| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9824992324224747, | |
| "eval_steps": 500, | |
| "global_step": 3200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003070310101320233, | |
| "grad_norm": 1.5795737504959106, | |
| "learning_rate": 1.0204081632653061e-07, | |
| "loss": 1.4209, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.006140620202640466, | |
| "grad_norm": 1.6946674585342407, | |
| "learning_rate": 2.0408163265306121e-07, | |
| "loss": 1.4115, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0092109303039607, | |
| "grad_norm": 2.268678903579712, | |
| "learning_rate": 3.0612244897959183e-07, | |
| "loss": 1.5281, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.012281240405280933, | |
| "grad_norm": 1.6933097839355469, | |
| "learning_rate": 4.0816326530612243e-07, | |
| "loss": 1.4233, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.015351550506601168, | |
| "grad_norm": 1.591718077659607, | |
| "learning_rate": 5.10204081632653e-07, | |
| "loss": 1.4581, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0184218606079214, | |
| "grad_norm": 1.4630407094955444, | |
| "learning_rate": 6.122448979591837e-07, | |
| "loss": 1.5041, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.021492170709241632, | |
| "grad_norm": 1.8335317373275757, | |
| "learning_rate": 7.142857142857143e-07, | |
| "loss": 1.4093, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.024562480810561865, | |
| "grad_norm": 1.1747108697891235, | |
| "learning_rate": 8.163265306122449e-07, | |
| "loss": 1.4103, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0276327909118821, | |
| "grad_norm": 1.8263486623764038, | |
| "learning_rate": 9.183673469387755e-07, | |
| "loss": 1.4307, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.030703101013202335, | |
| "grad_norm": 1.6976567506790161, | |
| "learning_rate": 9.993668882557771e-07, | |
| "loss": 1.3716, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03377341111452257, | |
| "grad_norm": 1.7711292505264282, | |
| "learning_rate": 9.962013295346628e-07, | |
| "loss": 1.5421, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0368437212158428, | |
| "grad_norm": 1.3698827028274536, | |
| "learning_rate": 9.930357708135485e-07, | |
| "loss": 1.416, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.039914031317163035, | |
| "grad_norm": 2.1234116554260254, | |
| "learning_rate": 9.898702120924344e-07, | |
| "loss": 1.474, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.042984341418483264, | |
| "grad_norm": 1.4592777490615845, | |
| "learning_rate": 9.8670465337132e-07, | |
| "loss": 1.4286, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0460546515198035, | |
| "grad_norm": 2.3222646713256836, | |
| "learning_rate": 9.835390946502057e-07, | |
| "loss": 1.4442, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04912496162112373, | |
| "grad_norm": 1.4718642234802246, | |
| "learning_rate": 9.803735359290914e-07, | |
| "loss": 1.3608, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.05219527172244397, | |
| "grad_norm": 1.648229718208313, | |
| "learning_rate": 9.772079772079773e-07, | |
| "loss": 1.5419, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0552655818237642, | |
| "grad_norm": 2.1675705909729004, | |
| "learning_rate": 9.74042418486863e-07, | |
| "loss": 1.3704, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.058335891925084433, | |
| "grad_norm": 1.9648547172546387, | |
| "learning_rate": 9.708768597657486e-07, | |
| "loss": 1.5037, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.06140620202640467, | |
| "grad_norm": 2.2735514640808105, | |
| "learning_rate": 9.677113010446345e-07, | |
| "loss": 1.4135, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0644765121277249, | |
| "grad_norm": 1.4363802671432495, | |
| "learning_rate": 9.6454574232352e-07, | |
| "loss": 1.4943, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.06754682222904514, | |
| "grad_norm": 1.8715194463729858, | |
| "learning_rate": 9.613801836024058e-07, | |
| "loss": 1.3656, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.07061713233036537, | |
| "grad_norm": 1.812849521636963, | |
| "learning_rate": 9.582146248812915e-07, | |
| "loss": 1.3368, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.0736874424316856, | |
| "grad_norm": 1.711365818977356, | |
| "learning_rate": 9.550490661601772e-07, | |
| "loss": 1.344, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.07675775253300583, | |
| "grad_norm": 1.574011206626892, | |
| "learning_rate": 9.51883507439063e-07, | |
| "loss": 1.2724, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.07982806263432607, | |
| "grad_norm": 1.8186664581298828, | |
| "learning_rate": 9.487179487179486e-07, | |
| "loss": 1.3877, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.0828983727356463, | |
| "grad_norm": 2.6951262950897217, | |
| "learning_rate": 9.455523899968344e-07, | |
| "loss": 1.3745, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.08596868283696653, | |
| "grad_norm": 1.8013410568237305, | |
| "learning_rate": 9.423868312757202e-07, | |
| "loss": 1.5363, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.08903899293828677, | |
| "grad_norm": 1.856616735458374, | |
| "learning_rate": 9.392212725546059e-07, | |
| "loss": 1.3532, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.092109303039607, | |
| "grad_norm": 4.05745267868042, | |
| "learning_rate": 9.360557138334915e-07, | |
| "loss": 1.3142, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09517961314092724, | |
| "grad_norm": 1.9517884254455566, | |
| "learning_rate": 9.328901551123773e-07, | |
| "loss": 1.2781, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.09824992324224746, | |
| "grad_norm": 1.9139647483825684, | |
| "learning_rate": 9.29724596391263e-07, | |
| "loss": 1.3773, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1013202333435677, | |
| "grad_norm": 2.246847152709961, | |
| "learning_rate": 9.265590376701488e-07, | |
| "loss": 1.3832, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.10439054344488793, | |
| "grad_norm": 1.9175080060958862, | |
| "learning_rate": 9.233934789490344e-07, | |
| "loss": 1.5022, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.10746085354620817, | |
| "grad_norm": 2.2456302642822266, | |
| "learning_rate": 9.202279202279202e-07, | |
| "loss": 1.3191, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.1105311636475284, | |
| "grad_norm": 2.491514205932617, | |
| "learning_rate": 9.17062361506806e-07, | |
| "loss": 1.2719, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.11360147374884863, | |
| "grad_norm": 2.05846905708313, | |
| "learning_rate": 9.138968027856916e-07, | |
| "loss": 1.3671, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.11667178385016887, | |
| "grad_norm": 2.2819442749023438, | |
| "learning_rate": 9.107312440645773e-07, | |
| "loss": 1.2372, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.1197420939514891, | |
| "grad_norm": 1.681643009185791, | |
| "learning_rate": 9.075656853434631e-07, | |
| "loss": 1.1861, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.12281240405280934, | |
