AI Paper Atlas

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

Characterizing and Mitigating Reasoning Drift in Large Language Models

Yufeng Zhang, Xuepeng Wang, Lingxiang Wu, Jinqiao Wang

ICLR 2026 · Large language models › Reasoning and chain of thought · approach: Search and self-correction in reasoning

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.

Analyzes thousands of chain-of-thought reasoning paths to characterize "Reasoning Drift," a failure mode where LLMs lock into flawed reasoning patterns, and proposes mitigation methods.

Problem

Chain-of-thought reasoning in LLMs is unreliable because stochastic generation can lock models into flawed reasoning patterns (reasoning drift).

Approach

Empirically characterizes reasoning drift across thousands of sampled reasoning paths, showing its manifestation is an interplay of factors; mitigation details are not stated in the available abstract.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
apply reasoning-aware activation steeringR1-Distill-Llama-8B · AIME2024accuracy: 34.99 -> 55.56Improved
apply reasoning-aware activation steeringR1-Distill-Qwen-14B · AIME2025accuracy: 26.67 -> 36.67Improved
apply fine-grained library of steering vectors instead of the single uncertainty-management vector (CAAum) (ablation)R1-Distill-Qwen-14B · GPQA-Diamondaccuracy: 52.02 -> 57.57Improved
apply single uncertainty-management vector (CAAum) alone (ablation)R1-Distill-Qwen-14B · AIME2024 and GPQA-Diamondaccuracy: 54.44 -> 53.33 (AIME2024); 55.55 -> 52.02 (GPQA-D)Hurt
match steering vector to source and target category (vs random source or random target) (ablation)RAAS · GSM8K, AIME, GPQA-DiamondaccuracyImproved
apply fully random steering vectors (ablation)RAAS · GSM8K, AIME, GPQA-DiamondaccuracyHurt
transfer vectors from distilled models to non-distilled base modelsLlama3.1-8B · GSM8Kaccuracy: 83.91 -> 85.22Improved
transfer vectors from distilled models to non-distilled base modelsQwen2.5-14B · GPQA-Diamondaccuracy: 34.27 -> 35.18Improved
apply math-derived steering vectors to non-math tasksR1-Distill-Qwen-14B · ReCloraccuracy: 65.97 -> 66.51Improved
apply math-derived steering vectors to non-math tasksR1-Distill-Llama-8B · LogiQAaccuracy: 58.80 -> 61.00Improved
apply RAAS vs PiCSAR on AIME2025R1-Distill-Llama-8B · AIME2025accuracy: 31.70 vs 33.33No clear effect

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