Definition
Knowledge conflict occurs in RAG pipelines when the non-parametric data retrieved from an external data source contradicts the parametric knowledge embedded in the LLM's weights during pre-training. This discrepancy forces the model to choose between its internal 'world model' and the provided external context, often leading to performance degradation if not handled via prompt engineering or architectural weighting.
Differentiates between a model being 'wrong' (hallucination) and a model being 'conflicted' (internal training vs. external context).
"A judge weighing a new witness statement (Retrieved Context) against their own personal memory of the event (Parametric Weights)."
- Parametric Knowledge(Prerequisite)
- Non-parametric Knowledge(Component)
- Contextual Grounding(Mitigation Strategy)
- Hallucination(Potential Outcome)
Conceptual Overview
Knowledge conflict occurs in RAG pipelines when the non-parametric data retrieved from an external data source contradicts the parametric knowledge embedded in the LLM's weights during pre-training. This discrepancy forces the model to choose between its internal 'world model' and the provided external context, often leading to performance degradation if not handled via prompt engineering or architectural weighting.
Disambiguation
Differentiates between a model being 'wrong' (hallucination) and a model being 'conflicted' (internal training vs. external context).
Visual Analog
A judge weighing a new witness statement (Retrieved Context) against their own personal memory of the event (Parametric Weights).