llm quality_error ai_generated true

RAG: retrieved context irrelevant to query

ID: llm/rag-retrieval-mismatch

Also available as: JSON · Markdown
75%Fix Rate
80%Confidence
3Evidence
2023-01-01First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
any active — — —

Root Cause

RAG pipeline retrieving irrelevant documents.

generic

Workarounds

  1. 90% success Use hybrid search: combine dense embeddings with BM25 sparse retrieval
  2. 88% success Implement reranking step after initial retrieval

Dead Ends

Common approaches that don't work:

  1. Increase top-k to retrieve more documents 72% fail

    More noise, longer context, slower

  2. Use keyword search only 75% fail

    Misses semantic similarity