AI Paper Atlas

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

Intervene When It Doubts: Conjunction-Guided Interactive Reasoning

Qianyue Wang, Jinwu Hu, Yaofo Chen, Yufeng Wang, Bailin Chen, Huanxiang Lin, Yu Rong, Yuanqing Li, Zhiquan Wen, Mingkui Tan

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 conjunction-guided interactive reasoning to curb overthinking and overshoot in large reasoning models by intervening when the model shows doubt (self-validation and exploratory extension).

Problem

Large reasoning models suffer from inefficient reasoning (overthinking and overshoot) driven by excessive self-validation and exploratory extension triggered by model doubt, raising compute cost and sometimes hurting accuracy.

Approach

Intervene during reasoning only when the model doubts itself, using a conjunction of signals to decide when to intervene, rather than regulating reasoning through purely internal signals as prior efficient-reasoning methods do.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add conjunction-guided external feedback (CGI) to base LRMDeepSeek-R1-Distilled-Qwen2.5-7B · AIME24@32accuracy (%): 40.00 -> 73.85Improved
add CGI to base LRMDeepSeek-R1-Distilled-Qwen2.5-7B · AIME24@32avg reasoning length: 3374.50 -> 1571.70Improved
add feedback tags (ablation)CGI · MATH500accuracy (%): 86.40 -> 94.00Improved
add fine-grained action-oriented feedback instead of binary static feedback (ablation)CGI · GPQA Diamondaccuracy (%): 43.88 -> 65.22Improved
add GRPO training instead of unadapted base model (ablation)CGI · AIME24accuracy (%): 33.75 -> 73.85Improved
add GRPO instead of SFT on synthesized traces (ablation)CGI · AIME24accuracy (%): 63.33 -> 73.85Improved
intervene at transitional conjunctions instead of newline (ablation)CGI · AIME24accuracy (%): 56.35 -> 73.85Improved
intervene at transitional conjunctions instead of every sentence (ablation)CGI · LiveCodeBenchaccuracy (%): 15.43 -> 34.86Improved
intervene at fixed length instead of conjunctions (ablation)CGI · AIME24accuracy (%): 38.33 -> 73.85Improved
scale feedback LLM proxy from 7B to 72B (ablation)CGI · AIME24accuracy (%): 59.68 -> 73.85Improved
use even a 7B peer-sized proxy for feedback (ablation)CGI · GPQA Diamondaccuracy (%): 22.25 -> 44.95Improved
use human feedback instead of LLM proxyCGI · AIME25@3accuracy (%): 38.89 -> 61.11Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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