Get Started
Home
Topics
Search
Library
Research questionHow can hallucination detectors verify claims and localize unsupported spans without costly separate pipelines?A generated response may contain unsupported facts that need both whole-claim verification and precise localization within the text. Handling these granularities separately can require repeated verification and additional alignment between claims and spans.
AI
Alignment & Safety
Evaluation & Benchmarks
Inference Optimization
Natural Language Processing
Latest papersRecent research connected to this question, newest first.Enoki: Efficient Multi-Level Hallucination DetectionApplies to LLM hallucination detection using evidence, including LLM-based, encoder-based, and rule-based extraction regimes. The source reports experiments on claim verification and span/entity localization and introduces a dual-granularity annotated dataset; it does not establish performance beyond those reported settings.research paper · Sep 4, 2026
Related questions
How can we reliably detect when an LLM response is unsupported by its reference documents?How can black-box LLM APIs detect hallucinations without trusted context while limiting false alarms?How can we detect object hallucinations in vision-language models when visual grounding shifts across layers?How can large vision-language models reduce visual hallucinations without extra training or decoding passes?