JT-Math: A Multi-Stage Framework for Advanced Mathematical Reasoning in Large Language Models

• URL: http://arxiv.org/abs/2507.19748v1
• Date: Sat, 26 Jul 2025 02:45:10 GMT
• Title: JT-Math: A Multi-Stage Framework for Advanced Mathematical Reasoning in Large Language Models
• Authors: Yifan Hao, Fangning Chao, Yaqian Hao, Zhaojun Cui, Huan Bai, Haiyu Zhang, Yankai Liu, Chao Deng, Junlan Feng,
• Abstract summary: We introduce JT-Math-8B, a series of open-source models comprising base, instruct, and thinking versions.<n>The Instruct Model is optimized for direct, concise answers through Supervised Fine-Tuning (SFT) and a GRPO-based reinforcement learning (RL) method.<n>The Thinking Model is trained for complex problem-solving using a Long Chain-of-Thought (Long CoT) approach, combining SFT with a novel, multi-stage RL curriculum.
• Score: 15.054016881151254
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Mathematical reasoning is a cornerstone of artificial general intelligence and a primary benchmark for evaluating the capabilities of Large Language Models (LLMs). While state-of-the-art models show promise, they often falter when faced with complex problems that demand deep conceptual understanding and intricate, multi-step deliberation. To address this challenge, we introduce JT-Math-8B, a series of open-source models comprising base, instruct, and thinking versions, built upon a systematic, multi-stage optimization framework. Our pre-training corpus is a high-quality, 210B-token dataset curated through a dedicated data pipeline that uses model-based validation to ensure quality and diversity. The Instruct Model is optimized for direct, concise answers through Supervised Fine-Tuning (SFT) and a GRPO-based reinforcement learning (RL) method. The Thinking Model is trained for complex problem-solving using a Long Chain-of-Thought (Long CoT) approach, combining SFT with a novel, multi-stage RL curriculum that progressively increases task difficulty and context length up to 32K tokens. JT-Math-8B achieves state-of-the-art results among open-source models of similar size, surpassing prominent models like OpenAI's O1-mini and GPT-4o , and demonstrating superior performance on competition-level mathematics.

Related papers

• Unifying Multimodal Large Language Model Capabilities and Modalities via Model Merging [103.98582374569789]
  Model merging aims to combine multiple expert models into a single model, thereby reducing storage and serving costs.<n>Previous studies have primarily focused on merging visual classification models or Large Language Models (LLMs) for code and math tasks.<n>We introduce the model merging benchmark for MLLMs, which includes multiple tasks such as VQA, Geometry, Chart, OCR, and Grounding, providing both LoRA and full fine-tuning models.
  arXiv  Detail & Related papers  (2025-05-26T12:23:14Z)
• MergeBench: A Benchmark for Merging Domain-Specialized LLMs [19.49737955489798]
  MergeBench is an evaluation suite designed to assess model merging at scale.<n>It builds on state-of-the-art open-source language models, including Llama and Gemma families at 2B to 9B scales.<n>We assess eight representative merging methods across multi-task performance, forgetting and runtime efficiency.
  arXiv  Detail & Related papers  (2025-05-16T04:02:55Z)
• MathFimer: Enhancing Mathematical Reasoning by Expanding Reasoning Steps through Fill-in-the-Middle Task [49.355810887265925]
  We introduce MathFimer, a novel framework for mathematical reasoning step expansion.<n>We develop a specialized model, MathFimer-7B, on our carefully curated NuminaMath-FIM dataset.<n>We then apply these models to enhance existing mathematical reasoning datasets by inserting detailed intermediate steps into their solution chains.
  arXiv  Detail & Related papers  (2025-02-17T11:22:24Z)
• Large Language Models Meet Symbolic Provers for Logical Reasoning Evaluation [24.081573908824353]
  First-order logic (FOL) reasoning is pivotal for intelligent systems.<n>Existing benchmarks often rely on extensive human annotation or handcrafted templates.<n>We propose a novel framework called ProverGen that synergizes the generative strengths of Large Language Models with the rigor and precision of symbolic provers.
  arXiv  Detail & Related papers  (2025-02-10T15:31:54Z)
• URSA: Understanding and Verifying Chain-of-thought Reasoning in Multimodal Mathematics [23.80647785460245]
  Process Reward Models (PRMs) have shown promise in enhancing the mathematical reasoning capabilities of Large Language Models.<n>We take the first step toward unlocking the potential of PRMs in multimodal mathematical reasoning.<n>We introduce URSA, a three-stage Unfolding multimodal Process-Supervision Aided training framework.
  arXiv  Detail & Related papers  (2025-01-08T18:49:41Z)
• RADIOv2.5: Improved Baselines for Agglomerative Vision Foundation Models [60.596005921295806]
  Agglomerative models have emerged as a powerful approach to training vision foundation models.<n>We identify critical challenges including resolution mode shifts, teacher imbalance, idiosyncratic teacher artifacts, and an excessive number of output tokens.<n>We propose several novel solutions: multi-resolution training, mosaic augmentation, and improved balancing of teacher loss functions.
  arXiv  Detail & Related papers  (2024-12-10T17:06:41Z)
• LATTE: Learning to Think with Vision Specialists [103.5952731807559]
  We propose LATTE, a family of vision-language models that offload perception to state-of-the-art vision models.<n>By offloading perception to state-of-the-art vision models, our approach enables vision-language models to focus solely on reasoning over high-quality perceptual information.
  arXiv  Detail & Related papers  (2024-12-07T00:42:04Z)
• Enhancing Multi-Step Reasoning Abilities of Language Models through Direct Q-Function Optimization [49.362750475706235]
  Reinforcement Learning (RL) plays a crucial role in aligning large language models with human preferences and improving their ability to perform complex tasks.<n>We introduce Direct Q-function Optimization (DQO), which formulates the response generation process as a Markov Decision Process (MDP) and utilizes the soft actor-critic (SAC) framework to optimize a Q-function directly parameterized by the language model.<n> Experimental results on two math problem-solving datasets, GSM8K and MATH, demonstrate that DQO outperforms previous methods, establishing it as a promising offline reinforcement learning approach for aligning language models.
  arXiv  Detail & Related papers  (2024-10-11T23:29:20Z)
• Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications and Opportunities [89.40778301238642]
  Model merging is an efficient empowerment technique in the machine learning community. There is a significant gap in the literature regarding a systematic and thorough review of these techniques.
  arXiv  Detail & Related papers  (2024-08-14T16:58:48Z)
• MindStar: Enhancing Math Reasoning in Pre-trained LLMs at Inference Time [51.5039731721706]
  MindStar is a purely inference-based searching method for large language models. It formulates reasoning tasks as searching problems and proposes two search ideas to identify the optimal reasoning paths. It significantly enhances the reasoning abilities of open-source models, such as Llama-2-13B and Mistral-7B, and achieves comparable performance to GPT-3.5 and Grok-1.
  arXiv  Detail & Related papers  (2024-05-25T15:07:33Z)
• Mixture-of-Instructions: Aligning Large Language Models via Mixture Prompting [7.103987978402038]
  We introduce a novel technique termed Mixture-of-Instructions (MoI)<n>MoI employs a strategy of instruction packing combined with diverse system prompts to boost the alignment efficiency of language models.<n>Our methodology was applied to the open-source Qwen-7B-chat model, culminating in the development of Qwen-SFT-MoI.
  arXiv  Detail & Related papers  (2024-04-29T03:58:12Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.