Nemotron-Math: Efficient Long-Context Distillation of Mathematical Reasoning from Multi-Mode Supervision

TL;DR

Nemotron-Math is a large-scale dataset for mathematical reasoning that combines diverse problem sources and reasoning modes, enabling improved model training and state-of-the-art performance on benchmarks.

Contribution

The paper introduces Nemotron-Math, a comprehensive dataset with multi-mode reasoning traces and an efficient training strategy for long-context mathematical reasoning models.

Findings

Outperforms previous datasets on AoPS problems.

Enhances robustness and generalization with StackExchange data.

Achieves 100% maj@16 accuracy on AIME 2024 and 2025.

Abstract

High-quality mathematical reasoning supervision requires diverse reasoning styles, long-form traces, and effective tool integration, capabilities that existing datasets provide only in limited form. Leveraging the multi-mode generation ability of gpt-oss-120b, we introduce Nemotron-Math, a large-scale mathematical reasoning dataset containing 7.5M solution traces across high, medium, and low reasoning modes, each available both with and without Python tool-integrated reasoning (TIR). The dataset integrates 85K curated AoPS problems with 262K community-sourced StackExchange-Math problems, combining structured competition tasks with diverse real-world mathematical queries. We conduct controlled evaluations to assess the dataset quality. Nemotron-Math consistently outperforms the original OpenMathReasoning on matched AoPS problems. Incorporating StackExchange-Math substantially improves robustness and generalization, especially on HLE-Math, while preserving accuracy on math competition benchmarks. To support efficient long-context training, we develop a sequential bucketed strategy that accelerates 128K context-length fine-tuning by 2--3$\times$ without significant accuracy loss. Overall, Nemotron-Math enables state-of-the-art performance, including 100\% maj@16 accuracy on AIME 2024 and 2025 with Python TIR.