paper-with-me

Papers Mathematical Reasoning

“Mathematical Reasoning” 태그가 달린 논문 1,929편 · 필터 해제

From Symbolic Perception to Logical Deduction: A Framework for Guiding Language Models in Geometric Reasoning

2026-09-09 · Weichen Dai, Rafael Medeiros Cabral, Ziyi Shou, Yan Cao 외 arxiv

Plane geometry remains a significant challenge in AI, requiring the integration of visual perception and mathematical reasoning. While Large Multimodal Models (LMMs) naturally handle visuo-linguistic inputs, they are oft…

Mathematical Reasoning

Which Tokens Should SFT Actually Learn? A Token-Trimming Perspective on Mathematical Reasoning

2026-09-09 · Yaning Jia, Chunhui Zhang, Wenxuan Xu, Xingjian Diao 외 arxiv

Supervised fine-tuning (SFT) applies a uniform cross-entropy loss to all target tokens, even though different tokens provide unequal learning signals for mathematical reasoning. This uniform treatment can over-sharpen al…

Mathematical Reasoning

Difficulty-Adaptive Tree-Structured Policy Optimization for Expanding Reasoning Coverage in RLVR

2026-09-08 · Youngjun Yu, Sanghwan Jang, Hwanjo Yu hf

Reinforcement Learning with Verifiable Rewards (RLVR) has been central to the recent success of Large Reasoning Models. However, while RLVR significantly improves single-sample accuracy, it often fails to expand the mode…

Mathematical ReasoningReinforcement Learning

RISE: Recursive Improvement via Self-Extrapolating Policy Distillation

2026-09-04 · Yang Li, Semih Yavuz, Shafiq Joty arxiv

On-policy distillation (OPD) provides dense, per-token supervision for language model post-training, but its effectiveness is bottlenecked by teacher quality: external teachers suffer from distribution mismatch, while se…

Mathematical ReasoningCode Generation

Do LLMs Exhibit Coherent Knowledge Structures in Mathematical Reasoning? A Perspective from Knowledge Space Theory

2026-09-04 · Peng Cui, Heejin Do, Mrinmaya Sachan arxiv

Human knowledge is inherently structured and interdependent: mastery of a concept requires prior mastery of its prerequisites, a principle formalized by Knowledge Space Theory (KST). While LLMs achieve strong performance…

Mathematical Reasoning

First Things First: Teaching LLM-Based Agents to Prioritize Must-Haves before Nice-to-Haves

2026-09-04 · Tianjie Ju, Xinyue Xu, Wanxuan Sun, Lingxiao Diao 외 arxiv

Recent progress in multimodal large language models (MLLMs) has fueled significant enthusiasm in their potential to act as autonomous agents for real-world tasks. However, scenarios requiring agents to fulfill users' com…

Reinforcement LearningMathematical Reasoning

Extremely Sparse Supervision Incentivizes Reasoning Ability

2026-09-03 · Zhishuai Liu, Xingzi Xu, Mehmet Saygin Seyfioglu, Pan Xu 외 arxiv

Large language models demonstrate increasingly strong reasoning capabilities through effective post-training. Yet, prevailing post-training methods optimize over massive numbers of tokens, implicitly assuming that effect…

Reinforcement LearningMathematical Reasoning

FlowBalance: Verifier-Grounded Self-Improvement from On-Policy Reasoning Experience

2026-09-03 · Zixun Huang, Kishan Panaganti, Haitao Mi, Leowei Liang hf

A reasoning model can improve from its own on-policy experience, but this inner loop is fragile: terminal verifiers provide reliable yet sparse supervision, while dense same-model guidance can reinforce false confidence …

Mathematical Reasoning

DiagEvo: Diagnosis-Guided Self-Evolution via Hierarchical Error Memory

2026-09-01 · Xincheng Wei, Yifan Ding, Yoshua Li, Dongsheng Ma 외 hf

Self-play is an effective paradigm for language-model self-evolution, but without guidance, solver performance can plateau or decline across rounds. Unguided methods steer question generation with signals such as difficu…

Mathematical ReasoningQuestion Generation

More Capable, Less Faithful: A Multilingual Analysis of Mathematical (Un)Solvability Detection in LLMs

2026-08-31 · Maria-Eleni Zoumpoulidi, Nikolaos Xiros, Georgios Paraskevopoulos arxiv

