| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.382978723404255, | |
| "global_step": 140000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00014999965456413008, | |
| "loss": 0.6492, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001499986182599296, | |
| "loss": 0.5432, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00014999689109762647, | |
| "loss": 0.5137, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00014999447309426713, | |
| "loss": 0.4986, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00014999136427371623, | |
| "loss": 0.4877, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001499875646666566, | |
| "loss": 0.4785, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00014998307431058887, | |
| "loss": 0.4702, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00014997789324983099, | |
| "loss": 0.462, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00014997202153551792, | |
| "loss": 0.454, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0001499654592256012, | |
| "loss": 0.4466, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0001499582063848481, | |
| "loss": 0.4398, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00014995026308484122, | |
| "loss": 0.4341, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00014994162940397778, | |
| "loss": 0.4288, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00014993230542746872, | |
| "loss": 0.4247, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00014992229124733788, | |
| "loss": 0.4207, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0001499115869624212, | |
| "loss": 0.4175, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00014990019267836565, | |
| "loss": 0.4146, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00014988810850762824, | |
| "loss": 0.4122, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00014987533456947482, | |
| "loss": 0.4097, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000149861870989979, | |
| "loss": 0.4082, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0001498477179020209, | |
| "loss": 0.406, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00014983287544528573, | |
| "loss": 0.4042, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0001498173755173638, | |
| "loss": 0.4026, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0001498011561473246, | |
| "loss": 0.4014, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00014978424786805407, | |
| "loss": 0.3999, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00014976665084643018, | |
| "loss": 0.3985, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00014974836525612832, | |
| "loss": 0.3971, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00014972939127761998, | |
| "loss": 0.3961, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0001497097290981706, | |
| "loss": 0.3949, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00014968937891183796, | |
| "loss": 0.3938, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00014966834091947024, | |
| "loss": 0.3927, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0001496466594659672, | |
| "loss": 0.3924, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00014962424786577448, | |
| "loss": 0.3908, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00014960114910236375, | |
| "loss": 0.3898, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0001495774116605143, | |
| "loss": 0.389, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00014955294063453595, | |
| "loss": 0.3883, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00014952778314911323, | |
| "loss": 0.3875, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0001495019394525403, | |
| "loss": 0.3871, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0001494755172845744, | |
| "loss": 0.3857, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00014944830467991823, | |
| "loss": 0.3855, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00014942040664853227, | |
| "loss": 0.3843, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00014939182346575884, | |
| "loss": 0.3841, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0001493625554137024, | |
| "loss": 0.3833, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00014933266336950772, | |
| "loss": 0.3825, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0001493020278205035, | |
| "loss": 0.3822, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00014927070828846305, | |
| "loss": 0.3814, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00014923870508249748, | |
| "loss": 0.3809, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00014920601851846558, | |
| "loss": 0.3806, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0001491727163396211, | |
| "loss": 0.3797, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00014913866539908636, | |
| "loss": 0.3794, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00014910393208783659, | |
| "loss": 0.3788, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00014906851674867546, | |
| "loss": 0.3784, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00014903249260525252, | |
| "loss": 0.3777, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00014899571562788615, | |
| "loss": 0.3776, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00014895833328588902, | |
| "loss": 0.3768, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00014892019611930716, | |
| "loss": 0.3762, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00014888137873821385, | |
| "loss": 0.3761, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00014884188152572092, | |
| "loss": 0.3755, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0001488017048716499, | |
| "loss": 0.3752, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00014876084917252817, | |
| "loss": 0.3748, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00014871931483158502, | |
| "loss": 0.3747, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00014867718736049287, | |
| "loss": 0.374, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0001486342983275925, | |
| "loss": 0.3735, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00014859073190187644, | |
| "loss": 0.3731, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0001485464885133277, | |
| "loss": 0.3735, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00014850156859861065, | |
| "loss": 0.3725, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00014845597260106674, | |
| "loss": 0.3721, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00014840970097071006, | |
| "loss": 0.3719, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00014836284873135484, | |
| "loss": 0.3717, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00014831522856104197, | |
