| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.1190171487074485, | |
| "global_step": 20000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.893353809401928e-05, | |
| "loss": 5.7923, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.7867076188038565e-05, | |
| "loss": 5.0624, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.680061428205785e-05, | |
| "loss": 4.1352, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.573415237607713e-05, | |
| "loss": 3.453, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.466769047009641e-05, | |
| "loss": 3.0505, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.360122856411569e-05, | |
| "loss": 2.7572, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.253476665813497e-05, | |
| "loss": 2.5325, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.146830475215426e-05, | |
| "loss": 2.3672, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.040184284617354e-05, | |
| "loss": 2.2281, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.933538094019282e-05, | |
| "loss": 2.1289, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 3.8268919034212095e-05, | |
| "loss": 2.0405, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 3.720245712823138e-05, | |
| "loss": 1.9581, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.6135995222250665e-05, | |
| "loss": 1.8997, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.506953331626994e-05, | |
| "loss": 1.8383, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.400307141028923e-05, | |
| "loss": 1.7812, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.2936609504308506e-05, | |
| "loss": 1.7441, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.187014759832779e-05, | |
| "loss": 1.6926, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 3.0803685692347076e-05, | |
| "loss": 1.657, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.9737223786366354e-05, | |
| "loss": 1.6246, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.8670761880385632e-05, | |
| "loss": 1.5971, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 2.7604299974404917e-05, | |
| "loss": 1.5715, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.6537838068424198e-05, | |
| "loss": 1.5444, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 2.5471376162443476e-05, | |
| "loss": 1.5255, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.440491425646276e-05, | |
| "loss": 1.5065, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.3338452350482043e-05, | |
| "loss": 1.4829, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 2.2271990444501324e-05, | |
| "loss": 1.46, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 2.1205528538520606e-05, | |
| "loss": 1.4434, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 2.0139066632539887e-05, | |
| "loss": 1.4356, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 1.907260472655917e-05, | |
| "loss": 1.4199, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 1.800614282057845e-05, | |
| "loss": 1.3948, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.693968091459773e-05, | |
| "loss": 1.3924, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.5873219008617013e-05, | |
| "loss": 1.3707, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 1.4806757102636296e-05, | |
| "loss": 1.3681, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 1.3740295196655576e-05, | |
| "loss": 1.3581, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 1.2673833290674857e-05, | |
| "loss": 1.3525, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 1.1607371384694139e-05, | |
| "loss": 1.3468, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 1.054090947871342e-05, | |
| "loss": 1.3392, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 9.474447572732703e-06, | |
| "loss": 1.3218, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 8.407985666751985e-06, | |
| "loss": 1.3237, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 7.3415237607712656e-06, | |
| "loss": 1.3108, | |
| "step": 20000 | |
| } | |
| ], | |
| "max_steps": 23442, | |
| "num_train_epochs": 6, | |
| "total_flos": 8.486152068857856e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |