| from datasets import load_dataset | |
| import numpy as np | |
| def main(): | |
| raw_data = load_dataset("xglue", "nc") | |
| raw_data = raw_data.rename_column("news_body", "text") | |
| raw_data = raw_data.remove_columns(["news_title"]) | |
| raw_data = raw_data.rename_column("news_category", "label") | |
| labels = raw_data["train"].features["label"] | |
| for split, dataset in raw_data.items(): | |
| len_data = len(dataset) | |
| dataset = dataset.add_column("id", np.arange(len_data)) | |
| dataset = dataset.map(lambda x: {"label_text": labels.int2str(x["label"])}, num_proc=4) | |
| dataset.to_json(f"{split}.jsonl") | |
| if __name__ == "__main__": | |
| main() | |