| { | |
| "best_global_step": 21000, | |
| "best_metric": 0.919884471379488, | |
| "best_model_checkpoint": "./saved_best/sst2_BART-20000/full_text/checkpoint-21000", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 21000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.40724968910217285, | |
| "learning_rate": 1.3334285714285714e-05, | |
| "loss": 2.167, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9023333333333333, | |
| "eval_f1_macro": 0.9016989529876082, | |
| "eval_f1_weighted": 0.9025202254683811, | |
| "eval_loss": 1.5102022886276245, | |
| "eval_runtime": 400.0393, | |
| "eval_samples_per_second": 14.999, | |
| "eval_steps_per_second": 7.499, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.08096525818109512, | |
| "learning_rate": 6.667619047619049e-06, | |
| "loss": 1.3183, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.916, | |
| "eval_f1_macro": 0.9155205698472799, | |
| "eval_f1_weighted": 0.9161824380227166, | |
| "eval_loss": 1.3292620182037354, | |
| "eval_runtime": 400.5613, | |
| "eval_samples_per_second": 14.979, | |
| "eval_steps_per_second": 7.489, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.19997699558734894, | |
| "learning_rate": 9.523809523809524e-10, | |
| "loss": 0.8244, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.921, | |
| "eval_f1_macro": 0.919884471379488, | |
| "eval_f1_weighted": 0.9208676491467189, | |
| "eval_loss": 1.2811408042907715, | |
| "eval_runtime": 399.4772, | |
| "eval_samples_per_second": 15.02, | |
| "eval_steps_per_second": 7.51, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 21000, | |
| "total_flos": 1559792198160000.0, | |
| "train_loss": 1.4365767531622025, | |
| "train_runtime": 5688.3135, | |
| "train_samples_per_second": 7.384, | |
| "train_steps_per_second": 3.692 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.921, | |
| "eval_f1_macro": 0.919884471379488, | |
| "eval_f1_weighted": 0.9208676491467189, | |
| "eval_loss": 1.2811408042907715, | |
| "eval_runtime": 399.7755, | |
| "eval_samples_per_second": 15.008, | |
| "eval_steps_per_second": 7.504, | |
| "step": 21000 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 21000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1559792198160000.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |