1736
},
{
"epoch": 1.990836197021764,
"grad_norm": 0.09518136829137802,
"learning_rate": 3.5139487929404335e-06,
"loss": 0.9023909568786621,
"step": 1738
},
{
"epoch": 1.993127147766323,
"grad_norm": 0.10411088913679123,
"learning_rate": 3.5037575388071997e-06,
"loss": 1.18477463722229,
"step": 1740
},
{
"epoch": 1.995418098510882,
"grad_norm": 0.04860776290297508,
"learning_rate": 3.4935790159531786e-06,
"loss": 1.0331647396087646,
"step": 1742
},
{
"epoch": 1.997709049255441,
"grad_norm": 0.11631099134683609,
"learning_rate": 3.4834132892931405e-06,
"loss": 1.1645119190216064,
"step": 1744
},
{
"epoch": 2.0,
"grad_norm": 0.10443787276744843,
"learning_rate": 3.47326042366024e-06,
"loss": 1.174635887145996,
"step": 1746
},
{
"epoch": 2.002290950744559,
"grad_norm": 0.10160228610038757,
"learning_rate": 3.46312048380562e-06,
"loss": 1.040701985359192,
"step": 1748
},
{
"epoch": 2.004581901489118,
"grad_norm": 0.11828029900789261,
"learning_rate": 3.452993534397975e-06,
"loss": 1.2142775058746338,
"step": 1750
},
{
"epoch": 2.006872852233677,
"grad_norm": 0.10975845903158188,
"learning_rate": 3.4428796400231595e-06,
"loss": 1.1384990215301514,
"step": 1752
},
{
"epoch": 2.009163802978236,
"grad_norm": 0.3760664463043213,
"learning_rate": 3.4327788651837658e-06,
"loss": 1.1351079940795898,
"step": 1754
},
{
"epoch": 2.011454753722795,
"grad_norm": 0.046844061464071274,
"learning_rate": 3.422691274298714e-06,
"loss": 1.0275928974151611,
"step": 1756
},
{
"epoch": 2.013745704467354,
"grad_norm": 0.038912829011678696,
"learning_rate": 3.4126169317028445e-06,
"loss": 1.1176273822784424,
"step": 1758
},
{
"epoch": 2.016036655211913,
"grad_norm": 0.13894204795360565,
"learning_rate": 3.4025559016465003e-06,
"loss": 1.2932333946228027,
"step": 1760
},
{
"epoch": 2.018327605956472,
"grad_norm": 0.044606562703847885,
"learning_rate": 3.3925082482951253e-06,
"loss": 1.10593843460083,
"step": 1762
},
{
"epoch": 2.020618556701031,
"grad_norm": 0.4010094702243805,
"learning_rate": 3.382474035728852e-06,
"loss": 0.9274369478225708,
"step": 1764
},
{
"epoch": 2.02290950744559,
"grad_norm": 0.148308664560318,
"learning_rate": 3.372453327942092e-06,
"loss": 1.2721387147903442,
"step": 1766
},
{
"epoch": 2.025200458190149,
"grad_norm": 0.13406376540660858,
"learning_rate": 3.3624461888431315e-06,
"loss": 1.1050572395324707,
"step": 1768
},
{
"epoch": 2.027491408934708,
"grad_norm": 0.11722642928361893,
"learning_rate": 3.352452682253713e-06,
"loss": 1.1906096935272217,
"step": 1770
},
{
"epoch": 2.029782359679267,
"grad_norm": 0.10479209572076797,
"learning_rate": 3.3424728719086424e-06,
"loss": 1.179795265197754,
"step": 1772
},
{
"epoch": 2.0320733104238258,
"grad_norm": 0.042092155665159225,
"learning_rate": 3.332506821455376e-06,
"loss": 0.937066912651062,
"step": 1774
},
{
"epoch": 2.0343642611683848,
"grad_norm": 0.12035442888736725,
"learning_rate": 3.3225545944536132e-06,
"loss": 0.8189719319343567,
"step": 1776
},
{
"epoch": 2.0366552119129437,
"grad_norm": 0.10334596037864685,
"learning_rate": 3.3126162543748906e-06,
"loss": 1.0670874118804932,
"step": 1778
},
{
"epoch": 2.0389461626575027,
"grad_norm": 0.1044737845659256,
"learning_rate": 3.302691864602183e-06,
"loss": 1.1054072380065918,
"step": 1780
},
{
"epoch": 2.0412371134020617,
"grad_norm": 0.11545980721712112,
"learning_rate": 3.292781488429494e-06,
"loss": 1.1284576654434204,
"step": 1782
},
{
"epoch": 2.0435280641466207,
"grad_norm": 0.1243211105465889,
"learning_rate": 3.2828851890614534e-06,
"loss": 1.0694844722747803,
"step": 1784
},
{
"epoch": 2.0458190148911797,
"grad_norm": 0.0351092666387558,
"learning_rate": 3.2730030296129157e-06,
"loss": 1.1130070686340332,
"step": 1786
},
{
"epoch": 2.0481099656357387,
"grad_norm": 0.20803546905517578,
"learning_rate": 3.2631350731085526e-06,
"loss": 1.176312804222107,
"step": 1788
},
{
"epoch": 2.0504009163802976,
"grad_norm": 0.10609522461891174,
"learning_rate": 3.25328138248246e-06,
"loss": 1.115060567855835,
"step": 1790
},
{
"epoch": 2.0526918671248566,
"grad_norm": 0.2342589944601059,
"learning_rate": 3.2434420205777473e-06,
"loss": 0.6569336652755737,
"step": 1792
},
{
"epoch": 2.0549828178694156,
"grad_norm": 0.30435091257095337,
"learning_rate": 3.2336170501461433e-06,
"loss": 0.9199868440628052,
"step": 1794
},
{
"epoch": 2.0572737686139746,
"grad_norm": 0.14184576272964478,
"learning_rate": 3.2238065338475944e-06,
"loss": 0.9895941019058228,
"step": 1796
},
{
"epoch": 2.0595647193585336,
"grad_norm": 0.05588042736053467,
"learning_rate": 3.2140105342498577e-06,
"loss": 1.0880318880081177,
"step": 1798
},
{
"epoch": 2.0618556701030926,
"grad_norm": 0.1207311823964119,
"learning_rate": 3.2042291138281155e-06,
"loss": 1.1787630319595337,
"step": 1800
},
{
"epoch": 2.0641466208476515,
"grad_norm": 0.32253092527389526,
"learning_rate": 3.194462334964566e-06,
"loss": 0.6825082302093506,
"step": 1802
},
{
"epoch": 2.066437571592211,
"grad_norm": 0.11069416999816895,
"learning_rate": 3.184710259948032e-06,
"loss": 0.9402496814727783,
"step": 1804
},
{
"epoch": 2.06872852233677,
"grad_norm": 0.13846199214458466,
"learning_rate": 3.1749729509735555e-06,
"loss": 0.5835152864456177,
"step": 1806
},
{
"epoch": 2.071019473081329,
"grad_norm": 0.03694439306855202,
"learning_rate": 3.165250470142009e-06,
"loss": 1.0787923336029053,
"step": 1808
},
{
"epoch": 2.073310423825888,
"grad_norm": 0.2090117484331131,
"learning_rate": 3.155542879459702e-06,
"loss": 1.236689567565918,
"step": 1810
},
{
"epoch": 2.075601374570447,
"grad_norm": 0.105824314057827,
"learning_rate": 3.145850240837971e-06,
"loss": 1.1017699241638184,
"step": 1812
},
{
"epoch": 2.077892325315006,
"grad_norm": 0.10985516756772995,
"learning_rate": 3.136172616092801e-06,
"loss": 1.1386725902557373,
"step": 1814
},
{
"epoch": 2.080183276059565,
"grad_norm": 0.22561407089233398,
"learning_rate": 3.1265100669444184e-06,
"loss": 0.9274455308914185,
"step": 1816
},
{
"epoch": 2.082474226804124,
"grad_norm": 0.10488112270832062,
"learning_rate": 3.116862655016907e-06,
"loss": 1.1748979091644287,
"step": 1818
},
{
"epoch": 2.084765177548683,
"grad_norm": 0.10127851366996765,
"learning_rate": 3.107230441837812e-06,
"loss": 1.2072792053222656,
"step": 1820
},
{
"epoch": 2.087056128293242,
"grad_norm": 0.03725925460457802,
"learning_rate": 3.097613488837744e-06,
"loss": 1.1067794561386108,
"step": 1822
},
{
"epoch": 2.089347079037801,
"grad_norm": 0.24176861345767975,
"learning_rate": 3.0880118573499928e-06,
"loss": 1.04270601272583,
"step": 1824
},
{
"epoch": 2.09163802978236,
"grad_norm": 0.11087094247341156,
"learning_rate": 3.0784256086101265e-06,
"loss": 1.1116524934768677,
"step": 1826
},
{
"epoch": 2.093928980526919,
"grad_norm": 0.10241114348173141,
"learning_rate": 3.0688548037556203e-06,
"loss": 1.0798685550689697,
"step": 1828
},
{
"epoch": 2.0962199312714778,
"grad_norm": 0.12044376134872437,
"learning_rate": 3.0592995038254413e-06,
"loss": 1.1466073989868164,
"step": 1830
},
{
"epoch": 2.0985108820160367,
"grad_norm": 0.07996680587530136,
"learning_rate": 3.04975976975968e-06,
"loss": 1.0409008264541626,
"step": 1832
},
{
"epoch": 2.1008018327605957,
"grad_norm": 0.10078135132789612,
"learning_rate": 3.0402356623991487e-06,
"loss": 1.2206724882125854,
"step": 1834
},
{
"epoch": 2.1030927835051547,
"grad_norm": 0.12359591573476791,
"learning_rate": 3.0307272424850043e-06,
"loss": 1.1318553686141968,
"step": 1836
},
{
"epoch": 2.1053837342497137,
"grad_norm": 0.040033888071775436,
"learning_rate": 3.021234570658351e-06,
"loss": 1.0623432397842407,
"step": 1838
},
{
"epoch": 2.1076746849942727,
"grad_norm": 0.10040728002786636,
"learning_rate": 3.011757707459857e-06,
"loss": 1.1741615533828735,
"step": 1840
},
{
"epoch": 2.1099656357388317,
"grad_norm": 0.06718029826879501,
"learning_rate": 3.0022967133293716e-06,
"loss": 0.9849785566329956,
"step": 1842
},
{
"epoch": 2.1122565864833907,
"grad_norm": 0.14430569112300873,
"learning_rate": 2.992851648605536e-06,
"loss": 1.1471836566925049,
"step": 1844
},
{
"epoch": 2.1145475372279496,
"grad_norm": 0.20142653584480286,
"learning_rate": 2.9834225735254e-06,
"loss": 0.6019679307937622,
"step": 1846
},
{
"epoch": 2.1168384879725086,
"grad_norm": 0.1241523027420044,
