conflict-QA datasets

The 'conflict-QA datasets' consist of query-context instances used to evaluate the ability of large language models to navigate knowledge conflicts between retrieved contextual knowledge and parametric knowledge.

Papers using conflict-QA datasets (3)

Trust or Abstain? A Self-Aware RAG Approach2026Exploring Knowledge Conflicts for Faithful LLM Reasoning: Benchmark and Method2026When Abundance Conceals Weakness: Knowledge Conflict in Multilingual Models2026