AutoIndex learns retrieval indexing programs and lifts Recall@100 by 8.4%
Sam O'Nuallain · hf · 2026-07-23
AutoIndex learns retrieval preprocessing as executable programs
AutoIndex treats document representation as an optimization target instead of a fixed preprocessing step. It searches over representation programs—executable transforms that can slice, enrich, normalize, reweight, or reorganize documents before they are indexed.
- The system uses validation-guided program search: agents inspect failures of the current index pipeline and synthesize candidate updates.
- Only updates that improve retrieval quality are kept.
- Evaluation on CRUMB, a benchmark of heterogeneous retrieval tasks, keeps BM25 fixed across experiments.
Results:
- +8.4% average Recall@100 over static full-document BM25
- +8.3% average nDCG@10
- Best task gains of +30.5% Recall@100 and +43.6% nDCG@10
The authors argue that document representation should be explicitly optimized, not hard-coded before retrieval starts. Code is available on GitHub.
More from coding & agent
- Claude + Chromium + Playwright can now record demo videos automatically — yenkel · 2026-07-23
- Claude plus Chromium and Playwright MCP can now record demo videos for you — yenkel · 2026-07-23
- Frontend work with Claude Design and Claude Code is “actually so good” — trq212 · 2026-07-23
- Multiple LLMs on one project start treating collaboration structure as a core feature — repligate · 2026-07-23
- Blocks says its agent-friendly UI system is nearing real-time on-brand generation — round · 2026-07-23
- Codex needing iTunes access becomes the latest AI agent meme — KarelDoostrlnck · 2026-07-23