MATH-AI @ NeurIPS 2025

113 papers

Top Research Directions

Research directions most represented by MATH-AI @ NeurIPS 2025 papers.

Overcoming the Limits of Multi Step Language Model Reasoning 8
Shifting from Static Benchmarks to Dynamic Evaluation 4
Granular Verification via Atomic Fact Decomposition 2
Digit Level Tokenization for Numerical Reasoning 2
Verifier Feedback Loops for Automated Proof Refinement 2
Evaluating Generative AI with Large Language Models 2
Cross Domain Generalization of Advanced Reasoning Frameworks 2
Prompting Large Language Models for Zero Shot Inference 2
The Performance Decay of Multi-Step Reasoning Chains 2
Sequential Error Propagation in Multi-Step Language Model Reasoning 2
How Fine Tuning Reshapes Internal Model Circuitry 1
Latent Computation Tokens for Transformer Reasoning 1
Extending Abelian Frameworks to Non-Abelian Symmetries 1
Fact Verification Through Informational Redundancy 1
The Divergence between Human and Machine Task Difficulty 1

Top Authors

AuthorPapers
Dong Yu5
Haitao Mi5
Dawn Song4
Brando Miranda3
Dian Yu3
Haolin Liu3
Joshua B. Tenenbaum3
Kaiyu Yang3
Katherine M. Collins3
Linfeng Song3
Nouha Dziri3
Sanmi Koyejo3
Vikas Yadav3
Wenjie Ma3
Yiyou Sun3
Zhenwen Liang3
Aadim Nepal2
Adrian Weller2
Alex Gu2
Amrith Setlur2

Impactful Papers

Papers from MATH-AI @ NeurIPS 2025 that shaped research clusters.

Influential

Papers

Title
Think, Align, Select: Query–Key Scores for LLM Reasoning
ThinkEdit: Interpretable Weight Editing to Mitigate Overly Short Thinking in Reasoning Models
Tool-Assisted Multi-Turn Theorem Proving with LLMs
Towards Scaling Laws for Symbolic Regression
Towards Understanding Self-play for LLM Reasoning
TRACE: A Framework for Analyzing and Enhancing Stepwise Reasoning in Vision-Language Models
Tree-OPO: Off-policy Monte Carlo Tree-Guided Advantage Optimization for Multistep Reasoning
Understanding Tool-Integrated Reasoning
Usefulness-Driven Learning of Formal Mathematics
VeriBench-FTP: A Formal Theorem Proving Benchmark in Lean 4 for Code Verification
Why Reinforcement Learning Struggles with Expression Simplification: A Reward Analysis
Winning Gold at IMO 2025 with a Model-Agnostic Verification-and-Refinement Pipeline
You Need Reasoning to Learn Reasoning: The Limitations of Label-Free RL in Weak Base Models

« prev Page 3/3