"learning_rate": 2.9740095482240382e-06,
"loss": 1.193034052848816,
"step": 1848
},
{
"epoch": 2.1191294387170676,
"grad_norm": 0.035164572298526764,
"learning_rate": 2.964612632734163e-06,
"loss": 1.113813877105713,
"step": 1850
},
{
"epoch": 2.1214203894616266,
"grad_norm": 0.06513768434524536,
"learning_rate": 2.9552318869857464e-06,
"loss": 0.8912978172302246,
"step": 1852
},
{
"epoch": 2.1237113402061856,
"grad_norm": 0.04359002783894539,
"learning_rate": 2.9458673708056385e-06,
"loss": 1.1514525413513184,
"step": 1854
},
{
"epoch": 2.1260022909507446,
"grad_norm": 0.03506020829081535,
"learning_rate": 2.936519143917179e-06,
"loss": 1.0918513536453247,
"step": 1856
},
{
"epoch": 2.1282932416953035,
"grad_norm": 0.11882031708955765,
"learning_rate": 2.9271872659398224e-06,
"loss": 1.1890110969543457,
"step": 1858
},
{
"epoch": 2.1305841924398625,
"grad_norm": 0.12055717408657074,
"learning_rate": 2.9178717963887557e-06,
"loss": 1.0867390632629395,
"step": 1860
},
{
"epoch": 2.1328751431844215,
"grad_norm": 0.1105470135807991,
"learning_rate": 2.90857279467452e-06,
"loss": 1.144205093383789,
"step": 1862
},
{
"epoch": 2.1351660939289805,
"grad_norm": 0.15523314476013184,
"learning_rate": 2.899290320102631e-06,
"loss": 0.925602376461029,
"step": 1864
},
{
"epoch": 2.1374570446735395,
"grad_norm": 0.21147528290748596,
"learning_rate": 2.890024431873201e-06,
"loss": 0.8735249638557434,
"step": 1866
},
{
"epoch": 2.1397479954180985,
"grad_norm": 0.033189158886671066,
"learning_rate": 2.880775189080555e-06,
"loss": 1.0115811824798584,
"step": 1868
},
{
"epoch": 2.1420389461626574,
"grad_norm": 0.061869971454143524,
"learning_rate": 2.8715426507128687e-06,
"loss": 1.066664695739746,
"step": 1870
},
{
"epoch": 2.1443298969072164,
"grad_norm": 0.10503748059272766,
"learning_rate": 2.862326875651777e-06,
"loss": 1.129685401916504,
"step": 1872
},
{
"epoch": 2.1466208476517754,
"grad_norm": 0.10907737910747528,
"learning_rate": 2.853127922672008e-06,
"loss": 1.0200390815734863,
"step": 1874
},
{
"epoch": 2.1489117983963344,
"grad_norm": 0.1520504504442215,
"learning_rate": 2.8439458504410044e-06,
"loss": 1.0280253887176514,
"step": 1876
},
{
"epoch": 2.1512027491408934,
"grad_norm": 0.12227918952703476,
"learning_rate": 2.834780717518547e-06,
"loss": 1.050950527191162,
"step": 1878
},
{
"epoch": 2.1534936998854524,
"grad_norm": 0.03874862939119339,
"learning_rate": 2.825632582356387e-06,
"loss": 1.0017433166503906,
"step": 1880
},
{
"epoch": 2.1557846506300113,
"grad_norm": 0.04391537606716156,
"learning_rate": 2.8165015032978703e-06,
"loss": 1.0685606002807617,
"step": 1882
},
{
"epoch": 2.1580756013745703,
"grad_norm": 0.04769907891750336,
"learning_rate": 2.8073875385775663e-06,
"loss": 1.186462640762329,
"step": 1884
},
{
"epoch": 2.1603665521191293,
"grad_norm": 0.032127585262060165,
"learning_rate": 2.7982907463208896e-06,
"loss": 0.895263135433197,
"step": 1886
},
{
"epoch": 2.1626575028636883,
"grad_norm": 0.14522826671600342,
"learning_rate": 2.7892111845437416e-06,
"loss": 1.0644443035125732,
"step": 1888
},
{
"epoch": 2.1649484536082473,
"grad_norm": 0.11006541550159454,
"learning_rate": 2.780148911152135e-06,
"loss": 1.0343241691589355,
"step": 1890
},
{
"epoch": 2.1672394043528063,
"grad_norm": 0.03823330998420715,
"learning_rate": 2.7711039839418187e-06,
"loss": 1.0082577466964722,
"step": 1892
},
{
"epoch": 2.1695303550973652,
"grad_norm": 0.03993887081742287,
"learning_rate": 2.762076460597919e-06,
"loss": 1.0605108737945557,
"step": 1894
},
{
"epoch": 2.1718213058419242,
"grad_norm": 0.17623330652713776,
"learning_rate": 2.7530663986945605e-06,
"loss": 0.8171453475952148,
"step": 1896
},
{
"epoch": 2.174112256586483,
"grad_norm": 0.03557036072015762,
"learning_rate": 2.7440738556945122e-06,
"loss": 0.9885848760604858,
"step": 1898
},
{
"epoch": 2.176403207331042,
"grad_norm": 0.11463136225938797,
"learning_rate": 2.735098888948811e-06,
"loss": 0.8385453820228577,
"step": 1900
},
{
"epoch": 2.178694158075601,
"grad_norm": 0.20864970982074738,
"learning_rate": 2.726141555696399e-06,
"loss": 1.330157995223999,
"step": 1902
},
{
"epoch": 2.18098510882016,
"grad_norm": 0.10703415423631668,
"learning_rate": 2.7172019130637605e-06,
"loss": 1.1223435401916504,
"step": 1904
},
{
"epoch": 2.183276059564719,
"grad_norm": 0.10402292013168335,
"learning_rate": 2.708280018064551e-06,
"loss": 0.5987660884857178,
"step": 1906
},
{
"epoch": 2.1855670103092786,
"grad_norm": 0.1397983282804489,
"learning_rate": 2.699375927599245e-06,
"loss": 1.0533286333084106,
"step": 1908
},
{
"epoch": 2.1878579610538376,
"grad_norm": 0.130964994430542,
"learning_rate": 2.6904896984547614e-06,
"loss": 0.5356656908988953,
"step": 1910
},
{
"epoch": 2.1901489117983965,
"grad_norm": 0.11952181160449982,
"learning_rate": 2.6816213873041086e-06,
"loss": 1.1443605422973633,
"step": 1912
},
{
"epoch": 2.1924398625429555,
"grad_norm": 0.1906312108039856,
"learning_rate": 2.6727710507060166e-06,
"loss": 0.44051751494407654,
"step": 1914
},
{
"epoch": 2.1947308132875145,
"grad_norm": 0.10350323468446732,
"learning_rate": 2.6639387451045894e-06,
"loss": 1.1139777898788452,
"step": 1916
},
{
"epoch": 2.1970217640320735,
"grad_norm": 0.15858761966228485,
"learning_rate": 2.6551245268289293e-06,
"loss": 1.123609185218811,
"step": 1918
},
{
"epoch": 2.1993127147766325,
"grad_norm": 0.1445668339729309,
"learning_rate": 2.6463284520927834e-06,
"loss": 0.9767943620681763,
"step": 1920
},
{
"epoch": 2.2016036655211915,
"grad_norm": 0.10581092536449432,
"learning_rate": 2.63755057699419e-06,
"loss": 0.9770960807800293,
"step": 1922
},
{
"epoch": 2.2038946162657505,
"grad_norm": 0.11240138858556747,
"learning_rate": 2.6287909575151166e-06,
"loss": 1.1265208721160889,
"step": 1924
},
{
"epoch": 2.2061855670103094,
"grad_norm": 0.2250319868326187,
"learning_rate": 2.620049649521103e-06,
"loss": 0.9786791801452637,
"step": 1926
},
{
"epoch": 2.2084765177548684,
"grad_norm": 0.10950709134340286,
"learning_rate": 2.6113267087609018e-06,
"loss": 1.1055612564086914,
"step": 1928
},
{
"epoch": 2.2107674684994274,
"grad_norm": 0.1569567322731018,
"learning_rate": 2.6026221908661307e-06,
"loss": 1.0753288269042969,
"step": 1930
},
{
"epoch": 2.2130584192439864,
"grad_norm": 0.10235545039176941,
"learning_rate": 2.5939361513509124e-06,
"loss": 0.6077594757080078,
"step": 1932
},
{
"epoch": 2.2153493699885454,
"grad_norm": 0.12555944919586182,
"learning_rate": 2.5852686456115194e-06,
"loss": 1.0947010517120361,
"step": 1934
},
{
"epoch": 2.2176403207331044,
"grad_norm": 0.0357440747320652,
"learning_rate": 2.5766197289260277e-06,
"loss": 1.0835474729537964,
"step": 1936
},
{
"epoch": 2.2199312714776633,
"grad_norm": 0.035975709557533264,
"learning_rate": 2.5679894564539513e-06,
"loss": 1.0810333490371704,
"step": 1938
},
{
"epoch": 2.2222222222222223,
"grad_norm": 0.12334857881069183,
"learning_rate": 2.5593778832359062e-06,
"loss": 0.9912731051445007,
"step": 1940
},
{
"epoch": 2.2245131729667813,
"grad_norm": 0.03624502941966057,
"learning_rate": 2.550785064193246e-06,
"loss": 0.7342801094055176,
"step": 1942
},
{
"epoch": 2.2268041237113403,
"grad_norm": 0.10606859624385834,
"learning_rate": 2.5422110541277225e-06,
"loss": 1.1674457788467407,
"step": 1944
},
{
"epoch": 2.2290950744558993,
"grad_norm": 0.16007988154888153,
"learning_rate": 2.5336559077211285e-06,
"loss": 0.7999182939529419,
"step": 1946
},
{
"epoch": 2.2313860252004583,
"grad_norm": 0.21072755753993988,
"learning_rate": 2.5251196795349476e-06,
"loss": 0.7957800030708313,
"step": 1948
},
{
"epoch": 2.2336769759450172,
"grad_norm": 0.11326409876346588,
"learning_rate": 2.5166024240100175e-06,
"loss": 1.092682123184204,
"step": 1950
},
{
"epoch": 2.2359679266895762,
"grad_norm": 0.03987419232726097,
"learning_rate": 2.508104195466169e-06,
"loss": 1.0951638221740723,
"step": 1952
},
{
"epoch": 2.238258877434135,
"grad_norm": 0.2164270430803299,
"learning_rate": 2.4996250481018917e-06,
"loss": 0.6124863624572754,
"step": 1954
},
{
"epoch": 2.240549828178694,
"grad_norm": 0.1036994680762291,
"learning_rate": 2.491165035993977e-06,
"loss": 1.0700827836990356,
"step": 1956
},
{
"epoch": 2.242840778923253,
"grad_norm": 0.03502170741558075,
"learning_rate": 2.4827242130971822e-06,
"loss": 1.0163440704345703,