| "grad_norm": 1.7323535680770874, | |
| "learning_rate": 9.044001266223488e-07, | |
| "loss": 1.4128, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.12588271415412958, | |
| "grad_norm": 1.7624768018722534, | |
| "learning_rate": 9.012345679012346e-07, | |
| "loss": 1.2962, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.1289530242554498, | |
| "grad_norm": 2.0329411029815674, | |
| "learning_rate": 8.980690091801203e-07, | |
| "loss": 1.3528, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.13202333435677002, | |
| "grad_norm": 2.3826355934143066, | |
| "learning_rate": 8.949034504590059e-07, | |
| "loss": 1.3301, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.13509364445809027, | |
| "grad_norm": 2.101026773452759, | |
| "learning_rate": 8.917378917378917e-07, | |
| "loss": 1.2829, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.1381639545594105, | |
| "grad_norm": 2.447965383529663, | |
| "learning_rate": 8.885723330167775e-07, | |
| "loss": 1.2124, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.14123426466073075, | |
| "grad_norm": 2.949786901473999, | |
| "learning_rate": 8.854067742956631e-07, | |
| "loss": 1.2711, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.14430457476205097, | |
| "grad_norm": 2.295283079147339, | |
| "learning_rate": 8.822412155745489e-07, | |
| "loss": 1.1455, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.1473748848633712, | |
| "grad_norm": 2.41044282913208, | |
| "learning_rate": 8.790756568534346e-07, | |
| "loss": 1.2558, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.15044519496469144, | |
| "grad_norm": 1.8993618488311768, | |
| "learning_rate": 8.759100981323203e-07, | |
| "loss": 1.2903, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.15351550506601166, | |
| "grad_norm": 2.800321102142334, | |
| "learning_rate": 8.72744539411206e-07, | |
| "loss": 1.348, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1565858151673319, | |
| "grad_norm": 1.8116044998168945, | |
| "learning_rate": 8.695789806900917e-07, | |
| "loss": 1.3102, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.15965612526865214, | |
| "grad_norm": 2.3227620124816895, | |
| "learning_rate": 8.664134219689775e-07, | |
| "loss": 1.2579, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.16272643536997236, | |
| "grad_norm": 2.884206771850586, | |
| "learning_rate": 8.632478632478633e-07, | |
| "loss": 1.2379, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.1657967454712926, | |
| "grad_norm": 1.9436391592025757, | |
| "learning_rate": 8.600823045267489e-07, | |
| "loss": 1.2234, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.16886705557261283, | |
| "grad_norm": 2.247084617614746, | |
| "learning_rate": 8.569167458056347e-07, | |
| "loss": 1.3054, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.17193736567393306, | |
| "grad_norm": 1.5196994543075562, | |
| "learning_rate": 8.537511870845204e-07, | |
| "loss": 1.2672, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.1750076757752533, | |
| "grad_norm": 1.7843005657196045, | |
| "learning_rate": 8.505856283634061e-07, | |
| "loss": 1.2728, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.17807798587657353, | |
| "grad_norm": 2.1363468170166016, | |
| "learning_rate": 8.474200696422918e-07, | |
| "loss": 1.1388, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.18114829597789378, | |
| "grad_norm": 2.081279993057251, | |
| "learning_rate": 8.442545109211775e-07, | |
| "loss": 1.2837, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.184218606079214, | |
| "grad_norm": 2.1368634700775146, | |
| "learning_rate": 8.410889522000633e-07, | |
| "loss": 1.2227, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.18728891618053423, | |
| "grad_norm": 1.2492815256118774, | |
| "learning_rate": 8.379233934789491e-07, | |
| "loss": 1.2216, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.19035922628185448, | |
| "grad_norm": 1.7561103105545044, | |
| "learning_rate": 8.347578347578346e-07, | |
| "loss": 1.2899, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.1934295363831747, | |
| "grad_norm": 2.0109572410583496, | |
| "learning_rate": 8.315922760367204e-07, | |
| "loss": 1.2586, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.19649984648449492, | |
| "grad_norm": 1.9294732809066772, | |
| "learning_rate": 8.284267173156062e-07, | |
| "loss": 1.2466, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.19957015658581517, | |
| "grad_norm": 1.9305318593978882, | |
| "learning_rate": 8.252611585944919e-07, | |
| "loss": 1.2221, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2026404666871354, | |
| "grad_norm": 1.4620651006698608, | |
| "learning_rate": 8.220955998733776e-07, | |
| "loss": 1.182, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.20571077678845565, | |
| "grad_norm": 2.4326343536376953, | |
| "learning_rate": 8.189300411522634e-07, | |
| "loss": 1.1801, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.20878108688977587, | |
| "grad_norm": 1.9332904815673828, | |
| "learning_rate": 8.157644824311491e-07, | |
| "loss": 1.0918, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.2118513969910961, | |
| "grad_norm": 2.345059394836426, | |
| "learning_rate": 8.125989237100348e-07, | |
| "loss": 1.1408, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.21492170709241634, | |
| "grad_norm": 2.2262935638427734, | |
| "learning_rate": 8.094333649889204e-07, | |
| "loss": 1.1688, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.21799201719373656, | |
| "grad_norm": 1.9713890552520752, | |
| "learning_rate": 8.062678062678062e-07, | |
| "loss": 1.1289, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.2210623272950568, | |
| "grad_norm": 2.036687135696411, | |
| "learning_rate": 8.03102247546692e-07, | |
| "loss": 1.0987, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.22413263739637704, | |
| "grad_norm": 2.875053644180298, | |
| "learning_rate": 7.999366888255777e-07, | |
| "loss": 1.1466, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.22720294749769726, | |
| "grad_norm": 1.509848952293396, | |
| "learning_rate": 7.967711301044634e-07, | |
| "loss": 1.1387, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.2302732575990175, | |
| "grad_norm": 1.844084620475769, | |
| "learning_rate": 7.936055713833491e-07, | |
| "loss": 1.2311, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.23334356770033773, | |
| "grad_norm": 1.9919129610061646, | |
| "learning_rate": 7.904400126622348e-07, | |
| "loss": 1.2468, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.23641387780165796, | |
| "grad_norm": 1.9697023630142212, | |
| "learning_rate": 7.872744539411206e-07, | |
| "loss": 1.1679, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.2394841879029782, | |
| "grad_norm": 1.8078380823135376, | |
| "learning_rate": 7.841088952200063e-07, | |
| "loss": 1.1574, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.24255449800429843, | |
| "grad_norm": 1.8240193128585815, | |
| "learning_rate": 7.80943336498892e-07, | |
| "loss": 1.2098, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.24562480810561868, | |
| "grad_norm": 2.1345813274383545, | |
| "learning_rate": 7.777777777777778e-07, | |
| "loss": 1.1673, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.2486951182069389, | |
| "grad_norm": 2.6286094188690186, | |
| "learning_rate": 7.746122190566636e-07, | |
| "loss": 1.1209, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.25176542830825915, | |
| "grad_norm": 2.819676637649536, | |
| "learning_rate": 7.714466603355491e-07, | |
| "loss": 1.1418, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.25483573840957935, | |
| "grad_norm": 3.308809995651245, | |
| "learning_rate": 7.682811016144349e-07, | |
| "loss": 1.0108, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.2579060485108996, | |
| "grad_norm": 2.4087624549865723, | |
| "learning_rate": 7.651155428933206e-07, | |
| "loss": 1.1166, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.26097635861221985, | |
| "grad_norm": 2.6052870750427246, | |
| "learning_rate": 7.619499841722064e-07, | |
| "loss": 1.1701, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.26404666871354004, | |
| "grad_norm": 1.932104468345642, | |
| "learning_rate": 7.587844254510921e-07, | |
| "loss": 1.1451, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.2671169788148603, | |
| "grad_norm": 1.9061886072158813, | |
| "learning_rate": 7.556188667299778e-07, | |
| "loss": 1.2128, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.27018728891618055, | |
| "grad_norm": 2.2594542503356934, | |
| "learning_rate": 7.524533080088635e-07, | |
| "loss": 1.2324, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.27325759901750074, | |
| "grad_norm": 2.510110855102539, | |
| "learning_rate": 7.492877492877493e-07, | |
| "loss": 1.1386, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.276327909118821, | |
| "grad_norm": 2.064319610595703, | |
| "learning_rate": 7.461221905666349e-07, | |
| "loss": 1.1055, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.27939821922014124, | |
| "grad_norm": 2.702468156814575, | |
| "learning_rate": 7.429566318455207e-07, | |
| "loss": 1.174, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.2824685293214615, | |
| "grad_norm": 2.1893582344055176, | |
| "learning_rate": 7.397910731244065e-07, | |
| "loss": 1.0903, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.2855388394227817, | |
| "grad_norm": 3.4419305324554443, | |
| "learning_rate": 7.366255144032922e-07, | |
| "loss": 1.0286, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.28860914952410194, | |
| "grad_norm": 1.8508418798446655, | |
| "learning_rate": 7.334599556821778e-07, | |
| "loss": 1.2314, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.2916794596254222, | |
| "grad_norm": 2.510098457336426, | |
| "learning_rate": 7.302943969610635e-07, | |
| "loss": 1.1322, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2947497697267424, | |
| "grad_norm": 2.328416347503662, | |
| "learning_rate": 7.271288382399493e-07, | |
| "loss": 1.025, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.29782007982806263, | |
| "grad_norm": 2.1068317890167236, | |
| "learning_rate": 7.239632795188351e-07, | |
| "loss": 1.0147, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.3008903899293829, | |
| "grad_norm": 1.842786192893982, | |
| "learning_rate": 7.207977207977207e-07, | |
| "loss": 1.0978, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.3039607000307031, | |
| "grad_norm": 1.274963140487671, | |
| "learning_rate": 7.176321620766065e-07, | |
| "loss": 0.9776, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.30703101013202333, | |
| "grad_norm": 2.67863392829895, | |
| "learning_rate": 7.144666033554923e-07, | |
| "loss": 1.1785, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3101013202333436, | |
| "grad_norm": 1.860278844833374, | |
| "learning_rate": 7.113010446343779e-07, | |
| "loss": 1.2182, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.3131716303346638, | |
| "grad_norm": 1.9876034259796143, | |
| "learning_rate": 7.081354859132636e-07, | |
| "loss": 1.166, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.316241940435984, | |
| "grad_norm": 2.1570093631744385, | |
| "learning_rate": 7.049699271921494e-07, | |
| "loss": 1.0908, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.3193122505373043, | |
| "grad_norm": 3.1807639598846436, | |
| "learning_rate": 7.018043684710351e-07, | |
| "loss": 1.0003, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.3223825606386245, | |
| "grad_norm": 3.396947145462036, | |
| "learning_rate": 6.986388097499209e-07, | |
| "loss": 1.1049, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.3254528707399447, | |
| "grad_norm": 2.2146832942962646, | |
| "learning_rate": 6.954732510288066e-07, | |
| "loss": 1.0274, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.328523180841265, | |
| "grad_norm": 1.499233365058899, | |
| "learning_rate": 6.923076923076922e-07, | |
| "loss": 0.9735, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.3315934909425852, | |
| "grad_norm": 1.6564570665359497, | |
| "learning_rate": 6.89142133586578e-07, | |
| "loss": 1.1102, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.3346638010439054, | |
| "grad_norm": 1.7835310697555542, | |
| "learning_rate": 6.859765748654637e-07, | |
| "loss": 1.0519, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.33773411114522567, | |
| "grad_norm": 2.079343795776367, | |
| "learning_rate": 6.828110161443494e-07, | |
| "loss": 1.0187, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.3408044212465459, | |
| "grad_norm": 2.1256511211395264, | |
| "learning_rate": 6.796454574232352e-07, | |
| "loss": 1.0023, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.3438747313478661, | |
| "grad_norm": 2.375253915786743, | |
| "learning_rate": 6.764798987021209e-07, | |
| "loss": 1.2046, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.34694504144918636, | |
| "grad_norm": 2.2755587100982666, | |
| "learning_rate": 6.733143399810067e-07, | |
| "loss": 1.1124, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.3500153515505066, | |
| "grad_norm": 1.930176019668579, | |
| "learning_rate": 6.701487812598923e-07, | |
| "loss": 0.9806, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.3530856616518268, | |
| "grad_norm": 1.805470585823059, | |
| "learning_rate": 6.66983222538778e-07, | |
| "loss": 1.0932, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.35615597175314706, | |
| "grad_norm": 1.877125859260559, | |
| "learning_rate": 6.638176638176638e-07, | |
| "loss": 1.0703, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.3592262818544673, | |
| "grad_norm": 1.6196848154067993, | |
| "learning_rate": 6.606521050965496e-07, | |
| "loss": 1.14, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.36229659195578756, | |
| "grad_norm": 2.2336461544036865, | |
| "learning_rate": 6.574865463754352e-07, | |
| "loss": 1.0706, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.36536690205710776, | |
| "grad_norm": 2.2776904106140137, | |
| "learning_rate": 6.54320987654321e-07, | |
| "loss": 0.9214, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.368437212158428, | |
| "grad_norm": 1.7691733837127686, | |
| "learning_rate": 6.511554289332066e-07, | |
