Research map / Large language models / Language, safety and interpretability
Efficient Hallucination Detection: Adaptive Bayesian Estimation of Semantic Entropy with Guided Semantic Exploration
Qiyao Sun, Xingming Li, Xixiang He, Ao Cheng, Xuanyu Ji, Hailun Lu, Runke Huang, Qingyong Hu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use adaptive Bayesian semantic entropy with guided exploration instead of SESDLG | Llama-3.1-8B · TriviaQA (N=2) | AUROC: .741 -> .855 | Improved |
| use adaptive Bayesian semantic entropy instead of SESDLG at larger budget | Mistral-Small-24B · SimpleQA (N=5) | AUROC: .873 -> .871 | No clear effect |
| use adaptive Bayesian semantic entropy instead of SESDLG at low budget | Llama-2-7B · SimpleQA (N=2) | AUROC: .891 -> .895 | No clear effect |
| add weighted-perplexity adaptive prior (vs fixed K = Kobs + 1) (ablation) | Adaptive Bayesian SE · TriviaQA, Llama-3.1-8B (N=2) | AUROC: .812 -> .855 | Improved |
| add variance-based adaptive sampling with Bayesian framework (vs fixed sampling w/o Bayesian) (ablation) | Adaptive Bayesian SE · TriviaQA, Llama-3.1-8B (N=2) | AUROC: .759 -> .855 | Improved |
| add Bayesian framework to fixed sampling (ablation) | Fixed-budget sampling · TriviaQA, Llama-3.1-8B (N=2) | AUROC: .759 -> .798 | Improved |
| add guided semantic exploration with importance sampling (ablation) | Adaptive Bayesian SE · TriviaQA, Llama-3.1-8B (N=2) | AUROC: .823 -> .855 | Improved |
| add guided semantic exploration with importance sampling (ablation) | Adaptive Bayesian SE · TriviaQA, Llama-3.1-8B (N=5) | AUROC: .892 -> .913 | Improved |
| increase budget from N=2 to N=5 for relative gain over baselines | Adaptive Bayesian SE · Average over 12 settings | AUROC: 12.6% -> 6.3% avg improvement | Mixed |
Closest papers by meaning
- Entropy Distribution as a Fingerprint for Hallucinations in Generative Models (NeurIPS 2026 · similarity 0.93)
- Enhancing Uncertainty Modeling with Semantic Graph for Hallucination Detection (AAAI 2025 · similarity 0.92)
- Scalable Token-Level Hallucination Detection in Large Language Models (NeurIPS 2026 · similarity 0.92)
- Sample Transform Cost-Based Training-Free Hallucination Detector for Large Language Models (NeurIPS 2026 · similarity 0.92)
- HaDeMiF: Hallucination Detection and Mitigation in Large Language Models (ICLR 2025 · similarity 0.91)
- LLM-Check: Investigating Detection of Hallucinations in Large Language Models (NeurIPS 2024 · similarity 0.91)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
