| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.7659509280772079, | |
| "global_step": 45000, | |
| "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": 0.6484, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001499986182599296, | |
| "loss": 0.5436, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00014999689109762647, | |
| "loss": 0.5149, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00014999447309426713, | |
| "loss": 0.4996, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00014999136427371623, | |
| "loss": 0.4888, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001499875646666566, | |
| "loss": 0.4797, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00014998307431058887, | |
| "loss": 0.4711, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00014997789324983099, | |
| "loss": 0.463, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00014997202153551792, | |
| "loss": 0.4545, | |
| "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": 0.4337, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00014994162940397778, | |
| "loss": 0.4286, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00014993230542746872, | |
| "loss": 0.4242, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00014992229124733788, | |
| "loss": 0.4203, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0001499115869624212, | |
| "loss": 0.4173, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00014990019267836565, | |
| "loss": 0.4141, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00014988810850762824, | |
| "loss": 0.4116, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00014987533456947482, | |
| "loss": 0.4095, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000149861870989979, | |
| "loss": 0.4075, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0001498477758876907, | |
| "loss": 0.4057, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00014983293618814342, | |
| "loss": 0.4038, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00014981740726570863, | |
| "loss": 0.4021, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00014980118927365057, | |
| "loss": 0.4005, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00014978428237203426, | |
| "loss": 0.3991, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00014976668672772396, | |
| "loss": 0.3977, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00014974843976988143, | |
| "loss": 0.3968, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00014972946854455738, | |
| "loss": 0.3953, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00014970980911752974, | |
| "loss": 0.3945, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0001496894616828291, | |
| "loss": 0.3931, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0001496684264412764, | |
| "loss": 0.3921, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00014964670360048104, | |
| "loss": 0.3909, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00014962429337483864, | |
| "loss": 0.3903, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00014960119598552907, | |
| "loss": 0.3892, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00014957745991457127, | |
| "loss": 0.3885, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00014955303988623036, | |
| "loss": 0.3877, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00014952788514615529, | |
| "loss": 0.3871, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0001495020441939234, | |
| "loss": 0.3858, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0001494755172845744, | |
| "loss": 0.3854, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00014944830467991823, | |
| "loss": 0.3847, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00014942046312846905, | |
| "loss": 0.3839, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0001493918813157191, | |
| "loss": 0.3837, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0001493626146331152, | |
| "loss": 0.383, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00014933266336950772, | |
| "loss": 0.3821, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00014930208977431846, | |
| "loss": 0.382, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00014927077160993785, | |
| "loss": 0.381, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0001492387697710072, | |
| "loss": 0.3804, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0001492060845733717, | |
| "loss": 0.38, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00014917278375754084, | |
| "loss": 0.3793, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00014913880296235724, | |
| "loss": 0.3791, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00014910407237991193, | |
| "loss": 0.3787, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00014906865976817066, | |
| "loss": 0.3779, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00014903256547664165, | |
| "loss": 0.3778, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0001489958640925119, | |
| "loss": 0.377, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00014895840887839445, | |
| "loss": 0.3768, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.000148920273072621, | |
| "loss": 0.3761, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00014888145705157658, | |
| "loss": 0.3759, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0001488420408682806, | |
| "loss": 0.3755, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00014880186693119052, | |
| "loss": 0.3749, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00014876101394745023, | |
| "loss": 0.3747, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00014871948232026222, | |
| "loss": 0.3741, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00014867727245952692, | |
| "loss": 0.3739, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00014863447123337502, | |
| "loss": 0.3736, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00014859090751637814, | |
| "loss": 0.3733, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00014854666683481534, | |
| "loss": 0.3728, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0001485017496253243, | |
| "loss": 0.3725, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00014845624819224238, | |
| "loss": 0.3722, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00014840998061432743, | |
| "loss": 0.3719, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00014836313241655734, | |
| "loss": 0.3717, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00014831551629313194, | |
| "loss": 0.3711, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00014826722592314076, | |
| "loss": 0.3709, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00014821826178319038, | |
| "loss": 0.3709, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0001481687243030059, | |
| "loss": 0.3703, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00014811841542465117, | |
| "loss": 0.3698, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00014806743424503673, | |
| "loss": 0.3697, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0001480157812673262, | |
| "loss": 0.3692, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0001479634570013137, | |
| "loss": 0.369, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00014791046196341849, | |
| "loss": 0.3689, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0001478567966766803, | |
| "loss": 0.3685, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0001478024616707538, | |
| "loss": 0.3683, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00014774756815776608, | |
| "loss": 0.368, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00014769189666559521, | |
| "loss": 0.3679, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00014763555708173254, | |
| "loss": 0.3676, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0001475785499622271, | |
| "loss": 0.3669, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00014752087586971605, | |
| "loss": 0.3673, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0001474626527191014, | |
| "loss": 0.3668, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00014740364772589337, | |
| "loss": 0.3666, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00014734397748589426, | |
| "loss": 0.3665, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0001472836425880252, | |
| "loss": 0.366, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00014722264362776731, | |
| "loss": 0.3657, | |
| "step": 45000 | |
| } | |
| ], | |
| "max_steps": 500000, | |
| "num_train_epochs": 9, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |