Skip to content

Retrieval, embeddings and long context

BRIGHT

Retrieval requiring reasoning beyond lexical or simple semantic matching.

About BRIGHT

What it measures

nDCG@10 and domain breakdowns.

Scope & limitations

Good complement to BEIR; label query expansion, reasoning model and additional compute.

Evaluation environment

Retrieval/reranker or agentic retrieval system.

Source

Original benchmark

Dataset & task catalog ↗