Function load_and_process_dataset

dflash/benchmark.py:84–93

(data_name: str)

def load_and_process_dataset(data_name: str) -> list[dict]:
    if data_name not in DATASETS:
        raise ValueError(f"Unknown dataset '{data_name}'. Available: {list(DATASETS.keys())}")

    path = CACHE_DIR / f"{data_name}.jsonl"
    if not path.exists():
        _prepare_dataset(data_name)

    with open(path) as f:
        return [json.loads(line) for line in f]


def _limit_dataset(dataset: list[dict], max_samples: int | None) -> list[dict]: