| { | |
| "best_metric": 0.8555758683729433, | |
| "best_model_checkpoint": "berttiny-hate_speech18-bothpretrained\\run-0\\checkpoint-120", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 480, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.665803895748027e-05, | |
| "loss": 0.5773, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8555758683729433, | |
| "eval_loss": 0.5026670098304749, | |
| "eval_runtime": 10.8208, | |
| "eval_samples_per_second": 202.204, | |
| "eval_steps_per_second": 1.663, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.1105359304986847e-05, | |
| "loss": 0.4888, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8555758683729433, | |
| "eval_loss": 0.49199947714805603, | |
| "eval_runtime": 10.9074, | |
| "eval_samples_per_second": 200.598, | |
| "eval_steps_per_second": 1.65, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 5.5526796524934235e-06, | |
| "loss": 0.4831, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8555758683729433, | |
| "eval_loss": 0.4904670715332031, | |
| "eval_runtime": 11.1022, | |
| "eval_samples_per_second": 197.078, | |
| "eval_steps_per_second": 1.621, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.4814, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8555758683729433, | |
| "eval_loss": 0.4893935024738312, | |
| "eval_runtime": 10.9723, | |
| "eval_samples_per_second": 199.412, | |
| "eval_steps_per_second": 1.64, | |
| "step": 480 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 480, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "total_flos": 8009349605088.0, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 2.2210718609973694e-05, | |
| "num_train_epochs": 4, | |
| "per_device_train_batch_size": 64, | |
| "seed": 16 | |
| } | |
| } | |