| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0212765957446808, | |
| "global_step": 60000, | |
| "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 | |
| } | |
| ], | |
| "max_steps": 500000, | |
| "num_train_epochs": 9, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |