AI Paper Atlas

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

ICPO: Provable and Practical In-Context Policy Optimization for Test-Time Scaling

Tianrun Yu, yuxiao Yang, Zhaoyang Wang, Kaixiang Zhao, Porter Jenkins, Xuchao Zhang, Chetan Bansal, Huaxiu Yao, Weitong Zhang

ICLR 2026 · Large language models › Reasoning and chain of thought · approach: Test-time compute scaling

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 In-Context Policy Optimization (ICPO), where an LLM agent improves answers over multi-round self-reflection at inference using self-assessed or external rewards, without parameter updates, with a theoretical account of the mechanism.

Problem

Test-time scaling via multi-round self-reflection: how a model can improve its answers at inference through in-context feedback without modifying its parameters.

Approach

Treats multi-round in-context self-reflection as an in-context policy optimization process driven by self-assessed or externally observed rewards, and explains it theoretically via pretraining under a Fisher-weighted log-likelihood objective.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
apply ME-ICPO multi-round in-context self-improvementQwen2.5-Math-7B · AIME 2024Mean@16 (%): 11.04 -> 30.42Improved
apply ME-ICPOQwen2.5-Math-7B · AMCMean@16 (%): 41.42 -> 47.06Improved
apply ME-ICPOQwen2.5-Math-1.5B · AIME 2024Mean@16 (%): 6.46 -> 9.79Improved
apply ME-ICPOQwen2.5-Math-1.5B · MATH-L1Mean@16 (%): 49.27 -> 61.19Improved
apply ME-ICPOQwen2.5-Math-7B · MATH-L5Mean@16 (%): 30.58 -> 38.71Improved
add minimum-entropy response selection (ablation)ME-ICPO (Qwen2.5-Math-7B) · AIME 2024Accuracy (%): 5.77 -> 30.05Improved
add explicit reward tags in context (ablation)ME-ICPO (Qwen2.5-Math-7B) · AIME 2024Accuracy (%): 19.30 -> 30.05Improved
add reward without entropy selection (ablation)ME-ICPO (Qwen2.5-Math-7B) · AIME 2024Accuracy (%): 5.77 -> 6.21No clear effect
use ground-truth reward instead of self-assessed majority vote (ablation)ME-ICPO (Qwen2.5-Math-7B) · AIME 2024Accuracy (%): 30.05 -> 38.19Improved
apply majority voting on top of ME-ICPO outputsME-ICPO · AIME 2024Maj@kImproved

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