| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.011291460832745237, | |
| "eval_steps": 2, | |
| "global_step": 5, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0022582921665490474, | |
| "grad_norm": 0.45662736892700195, | |
| "learning_rate": 1e-05, | |
| "loss": 1.6438, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0022582921665490474, | |
| "eval_loss": 1.4655977487564087, | |
| "eval_runtime": 77.9257, | |
| "eval_samples_per_second": 9.573, | |
| "eval_steps_per_second": 4.787, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004516584333098095, | |
| "grad_norm": 0.4056294858455658, | |
| "learning_rate": 2e-05, | |
| "loss": 1.814, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.004516584333098095, | |
| "eval_loss": 1.4531971216201782, | |
| "eval_runtime": 8.185, | |
| "eval_samples_per_second": 91.142, | |
| "eval_steps_per_second": 45.571, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.006774876499647142, | |
| "grad_norm": 0.4204471707344055, | |
| "learning_rate": 3e-05, | |
| "loss": 1.7214, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00903316866619619, | |
| "grad_norm": 0.5042304992675781, | |
| "learning_rate": 4e-05, | |
| "loss": 1.6533, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00903316866619619, | |
| "eval_loss": 1.4372713565826416, | |
| "eval_runtime": 8.2455, | |
| "eval_samples_per_second": 90.473, | |
| "eval_steps_per_second": 45.237, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.011291460832745237, | |
| "grad_norm": 0.35187670588493347, | |
| "learning_rate": 5e-05, | |
| "loss": 1.839, | |
| "step": 5 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 5, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 2, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 753993451044864.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |