8 lines
212 B
Python
8 lines
212 B
Python
from utils import process_dataset
|
|
|
|
if __name__ == "__main__":
|
|
process_dataset(
|
|
dataset_name="Blaze7451/enwiki_structured_content",
|
|
output_subdir="english-wiki",
|
|
max_chunk_size=5,
|
|
) |