AI Paper Atlas

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

On Code-Induced Reasoning in LLMs

Abdul Waheed, Zhen Wu, Carolyn Rose, Daphne Ippolito

ICLR 2026 · Large language models › Agents, code and math · approach: Code generation benchmarks · 4 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
finetune on code or mixed code+NL data instead of NL-only or zero-shotLLM SFT (14 bases) · natural language tasksaccuracy: 64%Improved
finetune on code or mixed dataLLM SFT (14 bases) · math and code understanding tasksaccuracy: 86%Improved
finetune on code or mixed dataLLM SFT (14 bases) · code generationjudge scoreImproved
raise the fraction of code in mixed finetuningQwen3 SFT · math and other tasksaccuracyImproved
apply structural perturbations (whitespace removal, pseudocode, flowcharts) to code data (ablation)code-ft · math and code tasksaccuracyHurt
apply semantic perturbations (identifiers, keywords, comments) to code data (ablation)code-ft · math and code tasksaccuracyHurt
replace code with pseudocode or flowchart abstractions (ablation)code-ft · non-code-generation tasksaccuracyNo clear effect
replace code with natural language step-by-step procedures (ablation)code-ft · reasoning tasksaccuracyHurt
reduce token density of code data (compact encodings) (ablation)code-ft · reasoning tasks excluding code generationaccuracyNo clear effect
reduce token density of code data (ablation)code-ft · code generationjudge scoreHurt
use low-interpretability corrupted code with misleading signals (ablation)code-ft · reasoning tasksaccuracyNo clear effect
train on lower-level languages (Java, Rust) instead of high-scripting languages (ablation)single-language SFT · math tasksaccuracyImproved

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