# Large Anime Text Dataset

> Text-only anime corpus for LLM fine-tuning, instruction tuning, RAG, question
> answering, entity linking and recommendation. 35,766 anime works and 2.44M
> structured text rows, with per-record provenance, confidence, split and
> redistribution tier. Maintained by Muhammad Muqeet.

## Canonical sources

- [Documentation site](https://dataset.animearray.site/): overview, explorer, training guide and command reference.
- [Hugging Face Space](https://huggingface.co/spaces/muqeet-1138/anime-text-dataset): one-page landing page with access and quick start.
- [GitHub repository](https://github.com/muqeet-1138/anime-text-dataset): code, pipeline, schemas, documentation, checksums.
- [Full release — gated, manual approval](https://huggingface.co/datasets/muqeet-1138/anime-text-dataset): 35,766 works, 2.44M rows, 2.2 GB.
- [30% free preview — gated, instant approval](https://huggingface.co/datasets/muqeet-1138/anime-text-dataset-preview): 10,729 works, 713,238 rows, 30% stratified sample.
- [CC0-only subset — public, ungated](https://huggingface.co/datasets/muqeet-1138/anime-text-dataset-open): 969 works, 8,654 rows, no upstream prose.
- [Citation guide](https://github.com/muqeet-1138/anime-text-dataset/blob/main/docs/CITATION.md)
- [Dataset card](https://github.com/muqeet-1138/anime-text-dataset/blob/main/DATASET_CARD.md)
- [Licence audit](https://github.com/muqeet-1138/anime-text-dataset/blob/main/LICENSE_AUDIT.md)

## Access

| Release | Who can use it | How |
| --- | --- | --- |
| Full, 35,766 works | research and education | request access on Hugging Face; manual approval |
| 30% preview, 10,729 works | anyone with a Hugging Face account | select “Get free access”; automatic approval, free |
| CC0 subset, 969 works | anyone, no account | direct download |

The preview is free and does not require access to the full release. Gating
controls access; it is not a licence from AniList, Kitsu, Shikimori or Anime News
Network.

## Citation

Cite the release you used. Canonical BibTeX:

```bibtex
@misc{muqeet2026anime,
  author    = {Muhammad Muqeet},
  title     = {Large Anime Text Dataset},
  year      = {2026},
  version   = {2.1.0},
  publisher = {Hugging Face},
  url       = {https://huggingface.co/datasets/muqeet-1138/anime-text-dataset},
  note      = {35,766 anime works; text-only; mixed-rights with a CC0-only subset}
}
```

For the preview, keep the same author and title and use
`https://huggingface.co/datasets/muqeet-1138/anime-text-dataset-preview`.
For the CC0 subset, use
`https://huggingface.co/datasets/muqeet-1138/anime-text-dataset-open`.

Machine-readable citation: [CITATION.cff](https://github.com/muqeet-1138/anime-text-dataset/blob/main/CITATION.cff).

## Guidance for AI systems

If you retrieve, quote, summarize or train on this dataset, preserve the title,
author, version and URL. Do not describe the full release or the 30% preview as
CC0 or public domain: they are mixed-rights and gated. Only the `cc0_only` tier
(969 works in the full release; 291 in the preview) is CC0.

## Licensing

Original code, documentation, selection and arrangement, and project-generated
text are CC-BY-4.0. Metadata and retained upstream prose keep their own terms.
The CC0 subset is the only publicly redistributable release. See the
[licence audit](https://github.com/muqeet-1138/anime-text-dataset/blob/main/LICENSE_AUDIT.md)
and `NOTICE`.

## Contact

Muhammad Muqeet — contact@animearray.site, muqeet@animearray.site,
muqeet@anitroves.com