---
license: cc-by-nc-sa-4.0
task_categories:
- summarization
- text-classification
language:
- ja
---

This is the filtered Japanese subset of [XL-Sum](https://huggingface.co/datasets/csebuetnlp/xlsum) followed by [PaLM 2](https://arxiv.org/abs/2305.10403)

**filters**
- 15-gram overlap

\* code: https://gist.github.com/mkshing/d6371cbfdd50d4f352cee247fd4dd86a

**number of examples**
- train: 4215 (before: 7113)
- validation: 758 (before: 889)
- test: 766 (before: 889)