GauntletScore offers auditable fact-checking matching top published checkers on LLM-AggreFact
A new paper introduces GauntletScore, an auditable fact-checking method for verifying whether AI-generated sentences are supported by source documents, matching the strongest published checkers on the LLM-AggreFact benchmark.
Read the full story at arXiv cs.CL ↗
Timeline · 1 report
- 2026-11-18 05:00 · arXiv cs.CL
GauntletScore: Auditable Fact-Checking That Matches the Best Published Checkers on LLM-AggreFact