| language: | |
| - zht | |
| - zh | |
| size_categories: | |
| - 100K<n<1M | |
| task_categories: | |
| - text-generation | |
| - text2text-generation | |
| - summarization | |
| pretty_name: Traditional Chinese News Collection | |
| dataset_info: | |
| features: | |
| - name: title | |
| dtype: string | |
| - name: text | |
| dtype: string | |
| - name: category | |
| dtype: string | |
| splits: | |
| - name: train | |
| num_bytes: 2410523754 | |
| num_examples: 584884 | |
| download_size: 1628926746 | |
| dataset_size: 2410523754 | |
| configs: | |
| - config_name: default | |
| data_files: | |
| - split: train | |
| path: data/train-* | |
| # Dataset Card for Traditional Chinese News Collection | |
| - Contains common news/magazines/articles available online in Traditional Chinese. | |
| - Provides `title`, `text` (content), and `category` for each sample. | |
| - Note: `category` is labeled according to the source of the news. | |
| - Cleaned with custom rules and de-deplicated using [MinHash](https://en.wikipedia.org/wiki/MinHash). | |
| ## Dataset Details | |
| Dataset size: 557,764 samples. | |
| Available labels: | |
| - article | |
| - tech | |
| - science | |
| - daily-weekly | |
| Dataset source: | |
| - [benchang1110/technewstw](https://huggingface.co/datasets/benchang1110/technewstw) | |
| - [benchang1110/newstw](https://huggingface.co/datasets/benchang1110/newstw) | |
| - [benchang1110/humantw](benchang1110/humantw) | |
| - [benchang1110/sciencetw](https://huggingface.co/datasets/benchang1110/sciencetw) | |
| - The World Wide Web |