about
MathMixup: Boosting LLM Mathematical Reasoning via Curriculum and Decomposition (arxiv.org)
1 point by wslh 21 days ago | hide | past | pdf | discuss on HN

In plain words: It builds math practice problems by mixing and breaking apart existing ones at a chosen difficulty, checks each for clarity, then trains the AI from easy to hard. A mid-sized AI model trained this way averaged 52.6% across seven math tests, beating the previous best.

Abstract · MathMixup: Boosting LLM Mathematical Reasoning with Difficulty-Controllable Data Synthesis and Curriculum Learning

In mathematical reasoning tasks, the advancement of Large Language Models (LLMs) relies heavily on high-quality training data with clearly defined and well-graded difficulty levels. However, existing data synthesis methods often suffer from limited diversity and lack precise control over problem difficulty, making them insufficient for supporting efficient training paradigms such as curriculum learning. To address these challenges, we propose MathMixup, a novel data synthesis paradigm that systematically generates high-quality, difficulty-controllable mathematical reasoning problems through hybrid and decomposed strategies. Automated self-checking and manual screening are incorporated to ensure semantic clarity and a well-structured difficulty gradient in the synthesized data. Building on this, we construct the MathMixupQA dataset and design a curriculum learning strategy that leverages these graded problems, supporting flexible integration with other datasets. Experimental results show that MathMixup and its curriculum learning strategy significantly enhance the mathematical reasoning performance of LLMs. Fine-tuned Qwen2.5-7B achieves an average score of 52.6\% across seven mathematical benchmarks, surpassing previous state-of-the-art methods. These results fully validate the effectiveness and broad applicability of MathMixup in improving the mathematical reasoning abilities of LLMs and advancing data-centric curriculum learning.

Xuchen Li, Jing Chen, Xuzhao Li, Hao Liang, Xiaohuan Zhou, Taifeng Wang, Wentao Zhang
arXiv:2601.17006 · cs.LG, cs.AI · submitted Jan 14, 2026
abstract · pdf · html · Preprint, Under review

add comment on HN