WikiContradict: A Benchmark for Evaluating LLMs on Real-World Knowledge Conflicts from Wikipedia

Documenti analoghi

• FactReasoner: A Probabilistic Approach to Long-Form Factuality Assessment for Large Language Models
  di: Marinescu, Radu, et al.
  Pubblicazione: (2025)
• FactCorrector: A Graph-Inspired Approach to Long-Form Factuality Correction of Large Language Models
  di: Carnerero-Cano, Javier, et al.
  Pubblicazione: (2026)
• When in Doubt, Cascade: Towards Building Efficient and Capable Guardrails
  di: Nagireddy, Manish, et al.
  Pubblicazione: (2024)
• Answering the Wrong Question: Reasoning Trace Inversion for Abstention in LLMs
  di: Gourabathina, Abinitha, et al.
  Pubblicazione: (2026)
• Value Alignment from Unstructured Text
  di: Padhi, Inkit, et al.
  Pubblicazione: (2024)