| { | |
| "best_metric": 38.048973083496094, | |
| "best_model_checkpoint": "/content/drive/MyDrive/Colab Notebooks/results_experiment_4/checkpoint-210", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 210, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 1462.05419921875, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 50.9406, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 45.00191116333008, | |
| "eval_runtime": 68.5228, | |
| "eval_samples_per_second": 1.022, | |
| "eval_steps_per_second": 0.131, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 3524.53125, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 44.4536, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 39.292198181152344, | |
| "eval_runtime": 68.6397, | |
| "eval_samples_per_second": 1.02, | |
| "eval_steps_per_second": 0.131, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 3753.434814453125, | |
| "learning_rate": 0.0, | |
| "loss": 41.8109, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 38.048973083496094, | |
| "eval_runtime": 91.0281, | |
| "eval_samples_per_second": 0.769, | |
| "eval_steps_per_second": 0.099, | |
| "step": 210 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 210, | |
| "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": 220885183365120.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |