| """Fetch the exact released pilot dataset into the training script's layout.""" | |
| from pathlib import Path | |
| from huggingface_hub import hf_hub_download | |
| import shutil | |
| out=Path('data/sft');out.mkdir(parents=True,exist_ok=True) | |
| for remote,local in [('train.jsonl','train.jsonl'),('validation.jsonl','val.jsonl')]: | |
| path=hf_hub_download('guilindev/pacman-decision-oracle-v1',remote,repo_type='dataset', | |
| revision='2863a89c91bb31bcfd0d6ade1834b696a5870c79') | |
| shutil.copy(path,out/local) | |
| print('Downloaded 4096 training / 512 validation rows') | |