"step": 1958
},
{
"epoch": 2.245131729667812,
"grad_norm": 0.04055381193757057,
"learning_rate": 2.4743026332438835e-06,
"loss": 1.0671887397766113,
"step": 1960
},
{
"epoch": 2.247422680412371,
"grad_norm": 0.15526892244815826,
"learning_rate": 2.4659003501437302e-06,
"loss": 1.065315842628479,
"step": 1962
},
{
"epoch": 2.24971363115693,
"grad_norm": 0.10842100530862808,
"learning_rate": 2.4575174173833076e-06,
"loss": 1.0709346532821655,
"step": 1964
},
{
"epoch": 2.252004581901489,
"grad_norm": 0.1467980593442917,
"learning_rate": 2.449153888425786e-06,
"loss": 1.220167875289917,
"step": 1966
},
{
"epoch": 2.254295532646048,
"grad_norm": 0.0432235449552536,
"learning_rate": 2.4408098166105905e-06,
"loss": 1.1115278005599976,
"step": 1968
},
{
"epoch": 2.256586483390607,
"grad_norm": 0.07890545576810837,
"learning_rate": 2.4324852551530546e-06,
"loss": 1.2021079063415527,
"step": 1970
},
{
"epoch": 2.258877434135166,
"grad_norm": 0.10672356933355331,
"learning_rate": 2.424180257144082e-06,
"loss": 0.9553957581520081,
"step": 1972
},
{
"epoch": 2.261168384879725,
"grad_norm": 0.04205801710486412,
"learning_rate": 2.4158948755498097e-06,
"loss": 1.057405710220337,
"step": 1974
},
{
"epoch": 2.263459335624284,
"grad_norm": 0.03718230873346329,
"learning_rate": 2.407629163211264e-06,
"loss": 1.1039612293243408,
"step": 1976
},
{
"epoch": 2.265750286368843,
"grad_norm": 0.15723231434822083,
"learning_rate": 2.399383172844033e-06,
"loss": 1.0969688892364502,
"step": 1978
},
{
"epoch": 2.268041237113402,
"grad_norm": 0.1052173301577568,
"learning_rate": 2.3911569570379226e-06,
"loss": 0.9735721349716187,
"step": 1980
},
{
"epoch": 2.270332187857961,
"grad_norm": 0.28806793689727783,
"learning_rate": 2.3829505682566274e-06,
"loss": 0.8661947250366211,
"step": 1982
},
{
"epoch": 2.27262313860252,
"grad_norm": 0.043717917054891586,
"learning_rate": 2.374764058837388e-06,
"loss": 1.085542917251587,
"step": 1984
},
{
"epoch": 2.274914089347079,
"grad_norm": 0.11502456665039062,
"learning_rate": 2.366597480990663e-06,
"loss": 1.1323692798614502,
"step": 1986
},
{
"epoch": 2.277205040091638,
"grad_norm": 0.1218227744102478,
"learning_rate": 2.3584508867998023e-06,
"loss": 1.165204405784607,
"step": 1988
},
{
"epoch": 2.279495990836197,
"grad_norm": 0.04732652008533478,
"learning_rate": 2.350324328220697e-06,
"loss": 1.1347787380218506,
"step": 1990
},
{
"epoch": 2.281786941580756,
"grad_norm": 0.051126670092344284,
"learning_rate": 2.3422178570814656e-06,
"loss": 1.190324068069458,
"step": 1992
},
{
"epoch": 2.284077892325315,
"grad_norm": 0.03463529422879219,
"learning_rate": 2.3341315250821102e-06,
"loss": 0.9339711666107178,
"step": 1994
},
{
"epoch": 2.286368843069874,
"grad_norm": 0.18917769193649292,
"learning_rate": 2.326065383794202e-06,
"loss": 0.7177498936653137,
"step": 1996
},
{
"epoch": 2.288659793814433,
"grad_norm": 0.16513919830322266,
"learning_rate": 2.3180194846605367e-06,
"loss": 1.1773316860198975,
"step": 1998
},
{
"epoch": 2.290950744558992,
"grad_norm": 0.15770074725151062,
"learning_rate": 2.3099938789948147e-06,
"loss": 0.9601297378540039,
"step": 2000
},
{
"epoch": 2.293241695303551,
"grad_norm": 0.11050211638212204,
"learning_rate": 2.301988617981314e-06,
"loss": 0.625648021697998,
"step": 2002
},
{
"epoch": 2.29553264604811,
"grad_norm": 0.24079091846942902,
"learning_rate": 2.294003752674557e-06,
"loss": 0.862073540687561,
"step": 2004
},
{
"epoch": 2.297823596792669,
"grad_norm": 0.22309234738349915,
"learning_rate": 2.2860393339990005e-06,
"loss": 0.4852597117424011,
"step": 2006
},
{
"epoch": 2.3001145475372278,
"grad_norm": 0.05693449079990387,
"learning_rate": 2.2780954127486883e-06,
"loss": 1.1776564121246338,
"step": 2008
},
{
"epoch": 2.3024054982817868,
"grad_norm": 0.23189516365528107,
"learning_rate": 2.270172039586948e-06,
"loss": 0.8554977774620056,
"step": 2010
},
{
"epoch": 2.3046964490263457,
"grad_norm": 0.11551795154809952,
"learning_rate": 2.2622692650460516e-06,
"loss": 1.0636159181594849,
"step": 2012
},
{
"epoch": 2.3069873997709047,
"grad_norm": 0.03734152764081955,
"learning_rate": 2.25438713952691e-06,
"loss": 0.9173470735549927,
"step": 2014
},
{
"epoch": 2.3092783505154637,
"grad_norm": 0.10906286537647247,
"learning_rate": 2.2465257132987388e-06,
"loss": 0.8785110712051392,
"step": 2016
},
{
"epoch": 2.3115693012600227,
"grad_norm": 0.03761937841773033,
"learning_rate": 2.2386850364987357e-06,
"loss": 1.0721259117126465,
"step": 2018
},
{
"epoch": 2.3138602520045817,
"grad_norm": 0.15644510090351105,
"learning_rate": 2.230865159131774e-06,
"loss": 0.10478156059980392,
"step": 2020
},
{
"epoch": 2.3161512027491407,
"grad_norm": 0.18335962295532227,
"learning_rate": 2.2230661310700717e-06,
"loss": 1.056382179260254,
"step": 2022
},
{
"epoch": 2.3184421534936996,
"grad_norm": 0.3297891318798065,
"learning_rate": 2.215288002052882e-06,
"loss": 0.5145710706710815,
"step": 2024
},
{
"epoch": 2.320733104238259,
"grad_norm": 0.1070839911699295,
"learning_rate": 2.207530821686166e-06,
"loss": 1.017712950706482,
"step": 2026
},
{
"epoch": 2.323024054982818,
"grad_norm": 0.24218393862247467,
"learning_rate": 2.199794639442287e-06,
"loss": 0.7674975395202637,
"step": 2028
},
{
"epoch": 2.325315005727377,
"grad_norm": 0.15954658389091492,
"learning_rate": 2.1920795046596887e-06,
"loss": 1.3201217651367188,
"step": 2030
},
{
"epoch": 2.327605956471936,
"grad_norm": 0.12265370041131973,
"learning_rate": 2.1843854665425816e-06,
"loss": 0.5814427137374878,
"step": 2032
},
{
"epoch": 2.329896907216495,
"grad_norm": 0.10698456317186356,
"learning_rate": 2.176712574160632e-06,
"loss": 0.9436638355255127,
"step": 2034
},
{
"epoch": 2.332187857961054,
"grad_norm": 0.19457845389842987,
"learning_rate": 2.1690608764486417e-06,
"loss": 1.0933640003204346,
"step": 2036
},
{
"epoch": 2.334478808705613,
"grad_norm": 0.0642176941037178,
"learning_rate": 2.1614304222062447e-06,
"loss": 0.8278051614761353,
"step": 2038
},
{
"epoch": 2.336769759450172,
"grad_norm": 0.039123423397541046,
"learning_rate": 2.1538212600975927e-06,
"loss": 1.037153959274292,
"step": 2040
},
{
"epoch": 2.339060710194731,
"grad_norm": 0.03765871003270149,
"learning_rate": 2.146233438651042e-06,
"loss": 0.9489548206329346,
"step": 2042
},
{
"epoch": 2.34135166093929,
"grad_norm": 0.03357570245862007,
"learning_rate": 2.138667006258849e-06,
"loss": 0.5923043489456177,
"step": 2044
},
{
"epoch": 2.343642611683849,
"grad_norm": 0.1226658970117569,
"learning_rate": 2.131122011176854e-06,
"loss": 1.0477964878082275,
"step": 2046
},
{
"epoch": 2.345933562428408,
"grad_norm": 0.278041273355484,
"learning_rate": 2.1235985015241832e-06,
"loss": 0.8828730583190918,
"step": 2048
},
{
"epoch": 2.348224513172967,
"grad_norm": 0.039722733199596405,
"learning_rate": 2.1160965252829344e-06,
"loss": 1.1027134656906128,
"step": 2050
},
{
"epoch": 2.350515463917526,
"grad_norm": 0.2350229173898697,
"learning_rate": 2.108616130297876e-06,
"loss": 0.710883617401123,
"step": 2052
},
{
"epoch": 2.352806414662085,
"grad_norm": 0.10858850926160812,
"learning_rate": 2.101157364276134e-06,
"loss": 1.2099969387054443,
"step": 2054
},
{
"epoch": 2.355097365406644,
"grad_norm": 0.10385426133871078,
"learning_rate": 2.0937202747868974e-06,
"loss": 1.068502426147461,
"step": 2056
},
{
"epoch": 2.357388316151203,
"grad_norm": 0.12905098497867584,
"learning_rate": 2.08630490926111e-06,
"loss": 0.9399811029434204,
"step": 2058
},
{
"epoch": 2.359679266895762,
"grad_norm": 0.1345222443342209,
"learning_rate": 2.0789113149911678e-06,
"loss": 1.1330405473709106,
"step": 2060
},
{
"epoch": 2.361970217640321,
"grad_norm": 0.3326592445373535,
"learning_rate": 2.071539539130619e-06,
"loss": 0.8196538686752319,
"step": 2062
},
{
"epoch": 2.3642611683848798,
"grad_norm": 0.03900545835494995,
"learning_rate": 2.0641896286938586e-06,
"loss": 1.1343563795089722,
"step": 2064
},
{
"epoch": 2.3665521191294387,
"grad_norm": 0.30328232049942017,
"learning_rate": 2.0568616305558362e-06,
"loss": 0.8793677687644958,
"step": 2066
},
{
"epoch": 2.3688430698739977,
"grad_norm": 0.11676283180713654,
"learning_rate": 2.0495555914517517e-06,
"loss": 1.1062278747558594,
"step": 2068
},
{
"epoch": 2.3711340206185567,
"grad_norm": 0.11668711155653,
"learning_rate": 2.042271557976758e-06,
"loss": 1.1123981475830078,
"step": 2070
},
{
"epoch": 2.3734249713631157,
"grad_norm": 0.41015803813934326,
"learning_rate": 2.0350095765856674e-06,
"loss": 0.6100105047225952,
"step": 2072
},
{
"epoch": 2.3757159221076747,
"grad_norm": 0.04040496423840523,
"learning_rate": 2.0277696935926465e-06,
"loss": 1.1077494621276855,
"step": 2074
},
{
"epoch": 2.3780068728522337,
"grad_norm": 0.11139100790023804,
"learning_rate": 2.020551955170932e-06,
"loss": 0.9434921145439148,
"step": 2076
},
{
"epoch": 2.3802978235967927,
"grad_norm": 0.10817384719848633,
"learning_rate": 2.0133564073525283e-06,
"loss": 0.7552821040153503,
"step": 2078
},
{
"epoch": 2.3825887743413516,
"grad_norm": 0.038600143045186996,
"learning_rate": 2.0061830960279182e-06,
"loss": 1.0198276042938232,
"step": 2080
},
{
"epoch": 2.3848797250859106,
"grad_norm": 0.13652552664279938,
"learning_rate": 1.9990320669457664e-06,
"loss": 1.120629072189331,
"step": 2082
},
{
"epoch": 2.3871706758304696,
"grad_norm": 0.05311482027173042,
"learning_rate": 1.9919033657126297e-06,
"loss": 1.1022953987121582,
"step": 2084
},
{
"epoch": 2.3894616265750286,
"grad_norm": 0.1396331489086151,
"learning_rate": 1.9847970377926723e-06,
"loss": 1.1291303634643555,
"step": 2086
},
{
"epoch": 2.3917525773195876,
"grad_norm": 0.03835023194551468,
"learning_rate": 1.9777131285073608e-06,
"loss": 1.0857717990875244,
"step": 2088
},
{
"epoch": 2.3940435280641466,
"grad_norm": 0.06169874966144562,
"learning_rate": 1.9706516830351915e-06,
"loss": 1.230233907699585,
"step": 2090
},
{
"epoch": 2.3963344788087055,
"grad_norm": 0.03643700107932091,
"learning_rate": 1.963612746411389e-06,
"loss": 1.1737711429595947,
"step": 2092
},
{
"epoch": 2.3986254295532645,
"grad_norm": 0.039749957621097565,
"learning_rate": 1.9565963635276326e-06,
"loss": 1.073061466217041,
"step": 2094
},
{
"epoch": 2.4009163802978235,
"grad_norm": 0.13701081275939941,
"learning_rate": 1.949602579131754e-06,
"loss": 1.0454206466674805,
"step": 2096
},
{
"epoch": 2.4032073310423825,
"grad_norm": 0.1076841652393341,
"learning_rate": 1.9426314378274663e-06,
"loss": 1.114311695098877,
"step": 2098
},
{
"epoch": 2.4054982817869415,
"grad_norm": 0.3381866216659546,
"learning_rate": 1.9356829840740705e-06,
"loss": 0.5539225339889526,
"step": 2100
},
{
"epoch": 2.4077892325315005,
"grad_norm": 0.03752079978585243,
"learning_rate": 1.928757262186177e-06,
"loss": 0.5570148229598999,
"step": 2102
},
{
"epoch": 2.4100801832760594,
"grad_norm": 0.11114965379238129,
"learning_rate": 1.9218543163334198e-06,
"loss": 1.1547082662582397,
"step": 2104
},
{
"epoch": 2.4123711340206184,
"grad_norm": 0.13616393506526947,
"learning_rate": 1.914974190540174e-06,
"loss": 1.0737618207931519,
"step": 2106
},
{
"epoch": 2.4146620847651774,
"grad_norm": 0.034298207610845566,
"learning_rate": 1.9081169286852827e-06,
"loss": 0.9443414807319641,
"step": 2108
},
{
"epoch": 2.4169530355097364,
"grad_norm": 0.2994533181190491,
"learning_rate": 1.9012825745017633e-06,
"loss": 0.6153253316879272,
"step": 2110
},
{
"epoch": 2.4192439862542954,
"grad_norm": 0.041227761656045914,
"learning_rate": 1.8944711715765446e-06,
"loss": 1.2450883388519287,
"step": 2112
},
{
"epoch": 2.4215349369988544,
"grad_norm": 0.11088487505912781,
"learning_rate": 1.8876827633501784e-06,
"loss": 1.1472076177597046,
"step": 2114
},
{
"epoch": 2.4238258877434133,
"grad_norm": 0.14382129907608032,
"learning_rate": 1.8809173931165614e-06,
"loss": 0.9431512355804443,
"step": 2116
},
{
"epoch": 2.4261168384879723,
"grad_norm": 0.12498413771390915,
"learning_rate": 1.8741751040226663e-06,
"loss": 1.1251745223999023,
"step": 2118
},
{
"epoch": 2.4284077892325313,
"grad_norm": 0.13061866164207458,
"learning_rate": 1.8674559390682629e-06,
"loss": 1.1227977275848389,
"step": 2120
},
{
"epoch": 2.4306987399770907,
"grad_norm": 0.08055675029754639,
"learning_rate": 1.8607599411056444e-06,
"loss": 0.8968017101287842,
"step": 2122
},
{
"epoch": 2.4329896907216497,
"grad_norm": 0.1616377979516983,
"learning_rate": 1.8540871528393497e-06,
"loss": 1.0171732902526855,
"step": 2124
},
{
"epoch": 2.4352806414662087,
"grad_norm": 0.048232708126306534,
"learning_rate": 1.8474376168259003e-06,
"loss": 1.138751745223999,
"step": 2126
},
{
"epoch": 2.4375715922107677,
"grad_norm": 0.11059226095676422,
"learning_rate": 1.84081137547352e-06,
"loss": 1.104867696762085,
"step": 2128
},
{
"epoch": 2.4398625429553267,
"grad_norm": 0.11222375184297562,
"learning_rate": 1.83420847104187e-06,
"loss": 0.5774714946746826,
"step": 2130
},
{
"epoch": 2.4421534936998857,
"grad_norm": 0.03512409329414368,
"learning_rate": 1.8276289456417784e-06,
"loss": 1.068396806716919,
"step": 2132
},
{
"epoch": 2.4444444444444446,
"grad_norm": 0.042172499001026154,
"learning_rate": 1.821072841234966e-06,
"loss": 1.0330781936645508,
"step": 2134
},
{
"epoch": 2.4467353951890036,
"grad_norm": 0.11471384018659592,
"learning_rate": 1.8145401996337877e-06,
"loss": 1.0819417238235474,
"step": 2136
},
{
"epoch": 2.4490263459335626,
"grad_norm": 0.04197245091199875,
"learning_rate": 1.80803106250096e-06,
"loss": 0.9713804125785828,
"step": 2138
},
{
"epoch": 2.4513172966781216,
"grad_norm": 0.14753484725952148,
"learning_rate": 1.8015454713492985e-06,
"loss": 1.2071187496185303,
"step": 2140
},
{
"epoch": 2.4536082474226806,
"grad_norm": 0.12013762444257736,
"learning_rate": 1.7950834675414498e-06,
"loss": 1.0779211521148682,
"step": 2142
},
{
"epoch": 2.4558991981672396,
"grad_norm": 0.11798571050167084,
"learning_rate": 1.788645092289627e-06,
"loss": 1.126075029373169,
"step": 2144
},
{
"epoch": 2.4581901489117985,
"grad_norm": 0.13205040991306305,
"learning_rate": 1.7822303866553536e-06,
"loss": 1.0150024890899658,
"step": 2146
},
{
"epoch": 2.4604810996563575,
"grad_norm": 0.2663130760192871,
"learning_rate": 1.7758393915491957e-06,
"loss": 0.542580783367157,
"step": 2148
},
{
"epoch": 2.4627720504009165,
"grad_norm": 0.0403556264936924,
"learning_rate": 1.7694721477305026e-06,
"loss": 1.0786938667297363,
"step": 2150
},
{
"epoch": 2.4650630011454755,
"grad_norm": 0.11474873125553131,
"learning_rate": 1.7631286958071444e-06,
"loss": 1.1576069593429565,
"step": 2152
},
{
"epoch": 2.4673539518900345,
"grad_norm": 0.4626518189907074,
"learning_rate": 1.7568090762352591e-06,
"loss": 0.6024706363677979,
"step": 2154
},
{
"epoch": 2.4696449026345935,
"grad_norm": 0.10904034972190857,
"learning_rate": 1.7505133293189898e-06,
"loss": 1.116260290145874,
"step": 2156
},
{
"epoch": 2.4719358533791524,
"grad_norm": 0.5043454766273499,
"learning_rate": 1.7442414952102279e-06,
"loss": 0.6629682779312134,
"step": 2158
},
{
"epoch": 2.4742268041237114,
"grad_norm": 0.04634617269039154,
"learning_rate": 1.7379936139083604e-06,
"loss": 1.1367168426513672,
"step": 2160
},
{
"epoch": 2.4765177548682704,
"grad_norm": 0.12527769804000854,
"learning_rate": 1.731769725260008e-06,
"loss": 0.9263055324554443,
"step": 2162
},
{
"epoch": 2.4788087056128294,
"grad_norm": 0.0392335020005703,
"learning_rate": 1.7255698689587774e-06,
"loss": 1.012625813484192,
"step": 2164
},
{
"epoch": 2.4810996563573884,
"grad_norm": 0.1595272570848465,
"learning_rate": 1.719394084545011e-06,
"loss": 1.2080172300338745,
"step": 2166
},
{
"epoch": 2.4833906071019474,
"grad_norm": 0.04092966765165329,
"learning_rate": 1.7132424114055212e-06,
"loss": 1.0967365503311157,
"step": 2168
},
{
"epoch": 2.4856815578465064,
"grad_norm": 0.09347018599510193,
"learning_rate": 1.7071148887733552e-06,
"loss": 1.0474753379821777,
"step": 2170
},
{
"epoch": 2.4879725085910653,
"grad_norm": 0.042294181883335114,
"learning_rate": 1.7010115557275297e-06,
"loss": 0.9783978462219238,
"step": 2172
},
{
"epoch": 2.4902634593356243,
"grad_norm": 0.300382524728775,
"learning_rate": 1.6949324511927993e-06,
"loss": 0.6601334810256958,
"step": 2174
},
{
"epoch": 2.4925544100801833,
"grad_norm": 0.10989909619092941,
"learning_rate": 1.6888776139393892e-06,
"loss": 1.036874532699585,
"step": 2176
},
{
"epoch": 2.4948453608247423,
"grad_norm": 0.05132288485765457,
"learning_rate": 1.6828470825827626e-06,
"loss": 1.1650950908660889,
"step": 2178
},
{
"epoch": 2.4971363115693013,
"grad_norm": 0.09411272406578064,
"learning_rate": 1.6768408955833653e-06,
