| { | |
| "best_metric": 18.46313833181642, | |
| "best_model_checkpoint": "../models/a_bf1k_10/checkpoint-440", | |
| "epoch": 3.1, | |
| "eval_steps": 220, | |
| "global_step": 440, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 8.699485778808594, | |
| "learning_rate": 9.727272727272728e-06, | |
| "loss": 3.3147, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 8.136752128601074, | |
| "learning_rate": 8.919191919191919e-06, | |
| "loss": 1.342, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "eval_LhPD_(mipa)": 7.1986371929013275, | |
| "eval_cer": 0.6239889864366144, | |
| "eval_cer_norm": 0.5221072755036582, | |
| "eval_loss": 1.3712860345840454, | |
| "eval_lvnshtn": 21.44000000000003, | |
| "eval_ped": 19.62285714285713, | |
| "eval_per": 0.6381009167551055, | |
| "eval_pfer": 20.690317459549295, | |
| "eval_runtime": 739.4673, | |
| "eval_samples_per_second": 0.473, | |
| "eval_steps_per_second": 0.06, | |
| "eval_time": 5664.729516983032, | |
| "eval_wefed": 65.59250000000009, | |
| "eval_wefer": 1.8089760650084719, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 7.045517921447754, | |
| "learning_rate": 7.808080808080808e-06, | |
| "loss": 0.9317, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "grad_norm": 6.565656661987305, | |
| "learning_rate": 6.6969696969696975e-06, | |
| "loss": 0.7632, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "eval_LhPD_(mipa)": 6.457400766165471, | |
| "eval_cer": 0.5521492110548145, | |
| "eval_cer_norm": 0.45695684003058124, | |
| "eval_loss": 1.075352430343628, | |
| "eval_lvnshtn": 19.11999999999999, | |
| "eval_ped": 17.90285714285715, | |
| "eval_per": 0.5793257313990053, | |
| "eval_pfer": 18.46313833181642, | |
| "eval_runtime": 758.5584, | |
| "eval_samples_per_second": 0.461, | |
| "eval_steps_per_second": 0.058, | |
| "eval_time": 11416.746544837952, | |
| "eval_wefed": 63.63178571428568, | |
| "eval_wefer": 1.6309992323741251, | |
| "step": 440 | |
| } | |
| ], | |
| "logging_steps": 110, | |
| "max_steps": 1100, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 220, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.81607989248e+18, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |