Showing 2026Show all
2 papers · 1 filter
cs.CL2026
Human-Anchored Factuality Evaluation with Strategic Annotation
Yu Wang, Craig Erickson, Kevin Small
LLM-based factuality judges provide scalable evaluation signals, but their metrics are often systematically biased relative to human judgments. We study human-anchored factuality e…
cs.CL2026
Evaluating the Utility of Grounding Documents with Reference-Free LLM-based Metrics
Yilun Hua, Giuseppe Castellucci, Peter Schulam +2
Retrieval Augmented Generation (RAG)'s success depends on the utility the LLM derives from the content used for grounding. Quantifying content utility does not have a definitive sp…