ReFACT benchmark reveals LLMs show a persistent salient distractor failure mode where 61% of incorrect error span predictions are semantically unrelated to actual errors, persisting across model sizes, and comparative judgment yields lower F1 than independent detection.
Fine-grained hallucination detection and editing for language models.arXivpreprint
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
representative citing papers
HalluScan benchmark tests hallucination detectors on LLMs, identifies NLI Verification as top performer with 0.88 AUROC, and introduces HalluScore (r=0.41 with humans) plus a routing method for 2x cost savings.
Researchers create a human-labeled dataset of obvious and elusive multimodal hallucinations and use learned activation-space probes to control their verifiability in MLLMs.
citing papers explorer
-
ReFACT: A Benchmark for Scientific Confabulation Detection with Positional Error Annotations
ReFACT benchmark reveals LLMs show a persistent salient distractor failure mode where 61% of incorrect error span predictions are semantically unrelated to actual errors, persisting across model sizes, and comparative judgment yields lower F1 than independent detection.
-
HalluScan: A Systematic Benchmark for Detecting and Mitigating Hallucinations in Instruction-Following LLMs
HalluScan benchmark tests hallucination detectors on LLMs, identifies NLI Verification as top performer with 0.88 AUROC, and introduces HalluScore (r=0.41 with humans) plus a routing method for 2x cost savings.
-
Steering the Verifiability of Multimodal AI Hallucinations
Researchers create a human-labeled dataset of obvious and elusive multimodal hallucinations and use learned activation-space probes to control their verifiability in MLLMs.