all-evidence
A judgment rule in RAG retrieval evaluation that requires the Top-K results to contain all required evidence for a query to count as a hit. It accurately reflects complete recall capability in multi-document query scenarios.
Timeline (last 90 days)
In deployment decisions, all-evidence should serve as the primary reference metric, because it is closer to the user's actual experience; any-hit can be retained as a diagnostic tool.
If the any-hit score is high while the all-evidence score is low, the problem lies in coverage, and one may consider increasing Top-K, improving the chunk strategy, or introducing multi-hop retrieval; if both are low, it indicates a problem with the retriever's basic capability.
All Facts (2)
In deployment decisions, all-evidence should serve as the primary reference metric, because it is closer to the user's actual experience; any-hit can be retained as a diagnostic tool.
50%UnverifiedIf the any-hit score is high while the all-evidence score is low, the problem lies in coverage, and one may consider increasing Top-K, improving the chunk strategy, or introducing multi-hop retrieval; if both are low, it indicates a problem with the retriever's basic capability.
50%