| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.002294630564479119, | |
| "eval_steps": 2, | |
| "global_step": 5, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0004589261128958238, | |
| "grad_norm": 3.2609922885894775, | |
| "learning_rate": 1e-05, | |
| "loss": 4.156, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0004589261128958238, | |
| "eval_loss": 0.27827903628349304, | |
| "eval_runtime": 243.9166, | |
| "eval_samples_per_second": 15.046, | |
| "eval_steps_per_second": 7.523, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0009178522257916476, | |
| "grad_norm": 3.390610694885254, | |
| "learning_rate": 2e-05, | |
| "loss": 4.0756, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0009178522257916476, | |
| "eval_loss": 0.2703548073768616, | |
| "eval_runtime": 140.5368, | |
| "eval_samples_per_second": 26.114, | |
| "eval_steps_per_second": 13.057, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0013767783386874712, | |
| "grad_norm": 4.031177520751953, | |
| "learning_rate": 3e-05, | |
| "loss": 4.4294, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0018357044515832951, | |
| "grad_norm": 3.4328739643096924, | |
| "learning_rate": 4e-05, | |
| "loss": 4.1462, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0018357044515832951, | |
| "eval_loss": 0.26450106501579285, | |
| "eval_runtime": 483.9395, | |
| "eval_samples_per_second": 7.584, | |
| "eval_steps_per_second": 3.792, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.002294630564479119, | |
| "grad_norm": 4.37693452835083, | |
| "learning_rate": 5e-05, | |
| "loss": 4.9677, | |
| "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": 3298217215131648.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |