| 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() |
|
|
|
|
