Synthesis by Design: Controlled Data Generation via Structural Guidance

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xu, Lei, Chen, Sirui, Huang, Yuxuan, Lu, Chaochao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910999845535744
author Xu, Lei
Chen, Sirui
Huang, Yuxuan
Lu, Chaochao
author_facet Xu, Lei
Chen, Sirui
Huang, Yuxuan
Lu, Chaochao
contents Mathematical reasoning remains challenging for LLMs due to complex logic and the need for precise computation. Existing methods enhance LLM reasoning by synthesizing datasets through problem rephrasing, but face issues with generation quality and problem complexity. To address this, we propose to extract structural information with generated problem-solving code from mathematical reasoning and guide data generation with structured solutions. Applied to MATH and GSM8K, our approach produces 39K problems with labeled intermediate steps and a 6.1K-problem benchmark of higher difficulty. Results on our benchmark show that model performance declines as reasoning length increases. Additionally, we conducted fine-tuning experiments using the proposed training data on a range of LLMs, and the results validate the effectiveness of our dataset. We hope the proposed method and dataset will contribute to future research in enhancing LLM reasoning capabilities. Our code and data are available at https://github.com/OpenCausaLab/StructuralGeneration.
format Preprint
id arxiv_https___arxiv_org_abs_2506_07664
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Synthesis by Design: Controlled Data Generation via Structural Guidance
Xu, Lei
Chen, Sirui
Huang, Yuxuan
Lu, Chaochao
Computation and Language
Artificial Intelligence
Mathematical reasoning remains challenging for LLMs due to complex logic and the need for precise computation. Existing methods enhance LLM reasoning by synthesizing datasets through problem rephrasing, but face issues with generation quality and problem complexity. To address this, we propose to extract structural information with generated problem-solving code from mathematical reasoning and guide data generation with structured solutions. Applied to MATH and GSM8K, our approach produces 39K problems with labeled intermediate steps and a 6.1K-problem benchmark of higher difficulty. Results on our benchmark show that model performance declines as reasoning length increases. Additionally, we conducted fine-tuning experiments using the proposed training data on a range of LLMs, and the results validate the effectiveness of our dataset. We hope the proposed method and dataset will contribute to future research in enhancing LLM reasoning capabilities. Our code and data are available at https://github.com/OpenCausaLab/StructuralGeneration.
title Synthesis by Design: Controlled Data Generation via Structural Guidance
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2506.07664