AI Paper Atlas

Research map / Large language models / Reasoning and chain of thought

Mind the Gap: Catching Hallucinations via Evidence Drop on the Reasoning Manifold

Qunjie Chen, Yufei Chen, Xiaodong Yue, Linye Li

ICML 2026 · Large language models › Reasoning and chain of thought · approach: Efficient and adaptive thinking

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes a step-wise, dynamics-based hallucination detector for LLM reasoning that tracks evidence change along the reasoning trajectory rather than averaging sequence-level uncertainty.

Problem

Hallucinations in LLM reasoning, where fluent chains of thought become factually or logically wrong; existing uncertainty-based detectors use sequence-level averaging that ignores step-wise reasoning dynamics and misclassifies hard-but-correct or easy-but-wrong samples.

Approach

Models reasoning as a trajectory on a reasoning manifold and detects hallucinations via the "evidence drop" across steps, replacing sequence-level averaging with a dynamic, step-wise signal. Full method details are not available from the abstract.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace sequence-average with worst-case Evidence Drop aggregation (ablation)Shannon entropy, Qwen3-8B · MATH (alpha=0.05)selective accuracy (%): 70.24 -> 88.26Improved
replace average with Evidence Drop (ablation)Shannon entropy, Qwen3-8B · MATHAURC (x1000): 288.8 -> 190.0Improved
replace average with Evidence Drop (ablation)LogTokU, Qwen3-4B · MATHAURC (x1000): 481.3 -> 299.1Improved
replace average with Evidence Drop (ablation)LogTokU, Qwen3-4B · GSM8K (alpha=0.05)selective accuracy (%): 0 -> 85.57Improved
replace average with Evidence Drop (ablation)LN-S, Qwen3-4B · GSM8K (alpha=0.05)selective accuracy (%): 95.45 -> 83.33Hurt
replace average with Evidence Drop (ablation)LN-S, Qwen3-8B · GSM8K (alpha=0.05)selective accuracy (%): 98.68 -> 94.44Hurt
replace average with Evidence Drop (ablation)LN-S, Qwen3-8B · MATH (alpha=0.05)selective accuracy (%): 69.79 -> 86.94Improved
replace average with Evidence Drop (ablation)Shannon entropy, Qwen3-8B · GSM8K (alpha=0.50)selective accuracy (%): 97.06 -> 93.72Hurt
use Shannon entropy instead of LN-S or LogTokU as the evidence proxy (ablation)Evidence Drop · GSM8K Qwen3-4B (alpha=0.05)selective accuracy (%): 83.33 -> 100Improved
replace average with Evidence Drop (ablation)Shannon entropy, Qwen3-8B · GSM8K step localizationstep-level localization accuracy (%): 27.66 -> 46.11Improved
replace average with Evidence Drop (ablation)Shannon entropy, Qwen3-8B · MATH step localizationstep-level localization accuracy (%): 24.62 -> 32.90Improved
replace average with Evidence Drop (ablation)Shannon entropy, Qwen3.5-27B · MATH (alpha=0.05)selective accuracy (%): 62.50 -> 92.92Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy