| { |
| "data": { |
| "dataset_name": "Maxscha/commitbench", |
| "cache_dir": null, |
| "prefix": "generate commit message: ", |
| "target_mode": "full", |
| "max_source_length": 512, |
| "max_target_length": 64, |
| "min_diff_chars": 1, |
| "min_message_chars": 1, |
| "num_proc": 8, |
| "max_train_samples": 500000, |
| "max_eval_samples": 2000, |
| "max_predict_samples": null |
| }, |
| "train": { |
| "model_name": "google-t5/t5-small", |
| "output_dir": "outputs/t5-small-commitbench", |
| "seed": 42, |
| "deterministic": false, |
| "learning_rate": 3e-05, |
| "num_train_epochs": 2.0, |
| "per_device_train_batch_size": 32, |
| "per_device_eval_batch_size": 64, |
| "gradient_accumulation_steps": 1, |
| "weight_decay": 0.01, |
| "warmup_ratio": 0.05, |
| "label_smoothing_factor": 0.0, |
| "max_grad_norm": 1.0, |
| "lr_scheduler_type": "linear", |
| "eval_strategy": "steps", |
| "save_strategy": "steps", |
| "eval_steps": 3000, |
| "save_steps": 3000, |
| "logging_steps": 250, |
| "save_total_limit": 2, |
| "metric_for_best_model": "eval_loss", |
| "greater_is_better": false, |
| "load_best_model_at_end": true, |
| "gradient_checkpointing": false, |
| "torch_compile": true, |
| "torch_compile_mode": "", |
| "group_by_length": true, |
| "dataloader_num_workers": 4, |
| "precision": "auto", |
| "resume_from_checkpoint": null, |
| "max_steps": -1 |
| }, |
| "generation": { |
| "num_beams": 4, |
| "max_new_tokens": 64, |
| "min_new_tokens": 0, |
| "length_penalty": 1.0, |
| "no_repeat_ngram_size": 3, |
| "early_stopping": true, |
| "do_sample": false, |
| "eval_num_beams": 1 |
| } |
| } |
