| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.3774104683195592, | |
| "eval_steps": 500, | |
| "global_step": 30500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5e-05, | |
| "loss": 1.3919, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.962080419845592e-05, | |
| "loss": 1.3494, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9241608396911834e-05, | |
| "loss": 1.3138, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8862412595367744e-05, | |
| "loss": 1.2302, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.848321679382366e-05, | |
| "loss": 1.126, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.8104020992279576e-05, | |
| "loss": 1.0285, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.772482519073549e-05, | |
| "loss": 0.8752, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.73456293891914e-05, | |
| "loss": 0.7864, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.696643358764732e-05, | |
| "loss": 0.719, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.658723778610323e-05, | |
| "loss": 0.6686, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.620804198455915e-05, | |
| "loss": 0.6423, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5828846183015065e-05, | |
| "loss": 0.6313, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5449650381470974e-05, | |
| "loss": 0.5966, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.507045457992689e-05, | |
| "loss": 0.5888, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.4691258778382806e-05, | |
| "loss": 0.5399, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.431206297683872e-05, | |
| "loss": 0.5202, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.393286717529464e-05, | |
| "loss": 0.4745, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.3553671373750554e-05, | |
| "loss": 0.5115, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.317447557220647e-05, | |
| "loss": 0.4806, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.2795279770662386e-05, | |
| "loss": 0.4516, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.2416083969118295e-05, | |
| "loss": 0.46, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.203688816757421e-05, | |
| "loss": 0.3857, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.165769236603013e-05, | |
| "loss": 0.4283, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.127849656448604e-05, | |
| "loss": 0.3975, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.089930076294195e-05, | |
| "loss": 0.3958, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.052010496139787e-05, | |
| "loss": 0.3672, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.0140909159853784e-05, | |
| "loss": 0.3591, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.97617133583097e-05, | |
| "loss": 0.3712, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.9382517556765616e-05, | |
| "loss": 0.3761, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.9003321755221525e-05, | |
| "loss": 0.3603, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.862412595367744e-05, | |
| "loss": 0.3545, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.824493015213336e-05, | |
| "loss": 0.3377, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.786573435058927e-05, | |
| "loss": 0.3292, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.748653854904518e-05, | |
| "loss": 0.346, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.71073427475011e-05, | |
| "loss": 0.3362, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.6728146945957015e-05, | |
| "loss": 0.3556, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.634895114441293e-05, | |
| "loss": 0.3123, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.5969755342868846e-05, | |
| "loss": 0.319, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.5590559541324756e-05, | |
| "loss": 0.3262, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.521136373978067e-05, | |
| "loss": 0.3134, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.483216793823659e-05, | |
| "loss": 0.3031, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4452972136692504e-05, | |
| "loss": 0.2965, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.407377633514842e-05, | |
| "loss": 0.2744, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.3694580533604336e-05, | |
| "loss": 0.3039, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.331538473206025e-05, | |
| "loss": 0.3266, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.293618893051617e-05, | |
| "loss": 0.2847, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.255699312897208e-05, | |
| "loss": 0.2637, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.217779732742799e-05, | |
| "loss": 0.3004, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.179860152588391e-05, | |
| "loss": 0.2636, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.1419405724339825e-05, | |
| "loss": 0.2452, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.104020992279574e-05, | |
| "loss": 0.2533, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.066101412125165e-05, | |
| "loss": 0.2854, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.0281818319707566e-05, | |
| "loss": 0.267, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9902622518163482e-05, | |
| "loss": 0.2452, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9523426716619395e-05, | |
| "loss": 0.2679, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.914423091507531e-05, | |
| "loss": 0.2559, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.8765035113531223e-05, | |
| "loss": 0.2499, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.838583931198714e-05, | |
| "loss": 0.2279, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.8006643510443055e-05, | |
| "loss": 0.2397, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7627447708898968e-05, | |
| "loss": 0.256, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.7248251907354884e-05, | |
| "loss": 0.2538, | |
| "step": 30500 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 66429, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 1.4181162326949888e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |