M2A: Synergizing Mathematical and Agentic Reasoning in Large Language Models

• URL: http://arxiv.org/abs/2605.09879v1
• Date: Mon, 11 May 2026 02:05:30 GMT
• Title: M2A: Synergizing Mathematical and Agentic Reasoning in Large Language Models
• Abstract summary: We propose M2A, a novel paradigm that synergizes mathematical and agentic reasoning via model merging.<n>M2A operates directly in parameter space, identifies the feature subspace critical for agent behavior, and merges the mathematical reasoning task vector only along its null space.<n>Unlike SFT or RL, M2A requires no additional gradient-update and exposes the coefficient merging as a simple knob for controlling reasoning length.
• Score: 9.712139528293946
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: While reasoning has become a central capability of large language models (LLMs), the reasoning patterns required for different scenarios are often misaligned. Mathematical reasoning typically relies on intrinsic logic to solve closed-world problems in a single response, whereas agentic reasoning requires not only internal reasoning but also multi-turn interaction with external environments, interleaving thought and action. This misalignment prevents mathematical and agentic reasoning from effectively benefiting from each other, often yielding unstable reasoning behavior and only limited performance gains under multi-task learning. In this paper, we propose M2A, a novel paradigm that synergizes mathematical and agentic reasoning via model merging. To avoid overfitting to superficial reasoning patterns under joint training, M2A operates directly in parameter space: it identifies the feature subspace critical for agent behavior, and merges the mathematical reasoning task vector only along its null space, thereby injecting reasoning capability along directions that do not perturb agent behavior. Unlike SFT or RL, M2A requires no additional gradient-update and exposes the merging coefficient as a simple knob for controlling reasoning length. Experiments in a challenging real-world coding agent setting show that our method effectively extends agentic reasoning depth and delivers substantial performance improvements. Applied to a fine-tuned Qwen3-8B, M2A improves its SWE-Bench Verified resolved rate from 44.0% to 51.2% without retraining the model. Code is available at https://github.com/laplucky/M2A.git.

Related papers

• Neuro-symbolic PRM: Enhancing Scientific Reasoning via Structured Traces and Symbolic Verification [10.892972402027494]
  Intermediate reasoning steps are syntactically well-formed, mathematically executable, and unit-consistent, yet contextually ungrounded.<n>Current approaches either rely on formal verifiers that cannot assess semantic intent, or burden Process Reward Models (PRMs) with the dual task of checking both arithmetic and logic.<n>We propose a neuro-symbolic framework that cleanly decouples reasoning into two formal dimensions: Symbolic Validity ($V$) and Semantic Groundedness ($G$)
  arXiv  Detail & Related papers  (2026-08-26T19:10:12Z)
• Switch-Reasoner: Learn When to Think in Multitask Mixtures via Reinforcement Learning [72.45536737928015]
  We propose Switch-Reasoner, a framework that learns to adaptively select reasoning modes for MLLMs.<n>It treats thinking as a virtual tool invocation and allows the model to either answer directly or invoke explicit reasoning before answering.<n>Experiments on 11 multimodal tasks show that Switch-Reasoner reduces unnecessary reasoning while maintaining strong performance.
  arXiv  Detail & Related papers  (2026-07-09T15:04:05Z)
• MMEmb-R1: Reasoning-Enhanced Multimodal Embedding with Pair-Aware Selection and Adaptive Control [36.06412693943508]
  We propose MMEmb-R1, an adaptive reasoning-based multimodal embedding framework.<n>We formulate reasoning as a latent variable and introduce pair-aware reasoning selection to identify reasoning paths beneficial for query-target alignment.<n>Experiments on the MMEB-V2 benchmark demonstrate that our model achieves a score of 71.2 with only 4B parameters, establishing a new state-of-the-art while significantly reducing reasoning overhead and inference latency.
  arXiv  Detail & Related papers  (2026-04-07T17:55:17Z)
• To Think or Not To Think, That is The Question for Large Reasoning Models in Theory of Mind Tasks [56.11584171938381]
  Theory of Mind (ToM) assesses whether models can infer hidden mental states such as beliefs, desires, and intentions.<n>Recent progress in Large Reasoning Models (LRMs) has boosted step-by-step inference in mathematics and coding.<n>We present a systematic study of nine advanced Large Language Models (LLMs) comparing reasoning models with non-reasoning models.
  arXiv  Detail & Related papers  (2026-02-11T08:16:13Z)
• Beyond Model Scaling: Test-Time Intervention for Efficient Deep Reasoning [34.912727372324625]
  Think-with-Me is a test-time interactive reasoning paradigm that introduces external feedback intervention into the reasoning process.<n>Think-with-Me pauses reasoning at points for external feedback, adaptively extending or terminating reasoning to reduce redundancy while preserving accuracy.<n>Experiments show that Think-with-Me achieves a superior balance between accuracy and reasoning length under limited context windows.
  arXiv  Detail & Related papers  (2026-01-16T13:00:42Z)
• GSM-Agent: Understanding Agentic Reasoning Using Controllable Environments [56.007498767771075]
  GSM-Agent is a novel benchmark for evaluating agentic reasoning in complex environments.<n>We analyze the agentic reasoning patterns by cluster the environment's document embeddings into nodes, and map each tool call to its nearest node.<n>We propose a tool-augmented test-time scaling method to improve LLM's agentic reasoning performance by adding tools to encourage models to revisit.
  arXiv  Detail & Related papers  (2025-09-26T07:24:37Z)
• Let LRMs Break Free from Overthinking via Self-Braking Tuning [68.93713497579853]
  Large reasoning models (LRMs) have significantly enhanced their reasoning capabilities by generating longer chains of thought.<n>This performance gain comes at the cost of a substantial increase in redundant reasoning during the generation process.<n>We propose a novel framework, Self-Braking Tuning (SBT), which tackles overthinking from the perspective of allowing the model to regulate its own reasoning process.
  arXiv  Detail & Related papers  (2025-05-20T16:53:40Z)
• Thinkless: LLM Learns When to Think [57.857534644932194]
  Reasoning Language Models, capable of extended chain-of-thought reasoning, have demonstrated remarkable performance on tasks requiring complex logical inference.<n>We propose Thinkless, a learnable framework that empowers an LLM to adaptively select between short-form and long-form reasoning.<n>On several benchmarks such as Minerva Algebra, MATH-500, and GSM8K, Thinkless is able to reduce the usage of long-chain thinking by 50% - 90%.
  arXiv  Detail & Related papers  (2025-05-19T17:24:16Z)
• Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models [49.61246073215651]
  Large Language Models (LLMs) have demonstrated remarkable capabilities in complex tasks.<n>Recent advancements in OpenAI o1 and DeepSeek-R1 have further improved performance in System-2 reasoning domains.<n>However, they also introduce significant computational overhead due to verbose and redundant outputs.
  arXiv  Detail & Related papers  (2025-03-20T17:59:38Z)

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.