| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.3104384943733023, | |
| "eval_steps": 10, | |
| "global_step": 50, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.006208769887466045, | |
| "eval_loss": 1.4452153444290161, | |
| "eval_runtime": 15.5313, | |
| "eval_samples_per_second": 69.859, | |
| "eval_steps_per_second": 8.757, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.06208769887466046, | |
| "grad_norm": 0.7810622453689575, | |
| "learning_rate": 9.330127018922194e-05, | |
| "loss": 1.2914, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06208769887466046, | |
| "eval_loss": 1.1287556886672974, | |
| "eval_runtime": 15.6194, | |
| "eval_samples_per_second": 69.465, | |
| "eval_steps_per_second": 8.707, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.12417539774932092, | |
| "grad_norm": 0.666254460811615, | |
| "learning_rate": 6.91341716182545e-05, | |
| "loss": 0.9854, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.12417539774932092, | |
| "eval_loss": 0.9413304924964905, | |
| "eval_runtime": 15.523, | |
| "eval_samples_per_second": 69.896, | |
| "eval_steps_per_second": 8.761, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.18626309662398138, | |
| "grad_norm": 0.4954511523246765, | |
| "learning_rate": 3.705904774487396e-05, | |
| "loss": 0.9005, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.18626309662398138, | |
| "eval_loss": 0.8872678875923157, | |
| "eval_runtime": 15.5192, | |
| "eval_samples_per_second": 69.913, | |
| "eval_steps_per_second": 8.763, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.24835079549864184, | |
| "grad_norm": 0.47105929255485535, | |
| "learning_rate": 1.0332332985438248e-05, | |
| "loss": 0.8699, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.24835079549864184, | |
| "eval_loss": 0.8705499172210693, | |
| "eval_runtime": 15.5382, | |
| "eval_samples_per_second": 69.828, | |
| "eval_steps_per_second": 8.753, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.3104384943733023, | |
| "grad_norm": 0.47211363911628723, | |
| "learning_rate": 0.0, | |
| "loss": 0.8423, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.3104384943733023, | |
| "eval_loss": 0.867393970489502, | |
| "eval_runtime": 15.5747, | |
| "eval_samples_per_second": 69.664, | |
| "eval_steps_per_second": 8.732, | |
| "step": 50 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 50, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 13, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.92436412350464e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |