Datasets:

jimx
/

simple-wiki-search

license: cc-by-sa-3.0
task_categories:
  - question-answering
  - text-retrieval
language:
  - en
tags:
  - wikipedia
  - agentic-search
  - retrieval
  - offline-index
pretty_name: SimpleWikiSearch
size_categories:
  - 10M<n<100M

SimpleWikiSearch Indexes

Offline Wikipedia indexes for SimpleWikiSearch (arXiv:2607.26070).

These files are binary search indexes (not row-oriented tables). Download and extract them locally; they are not intended for load_dataset().

Paper page: https://huggingface.co/papers/2607.26070

Files

Archive	Size (approx.)	Contents
`enwiki-20260601-tantivy-chunk-1536.tar.zst`	~22 GB	Tantivy keyword index + chunk text store
`enwiki-20260601-faiss-hnsw-Qwen3-Embedding-0.6B.tar.zst`	~19 GB	FAISS HNSW dense index (`Qwen3-Embedding-0.6B`, dim=1024, fp16)

Corpus snapshot: enwiki-20260601 — 7,189,602 articles / 10,837,506 section-aware chunks (target 1536 tokens).

Extract

tar -I zstd -xf enwiki-20260601-faiss-hnsw-Qwen3-Embedding-0.6B.tar.zst
tar -I zstd -xf enwiki-20260601-tantivy-chunk-1536.tar.zst

Expected layout:

data/database/wikipedia-index/enwiki-20260601/
  faiss/hnsw-Qwen3-Embedding-0.6B-chunk-1536-fp16/
    index.faiss
    ids.pkl
  tantivy/chunk-1536/
    chunks/
    titles/

Full setup (embedding server, MCP tools, QA agent): see deploy.md in the GitHub repo.

License

Wikipedia text is available under CC BY-SA 3.0. Indexes derived from that dump inherit the same license obligations.

Citation

@misc{xiong2026simplewikisearchcleanofflinewikipedia,
      title={SimpleWikiSearch: A Clean Offline Wikipedia Environment for Agentic Search},
      author={Guanming Xiong and Penghui Zhang},
      year={2026},
      eprint={2607.26070},
      archivePrefix={arXiv},
      primaryClass={cs.IR},
      url={https://arxiv.org/abs/2607.26070},
}