| { | |
| "best_metric": 0.8279816513761468, | |
| "best_model_checkpoint": "microsoft/Multilingual-MiniLM-L12-H384-finetuned-sst2/run-4/checkpoint-3368", | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 3368, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 207.84652709960938, | |
| "learning_rate": 1.6360069774388828e-05, | |
| "loss": 0.5581, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 1.0724189281463623, | |
| "learning_rate": 1.4563049807957717e-05, | |
| "loss": 0.5489, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 2.0987462997436523, | |
| "learning_rate": 1.2766029841526607e-05, | |
| "loss": 0.5061, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.823394495412844, | |
| "eval_loss": 0.4638556241989136, | |
| "eval_runtime": 1.0938, | |
| "eval_samples_per_second": 797.217, | |
| "eval_steps_per_second": 50.283, | |
| "step": 1684 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "grad_norm": 2.370238780975342, | |
| "learning_rate": 1.0969009875095499e-05, | |
| "loss": 0.458, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 25.474912643432617, | |
| "learning_rate": 9.171989908664387e-06, | |
| "loss": 0.4448, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 2.603041887283325, | |
| "learning_rate": 7.374969942233276e-06, | |
| "loss": 0.4212, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8279816513761468, | |
| "eval_loss": 0.5386698842048645, | |
| "eval_runtime": 1.024, | |
| "eval_samples_per_second": 851.56, | |
| "eval_steps_per_second": 53.711, | |
| "step": 3368 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 5052, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 43392374384520.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 1.8157089740819938e-05, | |
| "num_train_epochs": 3, | |
| "per_device_train_batch_size": 4, | |
| "seed": 35 | |
| } | |
| } | |