| "loss": 1.0746, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.37150752225974826, | |
| "grad_norm": 2.1708719730377197, | |
| "learning_rate": 6.479898702120924e-07, | |
| "loss": 1.0607, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.37457783236106845, | |
| "grad_norm": 1.6801633834838867, | |
| "learning_rate": 6.448243114909781e-07, | |
| "loss": 1.0381, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.3776481424623887, | |
| "grad_norm": 2.653822422027588, | |
| "learning_rate": 6.416587527698638e-07, | |
| "loss": 1.0508, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.38071845256370895, | |
| "grad_norm": 2.0415942668914795, | |
| "learning_rate": 6.384931940487496e-07, | |
| "loss": 1.0113, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.38378876266502915, | |
| "grad_norm": 1.906172752380371, | |
| "learning_rate": 6.353276353276354e-07, | |
| "loss": 0.9856, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.3868590727663494, | |
| "grad_norm": 1.486282229423523, | |
| "learning_rate": 6.321620766065209e-07, | |
| "loss": 0.9518, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.38992938286766965, | |
| "grad_norm": 1.6384541988372803, | |
| "learning_rate": 6.289965178854067e-07, | |
| "loss": 0.9473, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.39299969296898984, | |
| "grad_norm": 1.7707164287567139, | |
| "learning_rate": 6.258309591642925e-07, | |
| "loss": 1.0343, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.3960700030703101, | |
| "grad_norm": 2.7890567779541016, | |
| "learning_rate": 6.226654004431782e-07, | |
| "loss": 0.9901, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.39914031317163035, | |
| "grad_norm": 3.0801589488983154, | |
| "learning_rate": 6.194998417220639e-07, | |
| "loss": 0.9611, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.40221062327295054, | |
| "grad_norm": 1.4281337261199951, | |
| "learning_rate": 6.163342830009497e-07, | |
| "loss": 1.0773, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.4052809333742708, | |
| "grad_norm": 2.135127544403076, | |
| "learning_rate": 6.131687242798354e-07, | |
| "loss": 0.9285, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.40835124347559104, | |
| "grad_norm": 2.215280532836914, | |
| "learning_rate": 6.100031655587211e-07, | |
| "loss": 0.9311, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.4114215535769113, | |
| "grad_norm": 1.7742196321487427, | |
| "learning_rate": 6.068376068376067e-07, | |
| "loss": 0.9504, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.4144918636782315, | |
| "grad_norm": 1.6954290866851807, | |
| "learning_rate": 6.036720481164925e-07, | |
| "loss": 0.9653, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.41756217377955174, | |
| "grad_norm": 2.5020253658294678, | |
| "learning_rate": 6.005064893953783e-07, | |
| "loss": 0.8404, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.420632483880872, | |
| "grad_norm": 1.6127808094024658, | |
| "learning_rate": 5.97340930674264e-07, | |
| "loss": 1.0783, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.4237027939821922, | |
| "grad_norm": 1.9128029346466064, | |
| "learning_rate": 5.941753719531497e-07, | |
| "loss": 0.9965, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.42677310408351243, | |
| "grad_norm": 1.7032874822616577, | |
| "learning_rate": 5.910098132320354e-07, | |
| "loss": 0.9816, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.4298434141848327, | |
| "grad_norm": 2.7620840072631836, | |
| "learning_rate": 5.878442545109211e-07, | |
| "loss": 0.9708, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.4329137242861529, | |
| "grad_norm": 1.8601688146591187, | |
| "learning_rate": 5.846786957898069e-07, | |
| "loss": 1.0241, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.43598403438747313, | |
| "grad_norm": 2.5212254524230957, | |
| "learning_rate": 5.815131370686926e-07, | |
| "loss": 0.9622, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.4390543444887934, | |
| "grad_norm": 1.5986229181289673, | |
| "learning_rate": 5.783475783475783e-07, | |
| "loss": 1.018, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.4421246545901136, | |
| "grad_norm": 1.858396053314209, | |
| "learning_rate": 5.751820196264641e-07, | |
| "loss": 0.9908, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.4451949646914338, | |
| "grad_norm": 1.8508849143981934, | |
| "learning_rate": 5.720164609053498e-07, | |
| "loss": 0.95, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.4482652747927541, | |
| "grad_norm": 1.8969439268112183, | |
| "learning_rate": 5.688509021842354e-07, | |
| "loss": 0.9342, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.4513355848940743, | |
| "grad_norm": 1.766987919807434, | |
| "learning_rate": 5.656853434631212e-07, | |
| "loss": 1.0226, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.4544058949953945, | |
| "grad_norm": 1.9654399156570435, | |
| "learning_rate": 5.625197847420069e-07, | |
| "loss": 0.8694, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.45747620509671477, | |
| "grad_norm": 2.982687473297119, | |
| "learning_rate": 5.593542260208927e-07, | |
| "loss": 0.8769, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.460546515198035, | |
| "grad_norm": 2.498600721359253, | |
| "learning_rate": 5.561886672997784e-07, | |
| "loss": 1.0117, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.4636168252993552, | |
| "grad_norm": 2.473336696624756, | |
| "learning_rate": 5.530231085786641e-07, | |
| "loss": 0.8705, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.46668713540067547, | |
| "grad_norm": 2.05454421043396, | |
| "learning_rate": 5.498575498575498e-07, | |
| "loss": 0.9998, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.4697574455019957, | |
| "grad_norm": 2.6380317211151123, | |
| "learning_rate": 5.466919911364356e-07, | |
| "loss": 0.9217, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.4728277556033159, | |
| "grad_norm": 3.275218963623047, | |
| "learning_rate": 5.435264324153212e-07, | |
| "loss": 0.932, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.47589806570463616, | |
| "grad_norm": 1.7377240657806396, | |
| "learning_rate": 5.40360873694207e-07, | |
| "loss": 0.9129, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.4789683758059564, | |
| "grad_norm": 2.8902907371520996, | |
| "learning_rate": 5.371953149730927e-07, | |
| "loss": 0.9501, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.4820386859072766, | |
| "grad_norm": 1.725958228111267, | |
| "learning_rate": 5.340297562519785e-07, | |
| "loss": 0.9022, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.48510899600859686, | |
| "grad_norm": 1.855343222618103, | |
| "learning_rate": 5.308641975308642e-07, | |
| "loss": 0.9266, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.4881793061099171, | |
| "grad_norm": 1.868709921836853, | |
| "learning_rate": 5.276986388097498e-07, | |
| "loss": 0.8821, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.49124961621123736, | |
| "grad_norm": 3.291313886642456, | |
| "learning_rate": 5.245330800886356e-07, | |
| "loss": 0.9109, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.49431992631255756, | |
| "grad_norm": 2.1387109756469727, | |
| "learning_rate": 5.213675213675214e-07, | |
| "loss": 0.9324, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.4973902364138778, | |
| "grad_norm": 1.3971613645553589, | |
| "learning_rate": 5.18201962646407e-07, | |
| "loss": 0.9861, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.500460546515198, | |
| "grad_norm": 1.6886439323425293, | |
| "learning_rate": 5.150364039252928e-07, | |
| "loss": 0.8934, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.5035308566165183, | |
| "grad_norm": 2.2841763496398926, | |
| "learning_rate": 5.118708452041786e-07, | |
| "loss": 0.9155, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.5066011667178385, | |
| "grad_norm": 3.6136739253997803, | |
| "learning_rate": 5.087052864830642e-07, | |
| "loss": 0.8794, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.5096714768191587, | |
| "grad_norm": 2.126060962677002, | |
| "learning_rate": 5.055397277619499e-07, | |
| "loss": 0.897, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.512741786920479, | |
| "grad_norm": 1.4978396892547607, | |
| "learning_rate": 5.023741690408357e-07, | |
| "loss": 0.8214, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.5158120970217992, | |
| "grad_norm": 1.483849287033081, | |
| "learning_rate": 4.992086103197214e-07, | |
| "loss": 0.9382, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.5188824071231194, | |
| "grad_norm": 1.8234091997146606, | |
| "learning_rate": 4.960430515986072e-07, | |
| "loss": 1.016, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.5219527172244397, | |
| "grad_norm": 1.4991623163223267, | |
| "learning_rate": 4.928774928774928e-07, | |
| "loss": 0.9616, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.5250230273257599, | |
| "grad_norm": 2.0311944484710693, | |
| "learning_rate": 4.897119341563785e-07, | |
| "loss": 0.8459, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.5280933374270801, | |
| "grad_norm": 1.9169743061065674, | |
| "learning_rate": 4.865463754352643e-07, | |
| "loss": 0.8369, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.5311636475284004, | |
| "grad_norm": 2.482090473175049, | |
| "learning_rate": 4.833808167141501e-07, | |
| "loss": 0.8837, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.5342339576297206, | |
| "grad_norm": 2.389390230178833, | |
| "learning_rate": 4.802152579930357e-07, | |
| "loss": 0.8079, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.5373042677310408, | |
| "grad_norm": 1.729292392730713, | |
| "learning_rate": 4.770496992719214e-07, | |
| "loss": 0.9238, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.5403745778323611, | |
| "grad_norm": 2.0271520614624023, | |
| "learning_rate": 4.738841405508072e-07, | |
| "loss": 0.8741, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.5434448879336813, | |
| "grad_norm": 1.5236482620239258, | |
| "learning_rate": 4.707185818296929e-07, | |
| "loss": 0.9811, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.5465151980350015, | |
| "grad_norm": 1.9456485509872437, | |
| "learning_rate": 4.6755302310857865e-07, | |
| "loss": 0.8036, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.5495855081363218, | |
| "grad_norm": 1.8617053031921387, | |
| "learning_rate": 4.643874643874643e-07, | |
| "loss": 0.9359, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.552655818237642, | |
| "grad_norm": 1.4658828973770142, | |
| "learning_rate": 4.612219056663501e-07, | |
| "loss": 1.0115, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.5557261283389623, | |
| "grad_norm": 3.7697947025299072, | |
| "learning_rate": 4.580563469452358e-07, | |
| "loss": 0.9872, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.5587964384402825, | |
| "grad_norm": 1.5588624477386475, | |
| "learning_rate": 4.5489078822412155e-07, | |
| "loss": 0.885, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.5618667485416027, | |
| "grad_norm": 2.504701852798462, | |
| "learning_rate": 4.5172522950300727e-07, | |
| "loss": 0.8898, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.564937058642923, | |
| "grad_norm": 1.5156068801879883, | |
| "learning_rate": 4.48559670781893e-07, | |
| "loss": 0.8852, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.5680073687442432, | |
| "grad_norm": 1.6788108348846436, | |
| "learning_rate": 4.453941120607787e-07, | |
| "loss": 0.9429, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.5710776788455634, | |
| "grad_norm": 1.6702663898468018, | |
| "learning_rate": 4.422285533396644e-07, | |
| "loss": 0.8841, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.5741479889468837, | |
| "grad_norm": 1.7993017435073853, | |
| "learning_rate": 4.390629946185502e-07, | |
| "loss": 0.8859, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.5772182990482039, | |
| "grad_norm": 1.976618766784668, | |
| "learning_rate": 4.358974358974359e-07, | |
| "loss": 0.9707, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.5802886091495241, | |
| "grad_norm": 1.4222575426101685, | |
| "learning_rate": 4.3273187717632157e-07, | |
| "loss": 0.8285, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.5833589192508444, | |
| "grad_norm": 1.9608110189437866, | |
| "learning_rate": 4.295663184552073e-07, | |
| "loss": 0.8951, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.5864292293521646, | |
| "grad_norm": 1.8060814142227173, | |
| "learning_rate": 4.264007597340931e-07, | |
| "loss": 0.9487, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.5894995394534848, | |
| "grad_norm": 2.0148568153381348, | |
| "learning_rate": 4.2323520101297875e-07, | |
| "loss": 0.872, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.5925698495548051, | |
| "grad_norm": 2.4442741870880127, | |
| "learning_rate": 4.2006964229186447e-07, | |
| "loss": 0.8681, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.5956401596561253, | |
| "grad_norm": 2.112091541290283, | |
| "learning_rate": 4.1690408357075025e-07, | |
| "loss": 0.8961, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.5987104697574455, | |
| "grad_norm": 2.6487627029418945, | |
| "learning_rate": 4.13738524849636e-07, | |
| "loss": 0.8829, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.6017807798587658, | |
| "grad_norm": 1.5637843608856201, | |
| "learning_rate": 4.1057296612852165e-07, | |
| "loss": 0.9355, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.604851089960086, | |
| "grad_norm": 1.4878854751586914, | |
| "learning_rate": 4.0740740740740737e-07, | |
| "loss": 0.9063, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.6079214000614062, | |
| "grad_norm": 2.0287206172943115, | |
| "learning_rate": 4.0424184868629315e-07, | |
| "loss": 0.9765, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.6109917101627265, | |
| "grad_norm": 1.8464608192443848, | |
| "learning_rate": 4.010762899651788e-07, | |
| "loss": 1.0074, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.6140620202640467, | |
| "grad_norm": 2.208984851837158, | |
| "learning_rate": 3.9791073124406455e-07, | |
| "loss": 0.9381, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.6171323303653669, | |
