---
configs:
- config_name: default
  data_files:
  - split: train
    path: data.parquet
language:
- zh
---
# Wikipedia-zhtw Filtered
使用 [erhwenkuo/c4-chinese-zhtw](https://huggingface.co/datasets/erhwenkuo/c4-chinese-zhtw) 前 20% 的資料和搭配 [huggingface/datatrove](https://github.com/huggingface/datatrove) 再次清洗資料，移除許多過短、重複和包含雜訊的文章。