| "loss": 0.371, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00014826693414700317, | |
| "loss": 0.3714, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00014821796596588483, | |
| "loss": 0.3706, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0001481683245009831, | |
| "loss": 0.3702, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00014811801024223922, | |
| "loss": 0.3702, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00014806712632996593, | |
| "loss": 0.3695, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00014801546932299877, | |
| "loss": 0.3696, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00014796314103080835, | |
| "loss": 0.3694, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00014791014196985377, | |
| "loss": 0.3687, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0001478564726632144, | |
| "loss": 0.3689, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00014780224298665108, | |
| "loss": 0.3684, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0001477472361221548, | |
| "loss": 0.3681, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0001476915606197887, | |
| "loss": 0.3677, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00014763521702904748, | |
| "loss": 0.3677, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0001475782059060196, | |
| "loss": 0.3672, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00014752052781338187, | |
| "loss": 0.3673, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00014746218332039373, | |
| "loss": 0.3669, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00014740317300289185, | |
| "loss": 0.3669, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0001473436174579246, | |
| "loss": 0.3666, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00014728327857389895, | |
| "loss": 0.3659, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00014722227563107712, | |
| "loss": 0.3661, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00014716060923153354, | |
| "loss": 0.3656, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00014709840530350058, | |
| "loss": 0.3656, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00014703541514677053, | |
| "loss": 0.3652, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00014697176337755626, | |
| "loss": 0.365, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0001469074506240751, | |
| "loss": 0.365, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00014684260812587756, | |
| "loss": 0.3645, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00014677710855432052, | |
| "loss": 0.3647, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00014671081931749022, | |
| "loss": 0.3645, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00014664387167180547, | |
| "loss": 0.3638, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00014657626627801257, | |
| "loss": 0.3638, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0001465080038033494, | |
| "loss": 0.3634, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00014643908492153906, | |
| "loss": 0.3634, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00014636978991651827, | |
| "loss": 0.3632, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00014629956288627466, | |
| "loss": 0.3633, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00014622868150611156, | |
| "loss": 0.3627, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.00014615714647559958, | |
| "loss": 0.3627, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.00014608495850076063, | |
| "loss": 0.3625, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.00014601211829406085, | |
| "loss": 0.3622, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00014593862657440366, | |
| "loss": 0.3619, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0001458644840671227, | |
| "loss": 0.3617, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0001457896915039746, | |
| "loss": 0.3617, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.00014571424962313165, | |
| "loss": 0.3613, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0001456381591691749, | |
| "loss": 0.3612, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0001455614208930863, | |
| "loss": 0.3613, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00014548403555224174, | |
| "loss": 0.3608, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0001454061606181845, | |
| "loss": 0.3608, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00014532748473578173, | |
| "loss": 0.3606, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0001452481640974784, | |
| "loss": 0.3606, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00014516819948613718, | |
| "loss": 0.3603, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0001450875916909765, | |
| "loss": 0.3597, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00014500650464850112, | |
| "loss": 0.3599, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00014492461416110898, | |
| "loss": 0.3597, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00014484208289398622, | |
| "loss": 0.3597, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0001447589116616831, | |
| "loss": 0.3595, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00014467510128506612, | |
| "loss": 0.3592, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0001445908221251849, | |
| "loss": 0.3592, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0001445057372218952, | |
| "loss": 0.3589, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0001444203598262954, | |
| "loss": 0.359, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0001443340050193786, | |
| "loss": 0.3585, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.000144247015261805, | |
| "loss": 0.3586, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00014415939141212825, | |
| "loss": 0.3585, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00014407113433516052, | |
| "loss": 0.3583, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00014398224490196345, | |
| "loss": 0.3578, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00014389272398983985, | |
| "loss": 0.3579, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.000143802572482325, | |
| "loss": 0.3577, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00014371179126917793, | |
| "loss": 0.3579, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0001436203812463725, | |
| "loss": 0.3576, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00014352834331608877, | |
| "loss": 0.3575, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00014343567838670393, | |
| "loss": 0.3573, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00014334257457903268, | |
| "loss": 0.3572, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00014324865965072573, | |