"loss": 1.0749032497406006,
"step": 2180
},
{
"epoch": 2.4994272623138603,
"grad_norm": 0.17432832717895508,
"learning_rate": 1.6708590912463878e-06,
"loss": 0.6857452392578125,
"step": 2182
},
{
"epoch": 2.5017182130584192,
"grad_norm": 0.1278747320175171,
"learning_rate": 1.6649017077215146e-06,
"loss": 1.136509656906128,
"step": 2184
},
{
"epoch": 2.504009163802978,
"grad_norm": 0.12037646770477295,
"learning_rate": 1.6589687830026835e-06,
"loss": 1.1849336624145508,
"step": 2186
},
{
"epoch": 2.506300114547537,
"grad_norm": 0.12011449784040451,
"learning_rate": 1.653060354927844e-06,
"loss": 1.1057056188583374,
"step": 2188
},
{
"epoch": 2.508591065292096,
"grad_norm": 0.2916851341724396,
"learning_rate": 1.647176461178714e-06,
"loss": 0.5272727012634277,
"step": 2190
},
{
"epoch": 2.510882016036655,
"grad_norm": 0.20276512205600739,
"learning_rate": 1.6413171392805444e-06,
"loss": 0.8722376823425293,
"step": 2192
},
{
"epoch": 2.513172966781214,
"grad_norm": 0.11439058929681778,
"learning_rate": 1.6354824266018726e-06,
"loss": 1.018531322479248,
"step": 2194
},
{
"epoch": 2.515463917525773,
"grad_norm": 0.03520492836833,
"learning_rate": 1.6296723603542874e-06,
"loss": 0.988182783126831,
"step": 2196
},
{
"epoch": 2.517754868270332,
"grad_norm": 0.03757385164499283,
"learning_rate": 1.6238869775921947e-06,
"loss": 1.0032260417938232,
"step": 2198
},
{
"epoch": 2.520045819014891,
"grad_norm": 0.11876486241817474,
"learning_rate": 1.6181263152125761e-06,
"loss": 1.1592516899108887,
"step": 2200
},
{
"epoch": 2.52233676975945,
"grad_norm": 0.10025777667760849,
"learning_rate": 1.6123904099547593e-06,
"loss": 0.6935728788375854,
"step": 2202
},
{
"epoch": 2.524627720504009,
"grad_norm": 0.11772006005048752,
"learning_rate": 1.606679298400175e-06,
"loss": 1.0764268636703491,
"step": 2204
},
{
"epoch": 2.526918671248568,
"grad_norm": 0.22523917257785797,
"learning_rate": 1.6009930169721332e-06,
"loss": 0.3999437093734741,
"step": 2206
},
{
"epoch": 2.529209621993127,
"grad_norm": 0.03774823993444443,
"learning_rate": 1.595331601935586e-06,
"loss": 0.9670888185501099,
"step": 2208
},
{
"epoch": 2.531500572737686,
"grad_norm": 0.041391998529434204,
"learning_rate": 1.5896950893968968e-06,
"loss": 1.1194099187850952,
"step": 2210
},
{
"epoch": 2.533791523482245,
"grad_norm": 0.12182236462831497,
"learning_rate": 1.584083515303611e-06,
"loss": 0.6618677377700806,
"step": 2212
},
{
"epoch": 2.536082474226804,
"grad_norm": 0.047492437064647675,
"learning_rate": 1.5784969154442228e-06,
"loss": 1.135564923286438,
"step": 2214
},
{
"epoch": 2.538373424971363,
"grad_norm": 0.12321395426988602,
"learning_rate": 1.5729353254479544e-06,
"loss": 0.9604895710945129,
"step": 2216
},
{
"epoch": 2.540664375715922,
"grad_norm": 0.26825836300849915,
"learning_rate": 1.5673987807845224e-06,
"loss": 0.20692116022109985,
"step": 2218
},
{
"epoch": 2.542955326460481,
"grad_norm": 0.03937020152807236,
"learning_rate": 1.561887316763914e-06,
"loss": 0.7779680490493774,
"step": 2220
},
{
"epoch": 2.54524627720504,
"grad_norm": 0.12153687328100204,
"learning_rate": 1.5564009685361631e-06,
"loss": 1.1622121334075928,
"step": 2222
},
{
"epoch": 2.547537227949599,
"grad_norm": 0.048881255090236664,
"learning_rate": 1.5509397710911206e-06,
"loss": 1.0526154041290283,
"step": 2224
},
{
"epoch": 2.549828178694158,
"grad_norm": 0.11527767777442932,
"learning_rate": 1.545503759258239e-06,
"loss": 0.9147733449935913,
"step": 2226
},
{
"epoch": 2.552119129438717,
"grad_norm": 0.06676455587148666,
"learning_rate": 1.5400929677063453e-06,
"loss": 0.990266740322113,
"step": 2228
},
{
"epoch": 2.554410080183276,
"grad_norm": 0.0386880524456501,
"learning_rate": 1.534707430943422e-06,
"loss": 1.051877498626709,
"step": 2230
},
{
"epoch": 2.556701030927835,
"grad_norm": 0.037052009254693985,
"learning_rate": 1.5293471833163836e-06,
"loss": 0.7502873539924622,
"step": 2232
},
{
"epoch": 2.558991981672394,
"grad_norm": 0.12356662005186081,
"learning_rate": 1.5240122590108615e-06,
"loss": 1.1431405544281006,
"step": 2234
},
{
"epoch": 2.561282932416953,
"grad_norm": 0.10430527478456497,
"learning_rate": 1.5187026920509857e-06,
"loss": 0.8166168928146362,
"step": 2236
},
{
"epoch": 2.563573883161512,
"grad_norm": 0.04083002358675003,
"learning_rate": 1.5134185162991653e-06,
"loss": 1.0520750284194946,
"step": 2238
},
{
"epoch": 2.565864833906071,
"grad_norm": 0.10489920526742935,
"learning_rate": 1.5081597654558753e-06,
"loss": 1.0503392219543457,
"step": 2240
},
{
"epoch": 2.5681557846506298,
"grad_norm": 0.15466554462909698,
"learning_rate": 1.5029264730594366e-06,
"loss": 1.0287251472473145,
"step": 2242
},
{
"epoch": 2.5704467353951888,
"grad_norm": 0.15002210438251495,
"learning_rate": 1.49771867248581e-06,
"loss": 1.1313282251358032,
"step": 2244
},
{
"epoch": 2.5727376861397477,
"grad_norm": 0.03917040675878525,
"learning_rate": 1.4925363969483764e-06,
"loss": 1.2097923755645752,
"step": 2246
},
{
"epoch": 2.5750286368843067,
"grad_norm": 0.036840129643678665,
"learning_rate": 1.48737967949773e-06,
"loss": 1.0906928777694702,
"step": 2248
},
{
"epoch": 2.5773195876288657,
"grad_norm": 0.03971611335873604,
"learning_rate": 1.4822485530214636e-06,
"loss": 0.9993399977684021,
"step": 2250
},
{
"epoch": 2.579610538373425,
"grad_norm": 0.12046768516302109,
"learning_rate": 1.4771430502439585e-06,
"loss": 1.058572769165039,
"step": 2252
},
{
"epoch": 2.581901489117984,
"grad_norm": 0.21793264150619507,
"learning_rate": 1.4720632037261837e-06,
"loss": 1.0307724475860596,
"step": 2254
},
{
"epoch": 2.584192439862543,
"grad_norm": 0.040577035397291183,
"learning_rate": 1.4670090458654752e-06,
"loss": 1.0201236009597778,
"step": 2256
},
{
"epoch": 2.586483390607102,
"grad_norm": 0.08926853537559509,
"learning_rate": 1.4619806088953423e-06,
"loss": 1.028358817100525,
"step": 2258
},
{
"epoch": 2.588774341351661,
"grad_norm": 0.05179933086037636,
"learning_rate": 1.4569779248852514e-06,
"loss": 1.0471670627593994,
"step": 2260
},
{
"epoch": 2.59106529209622,
"grad_norm": 0.12112816423177719,
"learning_rate": 1.4520010257404282e-06,
"loss": 1.0303044319152832,
"step": 2262
},
{
"epoch": 2.593356242840779,
"grad_norm": 0.13048020005226135,
"learning_rate": 1.4470499432016564e-06,
"loss": 1.023011565208435,
"step": 2264
},
{
"epoch": 2.595647193585338,
"grad_norm": 0.20969820022583008,
"learning_rate": 1.4421247088450638e-06,
"loss": 0.3655018210411072,
"step": 2266
},
{
"epoch": 2.597938144329897,
"grad_norm": 0.15613162517547607,
"learning_rate": 1.4372253540819342e-06,
"loss": 1.1912338733673096,
"step": 2268
},
{
"epoch": 2.600229095074456,
"grad_norm": 0.036865267902612686,
"learning_rate": 1.4323519101584956e-06,
"loss": 1.14031982421875,
"step": 2270
},
{
"epoch": 2.602520045819015,
"grad_norm": 0.0848236158490181,
"learning_rate": 1.4275044081557336e-06,
"loss": 0.9510655999183655,
"step": 2272
},
{
"epoch": 2.604810996563574,
"grad_norm": 0.05090273916721344,
"learning_rate": 1.422682878989179e-06,
"loss": 1.06699538230896,
"step": 2274
},
{
"epoch": 2.607101947308133,
"grad_norm": 0.11556492745876312,
"learning_rate": 1.4178873534087224e-06,
"loss": 1.1918110847473145,
"step": 2276
},
{
"epoch": 2.609392898052692,
"grad_norm": 0.14172205328941345,
"learning_rate": 1.413117861998411e-06,
"loss": 0.5816162824630737,
"step": 2278
},
{
"epoch": 2.611683848797251,
"grad_norm": 0.0312827005982399,
"learning_rate": 1.4083744351762568e-06,
"loss": 1.0428073406219482,
"step": 2280
},
{
"epoch": 2.61397479954181,
"grad_norm": 0.035471122711896896,
"learning_rate": 1.4036571031940433e-06,
"loss": 0.9467089176177979,
"step": 2282
},
{
"epoch": 2.616265750286369,
"grad_norm": 0.21262601017951965,
"learning_rate": 1.398965896137126e-06,
"loss": 0.6726322174072266,
"step": 2284
},
{
"epoch": 2.618556701030928,
"grad_norm": 0.060128457844257355,
"learning_rate": 1.394300843924251e-06,
"loss": 1.120849847793579,
"step": 2286
},
{
"epoch": 2.620847651775487,
"grad_norm": 0.0908019170165062,
"learning_rate": 1.3896619763073558e-06,
"loss": 0.8834356069564819,
"step": 2288
},
{
"epoch": 2.623138602520046,
"grad_norm": 0.042698655277490616,
"learning_rate": 1.3850493228713829e-06,
"loss": 1.11699640750885,
"step": 2290
},
{
