illuin-tech/vidore-benchmark

Vision Document Retrieval (ViDoRe): Benchmark. Evaluation code for the ColPali paper.

71
/ 100
Verified

Evaluates end-to-end document retrieval pipelines (multi-stage retrieval, hybrid approaches, custom preprocessing) on ViDoRe v3 datasets through a modular `BasePipeline` interface. Integrates with MTEB for single-model retriever evaluation and tracks both retrieval metrics and computational costs (indexing/search time). Supports language filtering, custom pipeline implementations via Python or CLI, and community-driven result submission for leaderboard comparison.

265 stars and 3,130 monthly downloads. Available on PyPI.

Maintenance 10 / 25
Adoption 18 / 25
Maturity 25 / 25
Community 18 / 25

How are scores calculated?

Stars

265

Forks

35

Language

Python

License

MIT

Last pushed

Feb 27, 2026

Monthly downloads

3,130

Commits (30d)

0

Dependencies

10

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/rag/illuin-tech/vidore-benchmark"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.