# htext tanakh-v1: the Westminster Leningrad Codex, pointed and plain (HeOCR)

The Hebrew Bible from the Westminster Leningrad Codex, cut into lines, pointed (niqqud) and plain. Part of htext, HeOCR's public-domain Hebrew text for training Hebrew OCR.

- Version: `tanakh-v1`.
- Derived from the GitHub release tanakh-v1, not byte-identical to it: lines of eval-reserved units removed (reservations v7, sha256 91fd1ea5099d9b1cc3cd5cbaf22ab93158c8357dda03d266e17b1ef44c3caf34), and lines sharing text with HeOCR's evaluation sets removed, as were whole units with long shared runs; HeOCR's evaluation ran its own independent check on every lines file and found 0 overlap. The per-file counts are in the Hugging Face card.
- Licence: CC0-1.0 (htext-cc0).
- Credit: Westminster Leningrad Codex (public domain), via the Open Scriptures Hebrew Bible (https://github.com/openscriptures/morphhb).
- Text quality: transcription.
- `card_*` files are the GitHub release's own corpus card, rights index and checksums.

| File | Size |
| --- | ---: |
| `SHA256SUMS` | 836 bytes |
| `card_CORPUS_CARD.md` | 2,692 bytes |
| `card_KNOWN_ISSUES.md` | 1,517 bytes |
| `card_SHA256SUMS` | 543 bytes |
| `card_corpus_card.json` | 10,760 bytes |
| `card_tanakh_rights_index.v1.jsonl.gz` | 3,071 bytes |
| `tanakh_line_index_niqqud.v1.jsonl.gz` | 2,402,091 bytes |
| `tanakh_line_index_plain.v1.jsonl.gz` | 2,499,998 bytes |
| `tanakh_lines_niqqud_v1.txt` | 4,371,119 bytes |
| `tanakh_lines_plain_v1.txt` | 2,661,187 bytes |

- GitHub release: https://github.com/HeOCR/htext/releases/tag/tanakh-v1
- Hugging Face (the same files): https://huggingface.co/datasets/HeOCR/htext-tanakh/tree/tanakh-v1
- Documentation: https://github.com/HeOCR/htext/blob/main/docs/tanakh.md
- htext: https://heocr.org/htext

The rights fields are the project's reading of the law, not legal advice.