"epoch": 2.625429553264605,
"grad_norm": 0.11093872040510178,
"learning_rate": 1.3804629130340922e-06,
"loss": 0.8252977132797241,
"step": 2292
},
{
"epoch": 2.627720504009164,
"grad_norm": 0.4029269516468048,
"learning_rate": 1.375902776045869e-06,
"loss": 0.8700896501541138,
"step": 2294
},
{
"epoch": 2.6300114547537228,
"grad_norm": 0.4993189871311188,
"learning_rate": 1.3713689409895423e-06,
"loss": 0.9152309894561768,
"step": 2296
},
{
"epoch": 2.6323024054982818,
"grad_norm": 0.05592384934425354,
"learning_rate": 1.3668614367801968e-06,
"loss": 1.0523474216461182,
"step": 2298
},
{
"epoch": 2.6345933562428407,
"grad_norm": 0.26710832118988037,
"learning_rate": 1.3623802921649903e-06,
"loss": 1.022230625152588,
"step": 2300
},
{
"epoch": 2.6368843069873997,
"grad_norm": 0.05241448059678078,
"learning_rate": 1.3579255357229665e-06,
"loss": 1.0288628339767456,
"step": 2302
},
{
"epoch": 2.6391752577319587,
"grad_norm": 0.11045310646295547,
"learning_rate": 1.3534971958648763e-06,
"loss": 1.0415080785751343,
"step": 2304
},
{
"epoch": 2.6414662084765177,
"grad_norm": 0.1546574980020523,
"learning_rate": 1.3490953008329976e-06,
"loss": 1.096618890762329,
"step": 2306
},
{
"epoch": 2.6437571592210767,
"grad_norm": 0.1845928579568863,
"learning_rate": 1.34471987870095e-06,
"loss": 1.0915133953094482,
"step": 2308
},
{
"epoch": 2.6460481099656357,
"grad_norm": 0.11465723067522049,
"learning_rate": 1.3403709573735226e-06,
"loss": 1.0728447437286377,
"step": 2310
},
{
"epoch": 2.6483390607101946,
"grad_norm": 0.13408894836902618,
"learning_rate": 1.336048564586489e-06,
"loss": 1.1959388256072998,
"step": 2312
},
{
"epoch": 2.6506300114547536,
"grad_norm": 0.11020931601524353,
"learning_rate": 1.3317527279064346e-06,
"loss": 0.9788947105407715,
"step": 2314
},
{
"epoch": 2.6529209621993126,
"grad_norm": 0.056901715695858,
"learning_rate": 1.3274834747305816e-06,
"loss": 1.119939923286438,
"step": 2316
},
{
"epoch": 2.6552119129438716,
"grad_norm": 0.043660301715135574,
"learning_rate": 1.3232408322866112e-06,
"loss": 1.2462071180343628,
"step": 2318
},
{
"epoch": 2.6575028636884306,
"grad_norm": 0.12008506804704666,
"learning_rate": 1.319024827632493e-06,
"loss": 1.1180698871612549,
"step": 2320
},
{
"epoch": 2.6597938144329896,
"grad_norm": 0.03617572784423828,
"learning_rate": 1.3148354876563078e-06,
"loss": 0.8943448066711426,
"step": 2322
},
{
"epoch": 2.6620847651775486,
"grad_norm": 0.1361423134803772,
"learning_rate": 1.3106728390760822e-06,
"loss": 0.9939253330230713,
"step": 2324
},
{
"epoch": 2.6643757159221075,
"grad_norm": 0.12970031797885895,
"learning_rate": 1.3065369084396143e-06,
"loss": 1.1329681873321533,
"step": 2326
},
{
"epoch": 2.6666666666666665,
"grad_norm": 0.04504101350903511,
"learning_rate": 1.302427722124305e-06,
"loss": 1.0876107215881348,
"step": 2328
},
{
"epoch": 2.6689576174112255,
"grad_norm": 0.043641116470098495,
"learning_rate": 1.2983453063369886e-06,
"loss": 1.1243267059326172,
"step": 2330
},
{
"epoch": 2.6712485681557845,
"grad_norm": 0.12474491447210312,
"learning_rate": 1.2942896871137703e-06,
"loss": 0.150488942861557,
"step": 2332
},
{
"epoch": 2.673539518900344,
"grad_norm": 0.21792425215244293,
"learning_rate": 1.290260890319854e-06,
"loss": 1.049038290977478,
"step": 2334
},
{
"epoch": 2.675830469644903,
"grad_norm": 0.24270223081111908,
"learning_rate": 1.2862589416493828e-06,
"loss": 1.2486166954040527,
"step": 2336
},
{
"epoch": 2.678121420389462,
"grad_norm": 0.121235691010952,
"learning_rate": 1.282283866625271e-06,
"loss": 1.1332775354385376,
"step": 2338
},
{
"epoch": 2.680412371134021,
"grad_norm": 0.16410396993160248,
"learning_rate": 1.278335690599042e-06,
"loss": 1.1936967372894287,
"step": 2340
},
{
"epoch": 2.68270332187858,
"grad_norm": 0.12893368303775787,
"learning_rate": 1.27441443875067e-06,
"loss": 1.167176365852356,
"step": 2342
},
{
"epoch": 2.684994272623139,
"grad_norm": 0.7378900051116943,
"learning_rate": 1.2705201360884154e-06,
"loss": 0.5444666147232056,
"step": 2344
},
{
"epoch": 2.687285223367698,
"grad_norm": 0.14510443806648254,
"learning_rate": 1.2666528074486675e-06,
"loss": 1.0824750661849976,
"step": 2346
},
{
"epoch": 2.689576174112257,
"grad_norm": 0.10740603506565094,
"learning_rate": 1.2628124774957862e-06,
"loss": 1.043527364730835,
"step": 2348
},
{
"epoch": 2.691867124856816,
"grad_norm": 0.1193559393286705,
"learning_rate": 1.2589991707219407e-06,
"loss": 0.9171677827835083,
"step": 2350
},
{
"epoch": 2.6941580756013748,
"grad_norm": 0.04034031182527542,
"learning_rate": 1.255212911446962e-06,
"loss": 1.0557785034179688,
"step": 2352
},
{
"epoch": 2.6964490263459338,
"grad_norm": 0.09817221015691757,
"learning_rate": 1.2514537238181764e-06,
"loss": 0.9150309562683105,
"step": 2354
},
{
"epoch": 2.6987399770904927,
"grad_norm": 0.04312164708971977,
"learning_rate": 1.2477216318102638e-06,
"loss": 1.1289045810699463,
"step": 2356
},
{
"epoch": 2.7010309278350517,
"grad_norm": 0.13590708374977112,
"learning_rate": 1.2440166592250908e-06,
"loss": 1.1207735538482666,
"step": 2358
},
{
"epoch": 2.7033218785796107,
"grad_norm": 0.11693116277456284,
"learning_rate": 1.2403388296915748e-06,
"loss": 0.9732823371887207,
"step": 2360
},
{
"epoch": 2.7056128293241697,
"grad_norm": 0.09531773626804352,
"learning_rate": 1.2366881666655197e-06,
"loss": 1.0172736644744873,
"step": 2362
},
{
"epoch": 2.7079037800687287,
"grad_norm": 0.23191611468791962,
"learning_rate": 1.2330646934294733e-06,
"loss": 0.7734839916229248,
"step": 2364
},
{
"epoch": 2.7101947308132877,
"grad_norm": 0.13292023539543152,
"learning_rate": 1.2294684330925773e-06,
"loss": 1.0921151638031006,
"step": 2366
},
{
"epoch": 2.7124856815578466,
"grad_norm": 0.03602350875735283,
"learning_rate": 1.225899408590418e-06,
"loss": 1.0671606063842773,
"step": 2368
},
{
"epoch": 2.7147766323024056,
"grad_norm": 0.1325361579656601,
"learning_rate": 1.222357642684886e-06,
"loss": 0.9317104816436768,
"step": 2370
},
{
"epoch": 2.7170675830469646,
"grad_norm": 0.02299438789486885,
"learning_rate": 1.2188431579640229e-06,
"loss": 0.5671766996383667,
"step": 2372
},
{
"epoch": 2.7193585337915236,
"grad_norm": 0.1803787499666214,
"learning_rate": 1.2153559768418826e-06,
"loss": 1.208252191543579,
"step": 2374
},
{
"epoch": 2.7216494845360826,
"grad_norm": 0.10567279905080795,
"learning_rate": 1.2118961215583876e-06,
"loss": 0.9299669861793518,
"step": 2376
},
{
"epoch": 2.7239404352806416,
"grad_norm": 0.1304902285337448,
"learning_rate": 1.2084636141791853e-06,
"loss": 1.1629877090454102,
"step": 2378
},
{
"epoch": 2.7262313860252005,
"grad_norm": 0.11300843209028244,
"learning_rate": 1.2050584765955106e-06,
"loss": 0.5608739852905273,
"step": 2380
},
{
"epoch": 2.7285223367697595,
"grad_norm": 0.2323353886604309,
"learning_rate": 1.2016807305240418e-06,
"loss": 1.0366477966308594,
"step": 2382
},
{
"epoch": 2.7308132875143185,
"grad_norm": 0.1072358712553978,
"learning_rate": 1.198330397506765e-06,
"loss": 1.0595978498458862,
"step": 2384
},
{
"epoch": 2.7331042382588775,
"grad_norm": 0.16197116672992706,
"learning_rate": 1.1950074989108369e-06,
"loss": 1.25881028175354,
"step": 2386
},
{
"epoch": 2.7353951890034365,
"grad_norm": 0.042508866637945175,
"learning_rate": 1.1917120559284484e-06,
"loss": 1.033616304397583,
"step": 2388
},
{
"epoch": 2.7376861397479955,
"grad_norm": 0.16776804625988007,
"learning_rate": 1.188444089576689e-06,
"loss": 1.1445398330688477,
"step": 2390
},
{
"epoch": 2.7399770904925544,
"grad_norm": 0.040681999176740646,
"learning_rate": 1.1852036206974093e-06,
"loss": 1.0716893672943115,
"step": 2392
},
{
"epoch": 2.7422680412371134,
"grad_norm": 0.03599237650632858,
"learning_rate": 1.1819906699570951e-06,
"loss": 1.0026639699935913,
"step": 2394
},
{
"epoch": 2.7445589919816724,
"grad_norm": 0.20803894102573395,
"learning_rate": 1.178805257846731e-06,
"loss": 0.8390488624572754,
"step": 2396
},
{
"epoch": 2.7468499427262314,
"grad_norm": 0.045633960515260696,
"learning_rate": 1.1756474046816705e-06,
"loss": 1.0751760005950928,
"step": 2398
},
{
"epoch": 2.7491408934707904,
"grad_norm": 0.1319274753332138,
"learning_rate": 1.172517130601506e-06,
"loss": 0.9296597838401794,
"step": 2400
},
{
"epoch": 2.7514318442153494,
"grad_norm": 0.13140706717967987,
