| { | |
| "best_metric": 0.31043142080307007, | |
| "best_model_checkpoint": "llama_sequence_classification_1e-05/checkpoint-1030", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 1030, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.8064619898796082, | |
| "eval_pearson": 0.703103762183237, | |
| "eval_runtime": 15.2268, | |
| "eval_samples_per_second": 61.208, | |
| "eval_steps_per_second": 7.684, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.499979704618454, | |
| "eval_pearson": 0.8414151685022249, | |
| "eval_runtime": 15.4193, | |
| "eval_samples_per_second": 60.444, | |
| "eval_steps_per_second": 7.588, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 2.4271844660194173, | |
| "grad_norm": 9.565049171447754, | |
| "learning_rate": 5.145631067961165e-06, | |
| "loss": 0.8835, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.3854205012321472, | |
| "eval_pearson": 0.898057930574925, | |
| "eval_runtime": 15.2609, | |
| "eval_samples_per_second": 61.071, | |
| "eval_steps_per_second": 7.667, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.32877469062805176, | |
| "eval_pearson": 0.9038033088607682, | |
| "eval_runtime": 15.1169, | |
| "eval_samples_per_second": 61.653, | |
| "eval_steps_per_second": 7.74, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 4.854368932038835, | |
| "grad_norm": 7.450101852416992, | |
| "learning_rate": 2.9126213592233014e-07, | |
| "loss": 0.3221, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.31043142080307007, | |
| "eval_pearson": 0.9176986697209388, | |
| "eval_runtime": 15.6441, | |
| "eval_samples_per_second": 59.575, | |
| "eval_steps_per_second": 7.479, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 1030, | |
| "total_flos": 5.610584124311962e+16, | |
| "train_loss": 0.592618284873592, | |
| "train_runtime": 1052.021, | |
| "train_samples_per_second": 31.202, | |
| "train_steps_per_second": 0.979 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 1030, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "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": 5.610584124311962e+16, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |