| { | |
| "best_metric": 0.8555758683729433, | |
| "best_model_checkpoint": "berttiny-hate_speech18-bothpretrained\\run-2\\checkpoint-240", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 960, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.5453252643738987e-06, | |
| "loss": 0.637, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8555758683729433, | |
| "eval_loss": 0.5853705406188965, | |
| "eval_runtime": 10.8628, | |
| "eval_samples_per_second": 201.422, | |
| "eval_steps_per_second": 1.657, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6968835095825992e-06, | |
| "loss": 0.5701, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8555758683729433, | |
| "eval_loss": 0.545884907245636, | |
| "eval_runtime": 11.0146, | |
| "eval_samples_per_second": 198.645, | |
| "eval_steps_per_second": 1.634, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 8.484417547912996e-07, | |
| "loss": 0.5422, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8555758683729433, | |
| "eval_loss": 0.5303202867507935, | |
| "eval_runtime": 10.9437, | |
| "eval_samples_per_second": 199.933, | |
| "eval_steps_per_second": 1.645, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.5316, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8555758683729433, | |
| "eval_loss": 0.5260117650032043, | |
| "eval_runtime": 10.9305, | |
| "eval_samples_per_second": 200.173, | |
| "eval_steps_per_second": 1.647, | |
| "step": 960 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 960, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "total_flos": 6253424431104.0, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 3.3937670191651984e-06, | |
| "num_train_epochs": 4, | |
| "per_device_train_batch_size": 32, | |
| "seed": 33 | |
| } | |
| } | |