"learning_rate": 1.169414455569941e-06,
"loss": 1.083257794380188,
"step": 2402
},
{
"epoch": 2.7537227949599083,
"grad_norm": 0.10441871732473373,
"learning_rate": 1.1663393993746639e-06,
"loss": 0.70625901222229,
"step": 2404
},
{
"epoch": 2.7560137457044673,
"grad_norm": 0.21324773132801056,
"learning_rate": 1.1632919816272185e-06,
"loss": 1.1939747333526611,
"step": 2406
},
{
"epoch": 2.7583046964490263,
"grad_norm": 0.15752872824668884,
"learning_rate": 1.1602722217628847e-06,
"loss": 0.5383874177932739,
"step": 2408
},
{
"epoch": 2.7605956471935853,
"grad_norm": 0.04424149543046951,
"learning_rate": 1.157280139040546e-06,
"loss": 1.1126903295516968,
"step": 2410
},
{
"epoch": 2.7628865979381443,
"grad_norm": 0.11697188764810562,
"learning_rate": 1.1543157525425753e-06,
"loss": 1.0950360298156738,
"step": 2412
},
{
"epoch": 2.7651775486827033,
"grad_norm": 0.12432119995355606,
"learning_rate": 1.1513790811747077e-06,
"loss": 1.031628966331482,
"step": 2414
},
{
"epoch": 2.7674684994272623,
"grad_norm": 0.1950952112674713,
"learning_rate": 1.1484701436659228e-06,
"loss": 0.9780592322349548,
"step": 2416
},
{
"epoch": 2.7697594501718212,
"grad_norm": 0.10742440074682236,
"learning_rate": 1.1455889585683234e-06,
"loss": 1.1695256233215332,
"step": 2418
},
{
"epoch": 2.77205040091638,
"grad_norm": 0.03295076638460159,
"learning_rate": 1.1427355442570174e-06,
"loss": 0.9007450342178345,
"step": 2420
},
{
"epoch": 2.774341351660939,
"grad_norm": 0.12205695360898972,
"learning_rate": 1.1399099189300022e-06,
"loss": 1.1622676849365234,
"step": 2422
},
{
"epoch": 2.776632302405498,
"grad_norm": 0.12269830703735352,
"learning_rate": 1.1371121006080476e-06,
"loss": 0.6331859827041626,
"step": 2424
},
{
"epoch": 2.778923253150057,
"grad_norm": 0.12496553361415863,
"learning_rate": 1.1343421071345807e-06,
"loss": 0.8568001985549927,
"step": 2426
},
{
"epoch": 2.781214203894616,
"grad_norm": 0.13641129434108734,
"learning_rate": 1.1315999561755704e-06,
"loss": 1.0987757444381714,
"step": 2428
},
{
"epoch": 2.783505154639175,
"grad_norm": 0.2431728094816208,
"learning_rate": 1.1288856652194187e-06,
"loss": 1.1993837356567383,
"step": 2430
},
{
"epoch": 2.785796105383734,
"grad_norm": 0.1257048398256302,
"learning_rate": 1.1261992515768475e-06,
"loss": 1.093268871307373,
"step": 2432
},
{
"epoch": 2.788087056128293,
"grad_norm": 0.11494007706642151,
"learning_rate": 1.1235407323807863e-06,
"loss": 0.9766872525215149,
"step": 2434
},
{
"epoch": 2.790378006872852,
"grad_norm": 0.12698274850845337,
"learning_rate": 1.1209101245862641e-06,
"loss": 1.0670726299285889,
"step": 2436
},
{
"epoch": 2.792668957617411,
"grad_norm": 0.1961643397808075,
"learning_rate": 1.1183074449703026e-06,
"loss": 1.1611640453338623,
"step": 2438
},
{
"epoch": 2.79495990836197,
"grad_norm": 0.06117239221930504,
"learning_rate": 1.1157327101318079e-06,
"loss": 1.0661375522613525,
"step": 2440
},
{
"epoch": 2.797250859106529,
"grad_norm": 0.07389727979898453,
"learning_rate": 1.113185936491464e-06,
"loss": 0.653728723526001,
"step": 2442
},
{
"epoch": 2.799541809851088,
"grad_norm": 0.1278214305639267,
"learning_rate": 1.1106671402916311e-06,
"loss": 1.077947974205017,
"step": 2444
},
{
"epoch": 2.801832760595647,
"grad_norm": 0.11121320724487305,
"learning_rate": 1.1081763375962376e-06,
"loss": 1.074617862701416,
"step": 2446
},
{
"epoch": 2.804123711340206,
"grad_norm": 0.13351018726825714,
"learning_rate": 1.1057135442906802e-06,
"loss": 1.0540236234664917,
"step": 2448
},
{
"epoch": 2.806414662084765,
"grad_norm": 0.04395301640033722,
"learning_rate": 1.1032787760817245e-06,
"loss": 1.0289393663406372,
"step": 2450
},
{
"epoch": 2.808705612829324,
"grad_norm": 0.11833327263593674,
"learning_rate": 1.1008720484973997e-06,
"loss": 1.0666475296020508,
"step": 2452
},
{
"epoch": 2.810996563573883,
"grad_norm": 0.12046226114034653,
"learning_rate": 1.0984933768869045e-06,
"loss": 0.9002621173858643,
"step": 2454
},
{
"epoch": 2.813287514318442,
"grad_norm": 0.2076469361782074,
"learning_rate": 1.0961427764205054e-06,
"loss": 0.9490566253662109,
"step": 2456
},
{
"epoch": 2.815578465063001,
"grad_norm": 0.16561594605445862,
"learning_rate": 1.0938202620894444e-06,
"loss": 1.0374698638916016,
"step": 2458
},
{
"epoch": 2.81786941580756,
"grad_norm": 0.039882734417915344,
"learning_rate": 1.0915258487058393e-06,
"loss": 0.9988796710968018,
"step": 2460
},
{
"epoch": 2.820160366552119,
"grad_norm": 0.11536919325590134,
"learning_rate": 1.0892595509025897e-06,
"loss": 1.0317668914794922,
"step": 2462
},
{
"epoch": 2.822451317296678,
"grad_norm": 0.12474019825458527,
"learning_rate": 1.0870213831332856e-06,
"loss": 1.0597889423370361,
"step": 2464
},
{
"epoch": 2.824742268041237,
"grad_norm": 0.10003583133220673,
"learning_rate": 1.0848113596721147e-06,
"loss": 0.9430271983146667,
"step": 2466
},
{
"epoch": 2.827033218785796,
"grad_norm": 0.11832843720912933,
"learning_rate": 1.0826294946137705e-06,
"loss": 1.1333792209625244,
"step": 2468
},
{
"epoch": 2.8293241695303553,
"grad_norm": 0.12214350700378418,
"learning_rate": 1.0804758018733635e-06,
"loss": 1.1652965545654297,
"step": 2470
},
{
"epoch": 2.8316151202749142,
"grad_norm": 0.3238823711872101,
"learning_rate": 1.0783502951863302e-06,
"loss": 0.739303469657898,
"step": 2472
},
{
"epoch": 2.8339060710194732,
"grad_norm": 0.12782305479049683,
"learning_rate": 1.076252988108349e-06,
"loss": 1.1109859943389893,
"step": 2474
},
{
"epoch": 2.836197021764032,
"grad_norm": 0.12479905039072037,
"learning_rate": 1.0741838940152498e-06,
"loss": 1.060528039932251,
"step": 2476
},
{
"epoch": 2.838487972508591,
"grad_norm": 0.03563203662633896,
"learning_rate": 1.0721430261029334e-06,
"loss": 1.0834589004516602,
"step": 2478
},
{
"epoch": 2.84077892325315,
"grad_norm": 0.11884791404008865,
"learning_rate": 1.0701303973872825e-06,
"loss": 1.1230555772781372,
"step": 2480
},
{
"epoch": 2.843069873997709,
"grad_norm": 0.03811972215771675,
"learning_rate": 1.0681460207040816e-06,
"loss": 0.9433954358100891,
"step": 2482
},
{
"epoch": 2.845360824742268,
"grad_norm": 0.1846611499786377,
"learning_rate": 1.066189908708935e-06,
"loss": 1.1506900787353516,
"step": 2484
},
{
"epoch": 2.847651775486827,
"grad_norm": 0.1252901256084442,
"learning_rate": 1.0642620738771844e-06,
"loss": 1.1932969093322754,
"step": 2486
},
{
"epoch": 2.849942726231386,
"grad_norm": 0.03945187106728554,
"learning_rate": 1.0623625285038317e-06,
"loss": 1.2127320766448975,
"step": 2488
},
{
"epoch": 2.852233676975945,
"grad_norm": 0.12211194634437561,
"learning_rate": 1.0604912847034579e-06,
"loss": 1.1592764854431152,
"step": 2490
},
{
"epoch": 2.854524627720504,
"grad_norm": 0.03981486335396767,
"learning_rate": 1.0586483544101484e-06,
"loss": 0.9723308682441711,
"step": 2492
},
{
"epoch": 2.856815578465063,
"grad_norm": 0.03858442232012749,
"learning_rate": 1.0568337493774153e-06,
"loss": 1.0726711750030518,
"step": 2494
},
{
"epoch": 2.859106529209622,
"grad_norm": 0.11059273779392242,
"learning_rate": 1.055047481178123e-06,
"loss": 1.244250774383545,
"step": 2496
},
{
"epoch": 2.861397479954181,
"grad_norm": 0.149624302983284,
"learning_rate": 1.0532895612044148e-06,
"loss": 1.2564786672592163,
"step": 2498
},
{
"epoch": 2.86368843069874,
"grad_norm": 0.15256492793560028,
"learning_rate": 1.0515600006676388e-06,
"loss": 1.055551290512085,
"step": 2500
},
{
"epoch": 2.865979381443299,
"grad_norm": 0.0374947264790535,
"learning_rate": 1.0498588105982773e-06,
"loss": 1.071763515472412,
"step": 2502
},
{
"epoch": 2.868270332187858,
"grad_norm": 0.11409316956996918,
"learning_rate": 1.0481860018458777e-06,
"loss": 1.0046906471252441,
"step": 2504
},
{
"epoch": 2.870561282932417,
"grad_norm": 0.1608385592699051,
"learning_rate": 1.0465415850789818e-06,
"loss": 1.0260601043701172,
"step": 2506
},
{
"epoch": 2.872852233676976,
"grad_norm": 0.1228945255279541,
"learning_rate": 1.044925570785057e-06,
"loss": 1.1091501712799072,
"step": 2508
},
{
"epoch": 2.875143184421535,
"grad_norm": 0.23849347233772278,
"learning_rate": 1.0433379692704303e-06,
"loss": 0.8928807377815247,
"step": 2510
},
{
"epoch": 2.877434135166094,
"grad_norm": 0.05072511360049248,
"learning_rate": 1.0417787906602254e-06,
"loss": 1.1910967826843262,
