README.md

metadata

dataset_info:
  - config_name: default
    features:
      - name: translation_id
        dtype: string
      - name: translation
        dtype: string
      - name: language
        dtype: string
      - name: book_number
        dtype: int32
      - name: book
        dtype: string
      - name: book_abbr
        dtype: string
      - name: chapter
        dtype: int32
      - name: verse
        dtype: string
      - name: verse_number
        dtype: int32
      - name: verse_suffix
        dtype: string
      - name: verse_id
        dtype: string
      - name: text
        dtype: string
    splits:
      - name: train
        num_examples: 23188
    num_rows: 23188
configs:
  - config_name: default
    data_files: data/data.parquet

Aleppo Codex (כתר ארם צובא)

Description

The Aleppo Codex (Keter Aram Tzova) — the oldest and most authoritative manuscript of the Hebrew Bible, dating to approximately 930 CE. It was produced in Tiberias by the scribe Shlomo ben Buya'a, with vocalization added by Aaron ben Moses ben Asher, the most renowned Masorete. The codex represents the pinnacle of the Ben Asher Masoretic tradition.

Despite suffering damage in 1947 (approximately 40% of the manuscript was lost), the surviving portions are considered the most accurate witness to the Ben Asher text. It is the only surviving manuscript that ben Asher himself vocalized and annotated.

Important: This text is consonants only — without vocalization (niqqud) or cantillation marks (te'amim). The surviving Aleppo Codex pages cover only part of the Tanakh, but the electronic transcription (from Mechon Mamre) fills in the missing sections from other Masoretic manuscripts to create a complete Old Testament text.

Source: The electronic edition at Mechon Mamre, a digital transcription of the Aleppo Codex.

License: Public Domain.

Dataset Structure

Each row represents one Bible verse.

Column	Type	Description
translation_id	string	Translation identifier: aleppo_codex_he
translation	string	Full translation name
language	string	Language code: he
book_number	int32	Book number (1-39 for Old Testament)
book	string	Book name (English)
book_abbr	string	Book abbreviation
chapter	int32	Chapter number
verse	string	Verse number
verse_number	int32	Numeric part of verse
verse_suffix	string	Verse suffix or null
verse_id	string	Unique verse identifier ({book_abbr}_{chapter}_{verse})
text	string	Verse text in Hebrew script (consonants only, no niqqud)

Statistics

• Books: 39 (complete Old Testament / Tanakh)
• Verses: 23,188
• Language: Biblical Hebrew (hbo)
• Format: Parquet (ZSTD compressed)
• File size: 0.92 MB

Usage

from datasets import load_dataset

dataset = load_dataset("k-mktr/aleppo_codex_he", split="train")
print(dataset[0])
# {'translation_id': 'aleppo_codex_he', 'book': 'Genesis', 'chapter': 1, 'verse': '1', 'text': 'בראשית ברא אלהים את השמים ואת הארץ'}

License

Public Domain.

Notes

• This is a consonantal text (no vowel points or cantillation marks).
• For a fully vocalized Hebrew Old Testament, see k-mktr/westminster_leningrad_codex_he or k-mktr/hebrew_masoretic_ot_hbo.
• Verse count differs slightly from the fully vocalized versions due to differing verse division conventions between the Ben Asher tradition and the Leningrad Codex.