File size: 604 Bytes
1ffce31 |
1 2 3 4 5 6 7 8 9 10 11 12 13 |
from datasets import load_dataset
import os
from kowiki import process_text
for date in ['20230420', '20230501', '20230520', '20230601', '20230620', '20230701', '20230720', '20230801']:
dataset = load_dataset("psyche/kowiki", date, streaming=False, split="train")
def process_function(example):
example["text"] = process_text(example["text"])
return example
dataset = dataset.map(process_function)
os.makedirs(f"data/{date}", exist_ok=True)
for i in range(10):
dataset.shard(num_shards=10, index=i).to_parquet(f"data/{date}/train-{i:05d}-of-{10:05d}.parquet")
|