| "grad_norm": 3.5903429985046387, | |
| "learning_rate": 3.9474517252295033e-07, | |
| "loss": 0.9285, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.6202026404666872, | |
| "grad_norm": 1.8227884769439697, | |
| "learning_rate": 3.91579613801836e-07, | |
| "loss": 0.8484, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.6232729505680074, | |
| "grad_norm": 1.8685295581817627, | |
| "learning_rate": 3.884140550807217e-07, | |
| "loss": 0.8183, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.6263432606693276, | |
| "grad_norm": 1.505017876625061, | |
| "learning_rate": 3.8524849635960745e-07, | |
| "loss": 0.9065, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.6294135707706479, | |
| "grad_norm": 1.7611987590789795, | |
| "learning_rate": 3.820829376384932e-07, | |
| "loss": 0.8511, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.632483880871968, | |
| "grad_norm": 1.9972463846206665, | |
| "learning_rate": 3.789173789173789e-07, | |
| "loss": 0.8686, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.6355541909732882, | |
| "grad_norm": 2.034118890762329, | |
| "learning_rate": 3.757518201962646e-07, | |
| "loss": 0.7841, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.6386245010746086, | |
| "grad_norm": 1.953757882118225, | |
| "learning_rate": 3.725862614751503e-07, | |
| "loss": 0.893, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.6416948111759287, | |
| "grad_norm": 2.0740292072296143, | |
| "learning_rate": 3.694207027540361e-07, | |
| "loss": 0.8787, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.644765121277249, | |
| "grad_norm": 1.808714509010315, | |
| "learning_rate": 3.662551440329218e-07, | |
| "loss": 0.8674, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.6478354313785692, | |
| "grad_norm": 2.792292594909668, | |
| "learning_rate": 3.630895853118075e-07, | |
| "loss": 0.7246, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.6509057414798894, | |
| "grad_norm": 1.9077739715576172, | |
| "learning_rate": 3.5992402659069325e-07, | |
| "loss": 0.8878, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.6539760515812097, | |
| "grad_norm": 2.024315357208252, | |
| "learning_rate": 3.56758467869579e-07, | |
| "loss": 0.8197, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.65704636168253, | |
| "grad_norm": 1.799088954925537, | |
| "learning_rate": 3.535929091484647e-07, | |
| "loss": 0.7852, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.6601166717838501, | |
| "grad_norm": 2.362877368927002, | |
| "learning_rate": 3.504273504273504e-07, | |
| "loss": 0.8204, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.6631869818851704, | |
| "grad_norm": 1.3777751922607422, | |
| "learning_rate": 3.4726179170623615e-07, | |
| "loss": 0.8348, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.6662572919864906, | |
| "grad_norm": 1.9549330472946167, | |
| "learning_rate": 3.440962329851219e-07, | |
| "loss": 0.8312, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.6693276020878108, | |
| "grad_norm": 1.8688315153121948, | |
| "learning_rate": 3.4093067426400755e-07, | |
| "loss": 0.8096, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.6723979121891311, | |
| "grad_norm": 1.5635504722595215, | |
| "learning_rate": 3.3776511554289333e-07, | |
| "loss": 0.9582, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.6754682222904513, | |
| "grad_norm": 1.500148892402649, | |
| "learning_rate": 3.3459955682177905e-07, | |
| "loss": 0.8907, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.6785385323917715, | |
| "grad_norm": 1.767208456993103, | |
| "learning_rate": 3.314339981006647e-07, | |
| "loss": 0.912, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.6816088424930918, | |
| "grad_norm": 2.737919807434082, | |
| "learning_rate": 3.2826843937955045e-07, | |
| "loss": 0.7511, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.684679152594412, | |
| "grad_norm": 1.8129867315292358, | |
| "learning_rate": 3.2510288065843623e-07, | |
| "loss": 0.8095, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.6877494626957322, | |
| "grad_norm": 1.8896024227142334, | |
| "learning_rate": 3.219373219373219e-07, | |
| "loss": 0.7891, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.6908197727970525, | |
| "grad_norm": 1.7065414190292358, | |
| "learning_rate": 3.187717632162076e-07, | |
| "loss": 0.8119, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.6938900828983727, | |
| "grad_norm": 1.7079863548278809, | |
| "learning_rate": 3.1560620449509335e-07, | |
| "loss": 0.7894, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.6969603929996929, | |
| "grad_norm": 1.4625552892684937, | |
| "learning_rate": 3.1244064577397913e-07, | |
| "loss": 0.8951, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.7000307031010132, | |
| "grad_norm": 1.7978609800338745, | |
| "learning_rate": 3.092750870528648e-07, | |
| "loss": 0.898, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.7031010132023334, | |
| "grad_norm": 2.16448974609375, | |
| "learning_rate": 3.0610952833175053e-07, | |
| "loss": 0.8961, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.7061713233036536, | |
| "grad_norm": 1.6174335479736328, | |
| "learning_rate": 3.029439696106363e-07, | |
| "loss": 0.7894, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.7092416334049739, | |
| "grad_norm": 2.177285671234131, | |
| "learning_rate": 2.99778410889522e-07, | |
| "loss": 0.873, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.7123119435062941, | |
| "grad_norm": 2.4392199516296387, | |
| "learning_rate": 2.966128521684077e-07, | |
| "loss": 0.869, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.7153822536076143, | |
| "grad_norm": 2.829726219177246, | |
| "learning_rate": 2.9344729344729343e-07, | |
| "loss": 0.8769, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.7184525637089346, | |
| "grad_norm": 2.4519991874694824, | |
| "learning_rate": 2.9028173472617915e-07, | |
| "loss": 0.7622, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.7215228738102548, | |
| "grad_norm": 2.479762077331543, | |
| "learning_rate": 2.871161760050649e-07, | |
| "loss": 0.8554, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.7245931839115751, | |
| "grad_norm": 1.7480168342590332, | |
| "learning_rate": 2.839506172839506e-07, | |
| "loss": 0.8004, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.7276634940128953, | |
| "grad_norm": 1.6796847581863403, | |
| "learning_rate": 2.8078505856283633e-07, | |
| "loss": 0.7557, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.7307338041142155, | |
| "grad_norm": 1.8005452156066895, | |
| "learning_rate": 2.7761949984172205e-07, | |
| "loss": 0.7349, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.7338041142155358, | |
| "grad_norm": 1.5013504028320312, | |
| "learning_rate": 2.744539411206078e-07, | |
| "loss": 0.7608, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.736874424316856, | |
| "grad_norm": 1.9756343364715576, | |
| "learning_rate": 2.7128838239949345e-07, | |
| "loss": 0.7542, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.7399447344181762, | |
| "grad_norm": 1.6326251029968262, | |
