Q2D-Web leaderboard

Public leaderboard for Q2D-Web, a large-scale benchmark for first-stage retrieval in agentic RAG: ~190M web documents, ~70k agent-reformulated production queries in ten languages, three independent relevance-judgment sets.

This repo is the Space only, and self-contained: it renders the result JSONs committed under data/eval-results/. Scores are computed by the scoring pipeline (in the srch-search-api monorepo), which publishes the JSONs into this repo — there is no separate results dataset.

pip install -r requirements.txt
python app.py      # http://127.0.0.1:7860