| "loss": 0.3569, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00014315412048368257, | |
| "loss": 0.357, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001430589580109666, | |
| "loss": 0.3565, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014296355754965273, | |
| "loss": 0.3563, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014286715377317228, | |
| "loss": 0.3563, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00014277012952326458, | |
| "loss": 0.3563, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001426724857575198, | |
| "loss": 0.3561, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00014257422343964251, | |
| "loss": 0.3562, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0001424753435394421, | |
| "loss": 0.3558, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00014237584703282328, | |
| "loss": 0.3557, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0001422757349017764, | |
| "loss": 0.3554, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.00014217500813436775, | |
| "loss": 0.3554, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0001420736677247299, | |
| "loss": 0.3551, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.00014197191918990406, | |
| "loss": 0.3551, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.00014186935572468444, | |
| "loss": 0.3547, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00014176618163390153, | |
| "loss": 0.3545, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.00014166239793584186, | |
| "loss": 0.3548, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.0001415580056548087, | |
| "loss": 0.3546, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.00014145321642642885, | |
| "loss": 0.3544, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00014134782310325824, | |
| "loss": 0.3545, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0001412418267213298, | |
| "loss": 0.3547, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.0001411350140948281, | |
| "loss": 0.3547, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0001410275980904185, | |
| "loss": 0.354, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0001409195797682536, | |
| "loss": 0.3539, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.00014081096019443061, | |
| "loss": 0.3538, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00014070174044098087, | |
| "loss": 0.3537, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.00014059192158585924, | |
| "loss": 0.3534, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.00014048150471293355, | |
| "loss": 0.3539, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.00014037049091197368, | |
| "loss": 0.3532, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.00014025888127864108, | |
| "loss": 0.3531, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.00014014667691447777, | |
| "loss": 0.3532, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.00014003387892689548, | |
| "loss": 0.3529, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.00013992048842916485, | |
| "loss": 0.3529, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00013980673509364334, | |
| "loss": 0.3525, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.00013969216411821367, | |
| "loss": 0.3526, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.00013957700400522283, | |
| "loss": 0.3528, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.0001394612558912547, | |
| "loss": 0.3523, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00013934515417356422, | |
| "loss": 0.3525, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00013922823466086605, | |
| "loss": 0.3521, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.0001391107305894032, | |
| "loss": 0.352, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.0001389926431188934, | |
| "loss": 0.352, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.00013887421133451205, | |
| "loss": 0.3515, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.00013875496172903372, | |
| "loss": 0.3516, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.0001386351322358035, | |
| "loss": 0.3516, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.00013851472403749002, | |
| "loss": 0.3514, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.00013839398086947267, | |
| "loss": 0.3512, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.00013827241998328336, | |
| "loss": 0.3509, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00013815028397183435, | |
| "loss": 0.3515, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.00013802757404055875, | |
| "loss": 0.3509, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.00013790453853658956, | |
| "loss": 0.3509, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.00013778068554637154, | |
| "loss": 0.3509, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.00013765626228411418, | |
| "loss": 0.3504, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.00013753126997782485, | |
| "loss": 0.3507, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00013740570986112716, | |
| "loss": 0.351, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.00013727983599123606, | |
| "loss": 0.35, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00013715314510640312, | |
| "loss": 0.35, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00013702614521514468, | |
| "loss": 0.3502, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.00013689832855371767, | |
| "loss": 0.3502, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.0001367699503287586, | |
| "loss": 0.3499, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00013664101180730875, | |
| "loss": 0.3498, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00013651151426193926, | |
| "loss": 0.3494, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.00013638145897073862, | |
| "loss": 0.3497, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00013625084721730006, | |
| "loss": 0.3493, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0001361199431777653, | |
| "loss": 0.3498, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00013598822347904763, | |
| "loss": 0.3495, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.00013585595119916755, | |
| "loss": 0.3497, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00013572312764359914, | |
| "loss": 0.3493, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00013558975412325717, | |
| "loss": 0.3489, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.00013545583195448453, | |
| "loss": 0.3486, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00013532136245903893, | |
| "loss": 0.3492, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00013518634696408, | |
| "loss": 0.3484, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00013505078680215615, | |
