AI Paper Atlas

Research map / Large language models / Agents, code and math

What Really Improves Mathematical Reasoning: Structured Reasoning Signals Beyond Pure Code

Yuze Zhao, Junpeng Fang, Lu Yu, Zhenya Huang, Kai Zhang, Qing Cui, Qi Liu, JUN ZHOU, Enhong Chen

ICML 2026 · Large language models › Agents, code and math · approach: Mathematical and structural reasoning benchmarks · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
include standalone pure code in pretraining mix (fixed token budget) (ablation)10T-token MoE · mathematical benchmarks averagerelative score: -14.38%Hurt
include pure code corpus (ablation)10T-token MoE · MATH and OlympiadBenchrelative score: MATH -22.64%; OlympiadBench -47.16%Hurt
include pure code corpus (ablation)10T-token MoE · GSM8Krelative score: +0.4%No clear effect
include pure code corpus (ablation)10T-token MoE · PIQA and HellaSwagrelative score: -2.11%, -2.39%Hurt
include pure code corpus (ablation)10T-token MoE · programming benchmarksscoreImproved
include math corpus (ablation)10T-token MoE · CodeForces and LiveCodeBenchrelative score: CodeForces +37.11%; LiveCodeBench +11.26%Improved
include math corpus (ablation)10T-token MoE · CruxEval, MBPP, HellaSwagrelative score: CruxEval -17.30%; MBPP -6.12%; HellaSwag -2.94%Hurt
replace part of math data with structured cognitive-scaffold samples at fixed math budget10T-token MoE · mathematical reasoning averagerelative score: +17.56%Improved
add cognitive scaffolds10T-token MoE · OlympiadBench, MATH, College Math, MathBenchrelative score: +47.78%, +23.17%, +30.05%, +14.51%Improved
add cognitive scaffolds10T-token MoE · GSM8K and CMathrelative score: -6.29%, -2.00%Hurt
add cognitive scaffolds10T-token MoE · programming benchmarksscore: about -1%No clear effect
remove cognitive scaffolds (w/o cog) vs removing code or math (ablation)MoE routing · expert-routing JS divergencerouting shiftNo 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