"step": 2512
},
{
"epoch": 2.879725085910653,
"grad_norm": 0.11455868184566498,
"learning_rate": 1.0402480448982916e-06,
"loss": 1.0570517778396606,
"step": 2514
},
{
"epoch": 2.882016036655212,
"grad_norm": 0.3516460657119751,
"learning_rate": 1.0387457417471484e-06,
"loss": 0.6423954963684082,
"step": 2516
},
{
"epoch": 2.884306987399771,
"grad_norm": 0.06222286447882652,
"learning_rate": 1.0372718907879148e-06,
"loss": 0.615919828414917,
"step": 2518
},
{
"epoch": 2.88659793814433,
"grad_norm": 0.11802791804075241,
"learning_rate": 1.0358265014202562e-06,
"loss": 1.0918095111846924,
"step": 2520
},
{
"epoch": 2.888888888888889,
"grad_norm": 0.13629449903964996,
"learning_rate": 1.0344095828623187e-06,
"loss": 1.1885225772857666,
"step": 2522
},
{
"epoch": 2.891179839633448,
"grad_norm": 0.1508660912513733,
"learning_rate": 1.0330211441506725e-06,
"loss": 1.2527198791503906,
"step": 2524
},
{
"epoch": 2.893470790378007,
"grad_norm": 0.3050176799297333,
"learning_rate": 1.0316611941402554e-06,
"loss": 1.289149284362793,
"step": 2526
},
{
"epoch": 2.895761741122566,
"grad_norm": 0.03513256832957268,
"learning_rate": 1.0303297415043125e-06,
"loss": 1.0728578567504883,
"step": 2528
},
{
"epoch": 2.8980526918671248,
"grad_norm": 0.13335010409355164,
"learning_rate": 1.0290267947343467e-06,
"loss": 0.550873875617981,
"step": 2530
},
{
"epoch": 2.9003436426116838,
"grad_norm": 0.10280898958444595,
"learning_rate": 1.027752362140058e-06,
"loss": 0.9256083965301514,
"step": 2532
},
{
"epoch": 2.9026345933562427,
"grad_norm": 0.24355080723762512,
"learning_rate": 1.0265064518492971e-06,
"loss": 0.6109650135040283,
"step": 2534
},
{
"epoch": 2.9049255441008017,
"grad_norm": 0.11080548912286758,
"learning_rate": 1.0252890718080065e-06,
"loss": 0.9810810089111328,
"step": 2536
},
{
"epoch": 2.9072164948453607,
"grad_norm": 0.23204368352890015,
"learning_rate": 1.024100229780177e-06,
"loss": 0.7867921590805054,
"step": 2538
},
{
"epoch": 2.9095074455899197,
"grad_norm": 0.11744631826877594,
"learning_rate": 1.0229399333477933e-06,
"loss": 1.1211484670639038,
"step": 2540
},
{
"epoch": 2.9117983963344787,
"grad_norm": 0.04548896104097366,
"learning_rate": 1.0218081899107866e-06,
"loss": 1.0681244134902954,
"step": 2542
},
{
"epoch": 2.9140893470790377,
"grad_norm": 0.13808302581310272,
"learning_rate": 1.0207050066869884e-06,
"loss": 1.128277063369751,
"step": 2544
},
{
"epoch": 2.9163802978235966,
"grad_norm": 0.15436206758022308,
"learning_rate": 1.0196303907120835e-06,
"loss": 1.1815663576126099,
"step": 2546
},
{
"epoch": 2.9186712485681556,
"grad_norm": 0.12546345591545105,
"learning_rate": 1.0185843488395675e-06,
"loss": 1.1786937713623047,
"step": 2548
},
{
"epoch": 2.9209621993127146,
"grad_norm": 0.17458893358707428,
"learning_rate": 1.017566887740698e-06,
"loss": 1.1511332988739014,
"step": 2550
},
{
"epoch": 2.923253150057274,
"grad_norm": 0.19275349378585815,
"learning_rate": 1.0165780139044579e-06,
"loss": 1.0778021812438965,
"step": 2552
},
{
"epoch": 2.925544100801833,
"grad_norm": 0.24212509393692017,
"learning_rate": 1.0156177336375104e-06,
"loss": 0.9915075302124023,
"step": 2554
},
{
"epoch": 2.927835051546392,
"grad_norm": 0.21904876828193665,
"learning_rate": 1.0146860530641597e-06,
"loss": 1.1115221977233887,
"step": 2556
},
{
"epoch": 2.930126002290951,
"grad_norm": 0.16616712510585785,
"learning_rate": 1.013782978126312e-06,
"loss": 1.1849524974822998,
"step": 2558
},
{
"epoch": 2.93241695303551,
"grad_norm": 0.034661658108234406,
"learning_rate": 1.0129085145834374e-06,
"loss": 0.8701274394989014,
"step": 2560
},
{
"epoch": 2.934707903780069,
"grad_norm": 0.2624439299106598,
"learning_rate": 1.012062668012534e-06,
"loss": 0.7686551809310913,
"step": 2562
},
{
"epoch": 2.936998854524628,
"grad_norm": 0.25557151436805725,
"learning_rate": 1.0112454438080912e-06,
"loss": 0.9486254453659058,
"step": 2564
},
{
"epoch": 2.939289805269187,
"grad_norm": 0.07392004132270813,
"learning_rate": 1.010456847182055e-06,
"loss": 1.0521230697631836,
"step": 2566
},
{
"epoch": 2.941580756013746,
"grad_norm": 0.05467333644628525,
"learning_rate": 1.0096968831637976e-06,
"loss": 1.0537450313568115,
"step": 2568
},
{
"epoch": 2.943871706758305,
"grad_norm": 0.12582947313785553,
"learning_rate": 1.008965556600081e-06,
"loss": 1.0930761098861694,
"step": 2570
},
{
"epoch": 2.946162657502864,
"grad_norm": 0.1186356469988823,
"learning_rate": 1.0082628721550296e-06,
"loss": 1.0622632503509521,
"step": 2572
},
{
"epoch": 2.948453608247423,
"grad_norm": 0.03574733808636665,
"learning_rate": 1.0075888343100988e-06,
"loss": 1.1093733310699463,
"step": 2574
},
{
"epoch": 2.950744558991982,
"grad_norm": 0.13341714441776276,
"learning_rate": 1.0069434473640478e-06,
"loss": 0.6080751419067383,
"step": 2576
},
{
"epoch": 2.953035509736541,
"grad_norm": 0.11503258347511292,
"learning_rate": 1.0063267154329093e-06,
"loss": 1.136846899986267,
"step": 2578
},
{
"epoch": 2.9553264604811,
"grad_norm": 0.03894869238138199,
"learning_rate": 1.0057386424499674e-06,
"loss": 1.02817702293396,
"step": 2580
},
{
"epoch": 2.957617411225659,
"grad_norm": 0.03900009021162987,
"learning_rate": 1.005179232165728e-06,
"loss": 1.0276997089385986,
"step": 2582
},
{
"epoch": 2.959908361970218,
"grad_norm": 0.03746483847498894,
"learning_rate": 1.0046484881478987e-06,
"loss": 1.0998444557189941,
"step": 2584
},
{
"epoch": 2.9621993127147768,
"grad_norm": 0.13167481124401093,
"learning_rate": 1.0041464137813643e-06,
"loss": 1.077620506286621,
"step": 2586
},
{
"epoch": 2.9644902634593358,
"grad_norm": 0.11638258397579193,
"learning_rate": 1.0036730122681644e-06,
"loss": 1.0882108211517334,
"step": 2588
},
{
"epoch": 2.9667812142038947,
"grad_norm": 0.10818243771791458,
"learning_rate": 1.003228286627476e-06,
"loss": 1.0548460483551025,
"step": 2590
},
{
"epoch": 2.9690721649484537,
"grad_norm": 0.16958948969841003,
"learning_rate": 1.00281223969559e-06,
"loss": 1.097860336303711,
"step": 2592
},
{
"epoch": 2.9713631156930127,
"grad_norm": 0.24064916372299194,
"learning_rate": 1.0024248741258968e-06,
"loss": 0.8971477746963501,
"step": 2594
},
{
"epoch": 2.9736540664375717,
"grad_norm": 0.13208243250846863,
"learning_rate": 1.0020661923888674e-06,
"loss": 1.1392490863800049,
"step": 2596
},
{
"epoch": 2.9759450171821307,
"grad_norm": 0.14331702888011932,
"learning_rate": 1.001736196772038e-06,
"loss": 1.1136538982391357,
"step": 2598
},
{
"epoch": 2.9782359679266897,
"grad_norm": 0.11038228869438171,
"learning_rate": 1.0014348893799954e-06,
"loss": 1.282451868057251,
"step": 2600
},
{
"epoch": 2.9805269186712486,
"grad_norm": 0.12085051089525223,
"learning_rate": 1.0011622721343655e-06,
"loss": 1.0986806154251099,
"step": 2602
},
{
"epoch": 2.9828178694158076,
"grad_norm": 0.03834696114063263,
"learning_rate": 1.0009183467737969e-06,
"loss": 0.8867760896682739,
"step": 2604
},
{
"epoch": 2.9851088201603666,
"grad_norm": 0.04324295371770859,
"learning_rate": 1.0007031148539529e-06,
"loss": 1.044438362121582,
"step": 2606
},
{
"epoch": 2.9873997709049256,
"grad_norm": 0.308033287525177,
"learning_rate": 1.0005165777475022e-06,
"loss": 0.5716328620910645,
"step": 2608
},
{
"epoch": 2.9896907216494846,
"grad_norm": 0.034776829183101654,
"learning_rate": 1.000358736644108e-06,
"loss": 0.4520626366138458,
"step": 2610
},
{
"epoch": 2.9919816723940436,
"grad_norm": 0.052479542791843414,
"learning_rate": 1.00022959255042e-06,
"loss": 1.0296052694320679,
"step": 2612
},
{
"epoch": 2.9942726231386025,
"grad_norm": 0.05816009268164635,
"learning_rate": 1.0001291462900717e-06,
"loss": 1.0887324810028076,
"step": 2614
},
{
"epoch": 2.9965635738831615,
"grad_norm": 0.2066674381494522,
"learning_rate": 1.0000573985036708e-06,
"loss": 1.0183582305908203,
"step": 2616
},
{
"epoch": 2.9988545246277205,
"grad_norm": 0.12765929102897644,
"learning_rate": 1.0000143496487968e-06,
"loss": 1.177083969116211,
"step": 2618
},
{
"epoch": 3.0,
"step": 2619,
"total_flos": 4.569263266953429e+18,
"train_loss": 1.1290170093980534,
"train_runtime": 13562.7006,
"train_samples_per_second": 3.09,
"train_steps_per_second": 0.193
}
],
"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": 4.569263266953429e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}