| "loss": 0.3485, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.0001349149560595238, | |
| "loss": 0.3482, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.00013477831166543215, | |
| "loss": 0.3486, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.00013464112663151925, | |
| "loss": 0.3483, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.000134503402311746, | |
| "loss": 0.3479, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.00013436514006539586, | |
| "loss": 0.3479, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.00013422634125706136, | |
| "loss": 0.3477, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.00013408700725663073, | |
| "loss": 0.3478, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.00013394713943927426, | |
| "loss": 0.3475, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00013380702051639005, | |
| "loss": 0.3475, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.00013366637266202443, | |
| "loss": 0.3476, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.00013352491381339167, | |
| "loss": 0.3473, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00013338292669546847, | |
| "loss": 0.3472, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00013324041270961035, | |
| "loss": 0.3471, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.0001330979464650115, | |
| "loss": 0.347, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.00013295438506151565, | |
| "loss": 0.3468, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00013281030101961779, | |
| "loss": 0.3472, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.00013266569576136922, | |
| "loss": 0.3476, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.0001325205707139655, | |
| "loss": 0.3467, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.00013237492730973225, | |
| "loss": 0.3466, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.00013222876698611107, | |
| "loss": 0.3465, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00013208209118564532, | |
| "loss": 0.346, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.000131934901355966, | |
| "loss": 0.3463, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.00013178719894977722, | |
| "loss": 0.3461, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 0.0001316392823610171, | |
| "loss": 0.3459, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.00013149085814998288, | |
| "loss": 0.3459, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.00013134192777570792, | |
| "loss": 0.3459, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.00013119219272737532, | |
| "loss": 0.3455, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.00013104195243276762, | |
| "loss": 0.3452, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 0.0001308912083746959, | |
| "loss": 0.3455, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.0001307399620409431, | |
| "loss": 0.3449, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.00013058821492424954, | |
| "loss": 0.3455, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.00013043627351239063, | |
| "loss": 0.3448, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.00013028353032185398, | |
| "loss": 0.3448, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.00013013029085317334, | |
| "loss": 0.345, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.00012997655661876038, | |
| "loss": 0.3447, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 0.00012982232913590996, | |
| "loss": 0.3445, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.00012966760992678508, | |
| "loss": 0.3445, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.0001295124005184018, | |
| "loss": 0.3442, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.0001293567024426144, | |
| "loss": 0.344, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.00012920051723609998, | |
| "loss": 0.3439, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.0001290438464403435, | |
| "loss": 0.344, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.00012888700639334465, | |
| "loss": 0.3434, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.00012872937002614656, | |
| "loss": 0.3437, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.00012857125271973914, | |
| "loss": 0.3436, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 0.00012841265603467637, | |
| "loss": 0.3433, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.00012825358153624352, | |
| "loss": 0.3432, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.0001280943503701703, | |
| "loss": 0.3432, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.00012793432590746354, | |
| "loss": 0.3432, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.00012777382835231195, | |
| "loss": 0.343, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 0.0001276128592887616, | |
| "loss": 0.3429, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.00012745174365140025, | |
| "loss": 0.3427, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 0.00012728983727684876, | |
| "loss": 0.3426, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.00012712746417069506, | |
| "loss": 0.3425, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.00012696462593549614, | |
| "loss": 0.3426, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.0001268013241783996, | |
| "loss": 0.3421, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 0.00012663756051112788, | |
| "loss": 0.3421, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 0.00012647333654996226, | |
| "loss": 0.3419, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.00012630898373767156, | |
| "loss": 0.342, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 0.00012614417570078125, | |
| "loss": 0.3419, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 0.00012597858241938955, | |
| "loss": 0.3418, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.00012581253534795165, | |
| "loss": 0.3414, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 0.00012564603612528512, | |
| "loss": 0.3415, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 0.0001254790863946701, | |
| "loss": 0.3413, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 0.000125311687803833, | |
| "loss": 0.3412, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.00012514384200493036, | |
| "loss": 0.3414, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.00012497588768079038, | |
| "loss": 0.3414, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.00012480715332598534, | |
| "loss": 0.3411, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 0.00012463797674266635, | |
| "loss": 0.3409, | |
| "step": 140000 | |
| } | |
| ], | |
| "max_steps": 500000, | |
| "num_train_epochs": 9, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |