| { | |
| "best_metric": 0.2617274224758148, | |
| "best_model_checkpoint": "xlm-roberta-large-all-full-finetuned-toxicity-classification/checkpoint-4016", | |
| "epoch": 2.9987546699875467, | |
| "eval_steps": 500, | |
| "global_step": 6021, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.24906600249066002, | |
| "grad_norm": 13.982502937316895, | |
| "learning_rate": 1.83391463212091e-05, | |
| "loss": 0.4913, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.49813200498132004, | |
| "grad_norm": 9.81984806060791, | |
| "learning_rate": 1.6678292642418204e-05, | |
| "loss": 0.3674, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7471980074719801, | |
| "grad_norm": 9.316424369812012, | |
| "learning_rate": 1.5017438963627307e-05, | |
| "loss": 0.3317, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.9962640099626401, | |
| "grad_norm": 13.302652359008789, | |
| "learning_rate": 1.3356585284836408e-05, | |
| "loss": 0.3147, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.2968192994594574, | |
| "eval_runtime": 142.521, | |
| "eval_samples_per_second": 50.091, | |
| "eval_steps_per_second": 3.136, | |
| "step": 2008 | |
| }, | |
| { | |
| "epoch": 1.2450809464508095, | |
| "grad_norm": 16.878368377685547, | |
| "learning_rate": 1.1695731606045508e-05, | |
| "loss": 0.2468, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.4941469489414696, | |
| "grad_norm": 10.831304550170898, | |
| "learning_rate": 1.0034877927254609e-05, | |
| "loss": 0.237, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.7432129514321295, | |
| "grad_norm": 21.390304565429688, | |
| "learning_rate": 8.374024248463712e-06, | |
| "loss": 0.2271, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.9922789539227894, | |
| "grad_norm": 10.786124229431152, | |
| "learning_rate": 6.7131705696728125e-06, | |
| "loss": 0.2278, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.2617274224758148, | |
| "eval_runtime": 142.583, | |
| "eval_samples_per_second": 50.069, | |
| "eval_steps_per_second": 3.135, | |
| "step": 4016 | |
| }, | |
| { | |
| "epoch": 2.241095890410959, | |
| "grad_norm": 10.759682655334473, | |
| "learning_rate": 5.0523168908819146e-06, | |
| "loss": 0.1553, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.490161892901619, | |
| "grad_norm": 7.604818344116211, | |
| "learning_rate": 3.391463212091015e-06, | |
| "loss": 0.1573, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.739227895392279, | |
| "grad_norm": 5.243312358856201, | |
| "learning_rate": 1.7306095333001162e-06, | |
| "loss": 0.1422, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.988293897882939, | |
| "grad_norm": 11.100104331970215, | |
| "learning_rate": 6.975585450921775e-08, | |
| "loss": 0.1439, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.9987546699875467, | |
| "eval_loss": 0.31470784544944763, | |
| "eval_runtime": 142.6105, | |
| "eval_samples_per_second": 50.059, | |
| "eval_steps_per_second": 3.134, | |
| "step": 6021 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 6021, | |
| "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": 1.7951793851787264e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |