mlsanityData to 14 Oct 2026

Does Math Reasoning Improve General LLM Capabilities? Understanding Transferability of LLM Reasoning

arXiv 2507.00432 ↗First posted 1 Jul 2025cs.AI cs.CLcited by 106

On the map: LLM reasoning → Mathematical reasoning

Abstract (arXiv)

Math reasoning has become the poster child of progress in large language models (LLMs), with new models rapidly surpassing human-level performance on benchmarks like MATH and AIME. But as math leaderboards improve week by week, it is worth asking: do these gains reflect broader problem-solving ability or just narrow overfitting? To answer this question, we evaluate over 20 open-weight reasoning-tuned models across a broad suite of tasks, including math, scientific QA, agent planning, coding, and standard instruction-following. We surprisingly find that most models that succeed in math fail to transfer their gains to other domains. To rigorously study this phenomenon, we conduct controlled experiments on Qwen3-14B models using math-only data but different tuning methods. We find that reinforcement learning (RL)-tuned models generalize well across domains, while supervised fine-tuning (SFT)-tuned models often forget general capabilities. Latent-space representation and token-space distribution shift analyses reveal that SFT induces substantial representation and output drift, while RL preserves general-domain structure. Our results suggest a need to rethink standard post-training recipes, particularly the reliance on SFT-distilled data for advancing reasoning models.

Key facts. Cited by 106 papers in the map (data through 14 Oct 2026). Placed in Mathematical reasoning, one of 17 directions in LLM reasoning. Builds on 60 of its 72 references that are in the map. First posted on arXiv 1 Jul 2025.

Builds on

60 of 72 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025OMEGA: Can LLMs Reason Outside the Box in Math? Evaluating Exploratory, Compositional, and Transformative Generalization23 Jun
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models5 Feb
  3. 2023WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct18 Aug

Other references · 57

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  2. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  3. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,708
  4. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  5. 2019PIQA: Reasoning about Physical Commonsense in Natural Languagecited by 2,883
  6. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,879
  7. 2018Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answeringcited by 2,164
  8. 2019Fine-Tuning Language Models from Human Preferencescited by 2,150
  9. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 2,057
  10. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 2,004
  11. 2018CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledgecited by 1,822
  12. 2024LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Modelscited by 1,464
38 more
  1. 2025s1: Simple test-time scalingcited by 1,341
  2. 2023Instruction-Following Evaluation for Large Language Modelscited by 1,228
  3. 2024OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problemscited by 1,136
  4. 2023Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotationscited by 928
  5. 2017RACE: Large-scale ReAding Comprehension Dataset From Examinationscited by 861
  6. 2017Crowdsourcing Multiple Choice Science Questionscited by 765
  7. 2022Fine-Tuning can Distort Pretrained Features and Underperform Out-of-Distributioncited by 739
  8. 2023The Flan Collection: Designing Data and Methods for Effective Instruction Tuningcited by 739
  9. 2023MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Modelscited by 735
  10. 2018CoQA: A Conversational Question Answering Challengecited by 683
  11. 2025SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-trainingcited by 672
  12. 2025SimpleRL-Zoo: Investigating and Taming Zero Reinforcement Learning for Open Base Models in the Wildcited by 511
  13. 2023Faith and Fate: Limits of Transformers on Compositionalitycited by 503
  14. 2023MAmmoTH: Building Math Generalist Models through Hybrid Instruction Tuningcited by 487
  15. 2023HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Modelscited by 341
  16. 2025Demystifying Long Chain-of-Thought Reasoning in LLMscited by 326
  17. 2025OpenThoughts: Data Recipes for Reasoning Modelscited by 304
  18. 2023The Unlocking Spell on Base LLMs: Rethinking Alignment via In-Context Learningcited by 257
  19. 2023TheoremQA: A Theorem-driven Question Answering datasetcited by 192
  20. 2025Crossing the Reward Bridge: Expanding RL with Verifiable Rewards Across Diverse Domainscited by 156
  21. 2024MAmmoTH2: Scaling Instructions from the Webcited by 145
  22. 2023OpenAgents: An Open Platform for Language Agents in the Wildcited by 137
  23. 2025General-Reasoner: Advancing LLM Reasoning Across All Domainscited by 135
  24. 2019"Going on a vacation" takes longer than "Going for a walk": A Study of Temporal Commonsense Understandingcited by 123
  25. 2025Small Models Struggle to Learn from Strong Reasonerscited by 113
  26. 2019HEAD-QA: A Healthcare Dataset for Complex Reasoningcited by 101
  27. 2025Revisiting Reinforcement Learning for LLM Reasoning from A Cross-Domain Perspectivecited by 76
  28. 2024Latent Space Chain-of-Embedding Enables Output-free LLM Self-Evaluationcited by 57
  29. 2025When Thinking Fails: The Pitfalls of Reasoning for Instruction-Following in LLMscited by 53
  30. 2024Multi-step Problem Solving Through a Verifier: An Empirical Analysis on Model-induced Process Supervisioncited by 52
  31. 2025Spurious Forgetting in Continual Learning of Language Modelscited by 51
  32. 2025Unlearning Isn't Deletion: Investigating Reversibility of Machine Unlearning in LLMscited by 42
  33. 2024ACPBench: Reasoning about Action, Change, and Planningcited by 39
  34. 2024On the Impact of Fine-Tuning on Chain-of-Thought Reasoningcited by 32
  35. 2025General Scales Unlock AI Evaluation with Explanatory and Predictive Powercited by 31
  36. 2025TinyV: Reducing False Negatives in Verification Improves RL for LLM Reasoningcited by 21
  37. 2024OpenRFT: Adapting Reasoning Foundation Model for Domain-specific Tasks with Reinforcement Fine-Tuningcited by 13
  38. 2025Temporal Sampling for Forgotten Reasoning in LLMscited by 12

50 most cited of 57 All 57 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 46; 2026: 60.

106 papers.

Sort:

By area:

  1. 2026Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Modelscited by 455
  2. 2026Self-Distillation Enables Continual Learningcited by 247
  3. 2025R-Zero: Self-Evolving Reasoning LLM from Zero Datacited by 228
  4. 2025A Survey of Reinforcement Learning for Large Reasoning Modelscited by 180
  5. 2025RL's Razor: Why Online Reinforcement Learning Forgets Lesscited by 176
  6. 2025On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectificationcited by 127
  7. 2025On the Interplay of Pre-Training, Mid-Training, and RL on Reasoning Language Modelscited by 69
  8. 2025The Path Not Taken: RLVR Provably Learns Off the Principalscited by 56
  9. 2025First Return, Entropy-Eliciting Explorecited by 45
  10. 2025OpenMMReasoner: Pushing the Frontiers for Multimodal Reasoning with an Open and General Recipecited by 37
  11. 2025Front-Loading Reasoning: The Synergy between Pretraining and Post-Training Datacited by 27
  12. 2025RLVER: Reinforcement Learning with Verifiable Emotion Rewards for Empathetic Agentscited by 25
38 more
  1. 2025SFT Doesn't Always Hurt General Capabilities: Revisiting Domain-Specific Fine-Tuning in LLMscited by 24
  2. 2025RL Grokking Recipe: How Does RL Unlock and Transfer New Algorithms in LLMs?cited by 23
  3. 2026Sparse but Critical: A Token-Level Analysis of Distributional Shifts in RLVR Fine-Tuning of LLMscited by 20
  4. 2025Proximal Supervised Fine-Tuningcited by 16
  5. 2026Multi-Task GRPO: Reliable LLM Reasoning Across Taskscited by 14
  6. 2026Rethinking Generalization in Reasoning SFT: A Conditional Analysis on Optimization, Data, and Model Capabilitycited by 14
  7. 2026On the Direction of RLVR Updates for LLM Reasoning: Identification and Exploitationcited by 13
  8. 2025Train-before-Test Harmonizes Language Model Rankingscited by 12
  9. 2025Self-Jailbreaking: Language Models Can Reason Themselves Out of Safety Alignment After Benign Reasoning Trainingcited by 12
  10. 2025Beyond Scaling Law: A Data-Efficient Distillation Framework for Reasoningcited by 11
  11. 2025Routine: A Structural Planning Framework for LLM Agent System in Enterprisecited by 9
  12. 2025LaSeR: Reinforcement Learning with Last-Token Self-Rewardingcited by 9
  13. 2025Token Buncher: Shielding LLMs from Harmful Reinforcement Learning Fine-Tuningcited by 8
  14. 2025Beyond English-Centric Training: How Reinforcement Learning Improves Cross-Lingual Reasoning in LLMscited by 8
  15. 2025One-Token Rollout: Guiding Supervised Fine-Tuning of LLMs with Policy Gradientcited by 8
  16. 2025Debunk the Myth of SFT Generalizationcited by 8
  17. 2025Meaningless Tokens, Meaningful Gains: How Activation Shifts Enhance LLM Reasoningcited by 8
  18. 2025How and Why LLMs Generalize: A Fine-Grained Analysis of LLM Reasoning from Cognitive Behaviors to Low-Level Patternscited by 8
  19. 2026Learning Robust Reasoning through Guided Adversarial Self-Playcited by 8
  20. 2025Reinforcement Learning for Tool-Integrated Interleaved Thinking towards Cross-Domain Generalizationcited by 7
  21. 2025FAPO: Flawed-Aware Policy Optimization for Efficient and Reliable Reasoningcited by 7
  22. 2026OmniOPD: Logit-Free On-Policy Distillation via Speculative Verificationcited by 7
  23. 2026On the Non-decoupling of Supervised Fine-tuning and Reinforcement Learning in Post-trainingcited by 6
  24. 2026RISER: Orchestrating Latent Reasoning Skills for Adaptive Activation Steeringcited by 6
  25. 2026Paying Less Generalization Tax: A Cross-Domain Generalization Study of RL Training for LLM Agentscited by 6
  26. 2025Correct Reasoning Paths Visit Shared Decision Pivotscited by 5
  27. 2025Zero Reinforcement Learning Towards General Domainscited by 5
  28. 2025Reassessing the Role of Supervised Fine-Tuning: An Empirical Study in VLM Reasoningcited by 5
  29. 2025Same or Not? Enhancing Visual Perception in Vision-Language Modelscited by 5
  30. 2026Learning from Mistakes: Negative Reasoning Samples Enhance Out-of-Domain Generalizationcited by 5
  31. 2026ETS: Energy-Guided Test-Time Scaling for Training-Free RL Alignmentcited by 5
  32. 2026CodeScaler: Scaling Code LLM Training and Test-Time Inference via Reward Modelscited by 5
  33. 2026From Reasoning Traces to Reusable Modules: Understanding Compositional Generalization in Language Model Reasoningcited by 5
  34. 2026Transferability for General Reasoning: An Automated Curriculum for Multi-Domain RLVRcited by 5
  35. 2025Sample-efficient LLM Optimization with Reset Replaycited by 4
  36. 2025Parallel Scaling Law: Unveiling Reasoning Generalization through A Cross-Linguistic Perspectivecited by 4
  37. 2025Extending RLVR to Open-Ended Tasks via Verifiable Multiple-Choice Reformulationcited by 4
  38. 2025Reinforcement Learning Improves Traversal of Parametric Knowledge in LLMscited by 4

50 most cited of 106 All 106 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025A Survey on Large Language Models for Mathematical Reasoning
  2. 2024From Next-Token to Mathematics: The Learning Dynamics of Mathematical Reasoning in Language Models
  3. 2025Beyond Accuracy: Dissecting Mathematical Reasoning for LLMs Under Reinforcement Learning
  4. 2025Teaching LLMs According to Their Aptitude: Adaptive Reasoning for Mathematical Problem Solving
  5. 2025Does Learning Mathematical Problem-Solving Generalize to Broader Reasoning?
  6. 2025VAR-MATH: Probing True Mathematical Reasoning in LLMS via Symbolic Multi-Instance Benchmarks
  7. 2025Scaling Behaviors of LLM Reinforcement Learning Post-Training: An Empirical Study in Mathematical Reasoning
  8. 2026Beyond Accuracy: Evaluating Strategy Diversity in LLM Mathematical Reasoning