| { | |
| "best_metric": 0.9820850808867585, | |
| "best_model_checkpoint": "bert-base-uncased-finetuned-spam-real/run-0/checkpoint-1670", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 2505, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 1.6855844259262085, | |
| "learning_rate": 7.660219401214295e-06, | |
| "loss": 0.2399, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9690832834032355, | |
| "eval_loss": 0.1404273957014084, | |
| "eval_runtime": 321.4031, | |
| "eval_samples_per_second": 51.929, | |
| "eval_steps_per_second": 3.248, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.28391602635383606, | |
| "learning_rate": 6.618012679960649e-06, | |
| "loss": 0.1015, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 0.011191217228770256, | |
| "learning_rate": 5.575805958707004e-06, | |
| "loss": 0.0775, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9820850808867585, | |
| "eval_loss": 0.08610038459300995, | |
| "eval_runtime": 318.1892, | |
| "eval_samples_per_second": 52.453, | |
| "eval_steps_per_second": 3.281, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 0.1603321135044098, | |
| "learning_rate": 4.533599237453358e-06, | |
| "loss": 0.0478, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "grad_norm": 0.008322106674313545, | |
| "learning_rate": 3.4913925161997123e-06, | |
| "loss": 0.031, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9775314559616537, | |
| "eval_loss": 0.12152887880802155, | |
| "eval_runtime": 321.3639, | |
| "eval_samples_per_second": 51.935, | |
| "eval_steps_per_second": 3.249, | |
| "step": 2505 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 4175, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 5043584042416320.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 8.70242612246794e-06, | |
| "num_train_epochs": 5, | |
| "per_device_train_batch_size": 8, | |
| "seed": 20 | |
| } | |
| } | |