RAG-Stress: Probing the Limits of Evidence Reliance in Retrieval-Augmented Generation
arXiv:2610.11183v1 Announce Type: new Abstract: Following retrieved evidence does not guarantee factual correctness: misleading evidence can induce a model to replace an answer it previously gave correctly. Standard accuracy measures obscure this behavior by combining answer replacement with preexi…
Read the full story at arXiv cs.CL ↗
Timeline · 1 report
- 2026-10-09 04:00 · arXiv cs.CL
RAG-Stress: Probing the Limits of Evidence Reliance in Retrieval-Augmented Generation