Solvability detection is one of the most challenging aspects of mathematical reasoning for Large Language Models (LLMs). While prior work has studied this capability extensively, these analyses have been limited to Engli…

Mathematical Reasoning

Reactivating Test-Time Scaling for Plane Geometry Problem Solving

2026-08-31 · Xiaoqiang Kang, Shengen Wu, Maizhen Ning, Xiaobo Jin 외 arxiv

Plane geometry problem (PGP) solving has become a critical benchmark for multimodal reasoning because it requires accurate visual perception and precise multi-step symbolic deduction. Although test-time scaling (TTS) has…

Mathematical ReasoningMultimodal ReasoningVisual Grounding

WHALE: A Simple Recipe for Joint Harness-Weight Optimization

2026-08-31 · Haechan Kim, Yoonho Lee, Gisang Lee, Chelsea Finn 외 hf

Agent performance depends jointly on the model parameters and the executable harness code that manages context and control flow. Optimizing either component in isolation can leave the system bottlenecked by its frozen co…

Mathematical ReasoningQuestion Answering

DARTS: Decoder-Aware Representation Tuning via Surgery for Model Merging

2026-08-28 · Aaryan Ajay Sharma, Sai Nishanth Padala, Seganrasan Subramanian arxiv

Model merging combines multiple task-specific fine-tuned LLMs into a single multi-task model without additional training. However, merged models are known to suffer from representation bias: systematic drift between the …

Mathematical ReasoningInstruction FollowingCode Generation

NL2AGBench: Benchmarking LLM Auto-Formalization for AlphaGeometry

2026-08-28 · Samuel Xiao, Judy Song, Rory Hu, Ziliang Zong arxiv

Recent advances in large language models (LLMs) have demonstrated strong capabilities in natural language understanding and mathematical reasoning. However, their ability to translate informal mathematical problems into …

Natural Language UnderstandingMathematical Reasoning

Learning to Use Tools: Reinforcement Learning for Tool-Integrated Mathematical Reasoning

2026-08-28 · Minghui Xu, Zi Wang arxiv

Current large language models (LLMs) increasingly benefit from external tool integration, especially for tasks requiring reliable computation and verification. Motivated by this, we study calculator tool calling for impr…

Reinforcement LearningMathematical Reasoning

TTPO: Test-Time Policy Optimization

2026-08-27 · Aozhe Wang, Zhengxi Lu, Jianze Wang, Shangke Lv 외 hf

Recent prominent post-training methods, such as Reinforcement Learning (RL) and On-Policy Self-Distillation (OPSD), have driven rapid progress in mathematical reasoning for large language models, yet their reliance on gr…

Mathematical ReasoningReinforcement Learning

From Atomic to Agentic: Towards Interpretable Evaluation of LLMs' Agentic Mathematical Capabilities

2026-08-27 · Jiayi Kuang, Yinghui Li, Yunze Song, Keyu Chen 외 arxiv

Large Language Models (LLMs) are evolving from performing end-to-end mathematical reasoning to integrating agentic intelligence. However, most existing math benchmarks evaluate only final answers. This outcome-oriented e…

Mathematical Reasoning

INSPIRE: An Internalize-Then-Improve Approach for Example-Driven Mathematical Reasoning

2026-08-27 · Shuai Wang, Jiayi Kuang, Yinghui Li, Haojing Huang 외 arxiv

Mathematical reasoning has seen rapid progress in large language models (LLMs), yet existing methods optimize predominantly for final-answer correctness, raising the question whether models truly internalize mathematical…

Mathematical Reasoning

ProgRouter: Online Progress-Guided Orchestration for Multi-Agent LLM Workflows under Quality-Cost Tradeoffs

2026-08-26 · Somgyuan Li, Ahmed M. Abdelmoniem, Shiqiang Wang arxiv

Multi-agent large language model (LLM) workflows have emerged as a powerful paradigm for solving complex, open-ended tasks through collaborative reasoning among specialized LLM agents, but they incur substantial operatin…

Mathematical ReasoningQuestion AnsweringCode Generation

One Symptom, Three Levers: A Critical Review of On-Policy Self-Distillation

2026-08-26 · Justin Robert, Raheel Qader arxiv

On-policy distillation trains a language model on its own generations while a teacher scores them token by token. It combines the dense supervision of imitation learning with the on-policy sampling of reinforcement learn…

Mathematical ReasoningReinforcement Learning
1–20 / 1,929 다음 →