Q2D-Web: new retrieval benchmark with 190M web docs and 70k LLM-rewritten queries
antoine_chaffin · x · 2026-09-10
Researchers (antoinechaffin of LightOn) released Q2D-Web, a large-scale retrieval benchmark targeting web search — a use case existing benchmarks evaluate poorly.
- 190M web documents, with a subsampled split preserving ranking fidelity.
- 70k ecologically valid queries rewritten by LLMs (now among the largest sources of search queries), with deep RAG- and retrieval-oriented relevance annotations.
- Comes with a hosted leaderboard.
Related event: Perplexity Releases Q2D-Web, First Open Benchmark for Agentic RAG Retrieval(8 posts)→
More from Research
- Graph Machine: edge-based pretraining architecture that swaps dense Transformer layers — Lintai Hou · 2026-09-10
- ~300 neurons, no training: an artificial brain "primitive" for robot intelligence — chris_j_paxton · 2026-09-10
- MIT Press classic 'Visual Cortex and Deep Networks' (Poggio & Anselmi, 2016) now free — Limor_Raviv · 2026-09-10
- First systematic study of hybrid linear attention: PAS and ISP activation spikes explained — jiqizhixin · 2026-09-10
- Live from ICM 2026: Venkatesh, Vakil and Kontorovich on math in the age of AI — stevenstrogatz · 2026-09-10
- Hallucination rate as a kill gate: replay pipeline reveals pitfalls of LLM model swaps — dl_weekly · 2026-09-10