BibleLM Dataset

A high-performance, stateless Bible dataset optimized for edge-first RAG (Retrieval-Augmented Generation).

This dataset contains the processed Bible text, morphological data, and search indices used by the BibleLM project.

📚 What's inside?

• Combined Bible Index: Cleaned and tokenized text for BSB (Berean Standard Bible), KJV, WEB, and ASV.
• Search Engine State: Pre-computed BM25 term frequencies (bm25-state.json) allowing for <10ms search engine hydration on serverless platforms.
• Original Languages: Extensive morphological data for Hebrew (WLC/OSHB) and Greek (SBLGNT/OpenGNT), including word-by-word Strong's mappings and glosses.
• TSK Cross-References: Over 500,000 ranked thematic connections from the Treasury of Scripture Knowledge.

🛠 Technical Usage

The data is structured as flat JSON fragments to allow for efficient loading in Edge Functions (Vercel/Cloudflare).

// Example: Loading the BM25 state in a serverless function
import bm25State from './data/bm25-state.json';
import bibleIndex from './data/bible-full-index.json';
const engine = BM25Engine.createFromState(bm25State, bibleIndex);

⚖️ License & Attributions

This dataset is distributed under CC BY-NC 4.0 (Attribution-NonCommercial) because it incorporates the following open-source works:

1. OpenHebrewBible: CC BY-NC 4.0 (Source: eliranwong/OpenHebrewBible).
2. OpenGNT: CC BY-NC 4.0.
3. MorphHB: CC BY 4.0 (Source: OpenScriptures).
4. Berean Standard Bible (BSB): Public Domain (berean.bible).
5. KJV/WEB/ASV: Public Domain.
6. Strong's Concordance: Public Domain.
7. TSK: Public Domain.

Citation

If you use this dataset in your research or applications, please attribute the original sources listed above and the BibleLM project.