| "learning_rate": 2.6812282367837923e-07, | |
| "loss": 0.8952, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.7430150445194965, | |
| "grad_norm": 2.549936056137085, | |
| "learning_rate": 2.6495726495726495e-07, | |
| "loss": 0.9441, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.7460853546208167, | |
| "grad_norm": 2.071596622467041, | |
| "learning_rate": 2.617917062361507e-07, | |
| "loss": 0.8192, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.7491556647221369, | |
| "grad_norm": 2.2240512371063232, | |
| "learning_rate": 2.586261475150364e-07, | |
| "loss": 0.8697, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.7522259748234572, | |
| "grad_norm": 2.4233920574188232, | |
| "learning_rate": 2.5546058879392213e-07, | |
| "loss": 0.8084, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.7552962849247774, | |
| "grad_norm": 1.7786365747451782, | |
| "learning_rate": 2.5229503007280786e-07, | |
| "loss": 0.8905, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.7583665950260976, | |
| "grad_norm": 1.9785822629928589, | |
| "learning_rate": 2.491294713516936e-07, | |
| "loss": 0.7836, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.7614369051274179, | |
| "grad_norm": 1.8973301649093628, | |
| "learning_rate": 2.4596391263057925e-07, | |
| "loss": 0.8684, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.7645072152287381, | |
| "grad_norm": 1.6130449771881104, | |
| "learning_rate": 2.4279835390946503e-07, | |
| "loss": 0.7865, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.7675775253300583, | |
| "grad_norm": 1.6121731996536255, | |
| "learning_rate": 2.396327951883507e-07, | |
| "loss": 0.8248, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.7706478354313786, | |
| "grad_norm": 1.6531673669815063, | |
| "learning_rate": 2.3646723646723645e-07, | |
| "loss": 0.6813, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.7737181455326988, | |
| "grad_norm": 2.1497576236724854, | |
| "learning_rate": 2.3330167774612215e-07, | |
| "loss": 0.8148, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.776788455634019, | |
| "grad_norm": 1.9006414413452148, | |
| "learning_rate": 2.301361190250079e-07, | |
| "loss": 0.7169, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.7798587657353393, | |
| "grad_norm": 1.6517983675003052, | |
| "learning_rate": 2.2697056030389363e-07, | |
| "loss": 0.8157, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.7829290758366595, | |
| "grad_norm": 1.6298216581344604, | |
| "learning_rate": 2.2380500158277936e-07, | |
| "loss": 0.762, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.7859993859379797, | |
| "grad_norm": 1.198767066001892, | |
| "learning_rate": 2.2063944286166508e-07, | |
| "loss": 0.8029, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.7890696960393, | |
| "grad_norm": 1.9983083009719849, | |
| "learning_rate": 2.1747388414055078e-07, | |
| "loss": 0.7524, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.7921400061406202, | |
| "grad_norm": 1.9805610179901123, | |
| "learning_rate": 2.1430832541943653e-07, | |
| "loss": 0.7348, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.7952103162419404, | |
| "grad_norm": 2.2282845973968506, | |
| "learning_rate": 2.1114276669832223e-07, | |
| "loss": 0.7699, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.7982806263432607, | |
| "grad_norm": 1.6694320440292358, | |
| "learning_rate": 2.0797720797720798e-07, | |
| "loss": 0.716, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.8013509364445809, | |
| "grad_norm": 2.17622971534729, | |
| "learning_rate": 2.048116492560937e-07, | |
| "loss": 0.7734, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.8044212465459011, | |
| "grad_norm": 1.7180266380310059, | |
| "learning_rate": 2.016460905349794e-07, | |
| "loss": 0.8293, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.8074915566472214, | |
| "grad_norm": 2.1564178466796875, | |
| "learning_rate": 1.9848053181386516e-07, | |
| "loss": 0.746, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.8105618667485416, | |
| "grad_norm": 1.8624786138534546, | |
| "learning_rate": 1.9531497309275086e-07, | |
| "loss": 0.759, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.8136321768498619, | |
| "grad_norm": 3.1279971599578857, | |
| "learning_rate": 1.9214941437163658e-07, | |
| "loss": 0.7752, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.8167024869511821, | |
| "grad_norm": 2.345259189605713, | |
| "learning_rate": 1.889838556505223e-07, | |
| "loss": 0.7453, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.8197727970525023, | |
| "grad_norm": 2.008528709411621, | |
| "learning_rate": 1.8581829692940803e-07, | |
| "loss": 0.7598, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.8228431071538226, | |
| "grad_norm": 2.444312334060669, | |
| "learning_rate": 1.8265273820829376e-07, | |
| "loss": 0.8075, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.8259134172551428, | |
| "grad_norm": 2.321638345718384, | |
| "learning_rate": 1.7948717948717948e-07, | |
| "loss": 0.7458, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.828983727356463, | |
| "grad_norm": 1.5578280687332153, | |
| "learning_rate": 1.763216207660652e-07, | |
| "loss": 0.8062, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.8320540374577833, | |
| "grad_norm": 2.1762735843658447, | |
| "learning_rate": 1.7315606204495093e-07, | |
| "loss": 0.7908, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.8351243475591035, | |
| "grad_norm": 1.9740185737609863, | |
| "learning_rate": 1.6999050332383666e-07, | |
| "loss": 0.7735, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.8381946576604237, | |
| "grad_norm": 1.6252373456954956, | |
| "learning_rate": 1.6682494460272236e-07, | |
| "loss": 0.8089, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.841264967761744, | |
| "grad_norm": 2.2141306400299072, | |
| "learning_rate": 1.636593858816081e-07, | |
| "loss": 0.8623, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.8443352778630642, | |
| "grad_norm": 2.028036117553711, | |
| "learning_rate": 1.604938271604938e-07, | |
| "loss": 0.8883, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.8474055879643844, | |
| "grad_norm": 1.9973171949386597, | |
| "learning_rate": 1.5732826843937956e-07, | |
| "loss": 0.8589, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.8504758980657047, | |
| "grad_norm": 2.2462198734283447, | |
| "learning_rate": 1.5416270971826526e-07, | |
| "loss": 0.8815, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.8535462081670249, | |
| "grad_norm": 1.4654901027679443, | |
| "learning_rate": 1.5099715099715098e-07, | |
| "loss": 0.7683, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.8566165182683451, | |
| "grad_norm": 1.7927677631378174, | |
| "learning_rate": 1.4783159227603673e-07, | |
| "loss": 0.802, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.8596868283696654, | |
| "grad_norm": 2.216456890106201, | |
| "learning_rate": 1.4466603355492243e-07, | |
| "loss": 0.6807, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.8627571384709856, | |
| "grad_norm": 2.353961229324341, | |
| "learning_rate": 1.4150047483380816e-07, | |
| "loss": 0.8638, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.8658274485723058, | |
| "grad_norm": 1.4108779430389404, | |
| "learning_rate": 1.3833491611269388e-07, | |
| "loss": 0.7982, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.8688977586736261, | |
| "grad_norm": 2.021253824234009, | |
| "learning_rate": 1.351693573915796e-07, | |
| "loss": 0.771, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.8719680687749463, | |
| "grad_norm": 1.6902941465377808, | |
| "learning_rate": 1.3200379867046533e-07, | |
| "loss": 0.7988, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.8750383788762665, | |
| "grad_norm": 2.057300329208374, | |
| "learning_rate": 1.2883823994935106e-07, | |
| "loss": 0.7589, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.8781086889775868, | |
| "grad_norm": 2.0052621364593506, | |
| "learning_rate": 1.2567268122823676e-07, | |
| "loss": 0.7285, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.881178999078907, | |
| "grad_norm": 1.7238054275512695, | |
| "learning_rate": 1.225071225071225e-07, | |
| "loss": 0.7748, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.8842493091802271, | |
| "grad_norm": 1.6604859828948975, | |
| "learning_rate": 1.1934156378600823e-07, | |
| "loss": 0.7975, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.8873196192815475, | |
| "grad_norm": 2.2870965003967285, | |
| "learning_rate": 1.1617600506489395e-07, | |
| "loss": 0.822, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.8903899293828677, | |
| "grad_norm": 1.709179162979126, | |
| "learning_rate": 1.1301044634377967e-07, | |
| "loss": 0.7735, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.893460239484188, | |
| "grad_norm": 1.6187310218811035, | |
| "learning_rate": 1.0984488762266538e-07, | |
| "loss": 0.8106, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.8965305495855082, | |
| "grad_norm": 1.4298169612884521, | |
| "learning_rate": 1.0667932890155112e-07, | |
| "loss": 0.6901, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.8996008596868283, | |
| "grad_norm": 1.541094183921814, | |
| "learning_rate": 1.0351377018043685e-07, | |
| "loss": 0.775, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.9026711697881487, | |
| "grad_norm": 1.4745653867721558, | |
| "learning_rate": 1.0034821145932257e-07, | |
| "loss": 0.9154, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.9057414798894688, | |
| "grad_norm": 1.3007853031158447, | |
| "learning_rate": 9.718265273820828e-08, | |
| "loss": 0.7063, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.908811789990789, | |
| "grad_norm": 1.819677472114563, | |
| "learning_rate": 9.401709401709401e-08, | |
| "loss": 0.7288, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.9118821000921093, | |
| "grad_norm": 1.7065274715423584, | |
| "learning_rate": 9.085153529597973e-08, | |
| "loss": 0.7891, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.9149524101934295, | |
| "grad_norm": 1.572373390197754, | |
| "learning_rate": 8.768597657486546e-08, | |
| "loss": 0.802, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.9180227202947497, | |
| "grad_norm": 1.870667815208435, | |
| "learning_rate": 8.452041785375117e-08, | |
| "loss": 0.767, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.92109303039607, | |
| "grad_norm": 1.924141764640808, | |
| "learning_rate": 8.135485913263691e-08, | |
| "loss": 0.8292, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.9241633404973902, | |
| "grad_norm": 1.7546625137329102, | |
| "learning_rate": 7.818930041152264e-08, | |
| "loss": 0.7928, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.9272336505987104, | |
| "grad_norm": 2.3460092544555664, | |
| "learning_rate": 7.502374169040836e-08, | |
| "loss": 0.8534, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.9303039607000307, | |
| "grad_norm": 2.752711296081543, | |
| "learning_rate": 7.185818296929407e-08, | |
| "loss": 0.7184, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.9333742708013509, | |
| "grad_norm": 1.901767373085022, | |
| "learning_rate": 6.86926242481798e-08, | |
| "loss": 0.7939, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.9364445809026711, | |
| "grad_norm": 2.0688488483428955, | |
| "learning_rate": 6.552706552706552e-08, | |
| "loss": 0.7265, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.9395148910039914, | |
| "grad_norm": 1.7756198644638062, | |
| "learning_rate": 6.236150680595125e-08, | |
| "loss": 0.7499, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.9425852011053116, | |
| "grad_norm": 1.5019465684890747, | |
| "learning_rate": 5.919594808483697e-08, | |
| "loss": 0.8274, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.9456555112066318, | |
| "grad_norm": 1.6010905504226685, | |
| "learning_rate": 5.60303893637227e-08, | |
| "loss": 0.7893, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.9487258213079521, | |
| "grad_norm": 1.6651114225387573, | |
| "learning_rate": 5.286483064260842e-08, | |
| "loss": 0.7241, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.9517961314092723, | |
| "grad_norm": 2.506920576095581, | |
| "learning_rate": 4.969927192149414e-08, | |
| "loss": 0.8166, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.9548664415105925, | |
| "grad_norm": 2.0352845191955566, | |
| "learning_rate": 4.653371320037987e-08, | |
| "loss": 0.7791, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.9579367516119128, | |
| "grad_norm": 2.4386065006256104, | |
| "learning_rate": 4.336815447926559e-08, | |
| "loss": 0.7289, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.961007061713233, | |
| "grad_norm": 1.4857659339904785, | |
| "learning_rate": 4.020259575815131e-08, | |
| "loss": 0.8678, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.9640773718145532, | |
| "grad_norm": 1.967745065689087, | |
| "learning_rate": 3.7037037037037036e-08, | |
| "loss": 0.7746, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.9671476819158735, | |
| "grad_norm": 1.7988271713256836, | |
| "learning_rate": 3.387147831592276e-08, | |
| "loss": 0.7677, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.9702179920171937, | |
| "grad_norm": 1.9551514387130737, | |
| "learning_rate": 3.070591959480849e-08, | |
| "loss": 0.8411, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.9732883021185139, | |
| "grad_norm": 2.3268728256225586, | |
| "learning_rate": 2.7540360873694205e-08, | |
| "loss": 0.8246, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.9763586122198342, | |
| "grad_norm": 1.7236260175704956, | |
| "learning_rate": 2.4374802152579927e-08, | |
| "loss": 0.8471, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.9794289223211544, | |
| "grad_norm": 1.8868488073349, | |
| "learning_rate": 2.1209243431465652e-08, | |
| "loss": 0.7894, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.9824992324224747, | |
| "grad_norm": 2.07332706451416, | |
| "learning_rate": 1.8043684710351374e-08, | |
| "loss": 0.7217, | |
| "step": 3200 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 3257, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.17632691027968e+16, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |