claim
Datadog's hallucination detection system categorizes contradictions as claims made in an LLM-generated response that directly oppose the provided context, which is assumed to be correct.

Authors

Sources

Referenced by nodes (2)