| from datasets import load_dataset | |
| def main(): | |
| raw_data = load_dataset("amazon_reviews_multi", "de") | |
| raw_data = raw_data.rename_column("review_id", "id") | |
| raw_data = raw_data.rename_column("review_body", "text") | |
| raw_data = raw_data.rename_column("stars", "label") | |
| raw_data = raw_data.remove_columns(["product_id", "reviewer_id", "review_title", "language", "product_category"]) | |
| for split, dataset in raw_data.items(): | |
| dataset = dataset.map(lambda x: {"label_text": str(x["label"])}, num_proc=4) | |
| dataset.to_json(f"{split}.jsonl") | |
| if __name__ == "__main__": | |
| main() | |