Accepted Papers

1. Tales from a Graph: a Pipeline for Mathematical Problem Generation
2. Meta Thinker: Thinking What AI Thinks
3. SATBench: Benchmarking LLMs' Logical Reasoning via Automated Puzzle Generation from SAT Formulas
4. Probabilistic Soundness Guarantees in LLM Reasoning Chains
5. I-RAVEN-X: Benchmarking Generalization and Robustness of Analogical and Mathematical Reasoning in Large Language and Reasoning Models
6. Axiom-Aware FunSearch for Non-Constructive Mathematics
7. CauSciBench: Assessing LLM Causal Reasoning for Scientific Research
8. PRISM-Physics: Causal DAG-Based Process Evaluation for Physics Reasoning
9. Aryabhata: An exam-focused language model for JEE Math
10. Scratchpad Thinking: Alternation Between Storage and Computation in Latent Reasoning Models
11. Curiosity-driven RL for symbolic equation solving
12. Decompose, Adapt, and Evolve: Towards Efficient Scientific Equation Discovery with Large Language Models
13. Adaptive Coopetition: Leveraging Coarse Verifier Signals for Resilient Multi-Agent LLM Reasoning
14. Infinite-Dimensional HiPPO Provides an Explicit Formula for LSSLs
15. MathSticks: A Benchmark for Visual Symbolic Compositional Reasoning with Matchstick Puzzles
16. ThinkEdit: Interpretable Weight Editing to Mitigate Overly Short Thinking in Reasoning Models
17. Quagmires in SFT-RL Post-Training: When High SFT Scores Mislead and What to Use Instead
18. Investigating the interaction of linguistic and mathematical reasoning in language models using multilingual number puzzles
19. Towards Scaling Laws for Symbolic Regression
20. Tree-OPO: Off-policy Monte Carlo Tree-Guided Advantage Optimization for Multistep Reasoning
21. Minif2f in Rocq: Automatic Translation Between Proof Assistants — A Case Study
22. OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles
23. EchoRL: Learning to Plan through Experience for Efficient Reinforcement Learning
24. Limits of PRM-Guided Tree Search for Mathematical Reasoning with LLMs
25. CDE: Curiosity-Driven Exploration for Efficient Reinforcement Learning in Large Language Models
26. DreamPRM: Domain-Reweighted Process Reward Model for Multimodal Reasoning
27. RADAR: Reasoning–Ability and Difficulty-Aware Routing for Reasoning LLMs
28. Compute as Teacher: Turning Inference Compute Into Reference-Free Supervision
29. TRACE: A Framework for Analyzing and Enhancing Stepwise Reasoning in Vision-Language Models
30. Decoupling Reasoning from Proving: A New Framework for Tackling Olympiad-Level Mathematics
31. Analytical Lyapunov Function Discovery: An RL-based Generative Approach
32. You Need Reasoning to Learn Reasoning: The Limitations of Label-Free RL in Weak Base Models
33. PVSGym: A Proof Learning Environment
34. ProxyThinker: Test-Time Guidance through Small Visual Reasoners
35. Nested Depth Generalization in Transformers
36. Learning Modular Exponentiation with Transformers
37. \textsc{Gambit}: Generating Automated Mathematical Bounds, Inequalities, and Theorems
38. Specifying exact circuit algorithms in universal transformers
39. Tool-Assisted Multi-Turn Theorem Proving with LLMs
40. A Small Math Model: Recasting Strategy Choice Theory in an LLM-Inspired Architecture
41. Solving Inequality Proofs with Large Language Models
42. Beyond Accuracy: Evaluating Multimodal Mathematical and Scientific Reasoning Through Error Analysis and Self-Correction
43. Process-Verified Reinforcement Learning for Theorem Proving via Lean
44. LLM-Generated Search Heuristics Can Solve Open Instances of Combinatorial Design Problems
45. Expanding the Action Space of LLMs to Reason Beyond Language
46. Single-stream Policy Optimization
47. SpotIt: Evaluating Text-to-SQL Evaluation with Formal Verification
48. CircuitSense: A Hierarchical Circuit System Benchmark Bridging Visual Comprehension and Symbolic Reasoning in Engineering Design Process
49. Can Large Language Models Learn Formal Logic? A Data-Driven Training and Evaluation Framework
50. Why Reinforcement Learning Struggles with Expression Simplification: A Reward Analysis
51. AI Impact on Human Proof Formalization Workflows
52. One Token to Fool LLM-as-a-Judge
53. Modeling Chain-of-Thought Collapse in Pruned Language Models: Fidelity and Similarity Analysis for Mathematical Reasoning
54. LeanDojo-v2: A Comprehensive Library for AI-Assisted Theorem Proving in Lean
55. From EduVisBench to EduVisAgent: A Benchmark and Multi-Agent Framework for Reasoning-Driven Pedagogical Visualization
56. Learning Permuted Congruential Sequences with Transformers
57. In-the-Flow Agentic System Optimization for Effective Planning and Tool Use
58. SciML Agents: Write the Solver, Not the Solution
59. Concept Generalization in Humans and Large Language Models: Insights from the Number Game
60. Hilbert: Recursively Building Formal Proofs with Informal Reasoning
61. Risk-Sensitive Reinforcement Learning for Alleviating Exploration Dilemmas in Large Language Models
62. FormalML: A Benchmark for Evaluating Formal Subgoal Completion in Machine Learning Theory
63. PAG: Multi-Turn Reinforced LLM Self-Correction with Policy as Generative Verifier
64. Mirage or Method? How Model–Task Alignment Induces Divergent RL Conclusions
65. AI-Driven Mathematical Discovery for the Andrews–Curtis Conjecture
66. DAG-Math: Graph-Guided Mathematical Reasoning in LLMs
67. ProofOptimizer: Training Language Models to Simplify Proofs without Human Demonstrations
68. A Matter of Interest: Understanding Interestingness of Math Problems in Humans and Language Models
69. Co-rewarding: Stable Self-supervised RL for Eliciting Reasoning in Large Language Models
70. On the Design of KL-Regularized Policy Gradient Algorithms for LLM Reasoning
71. R-Zero: Self-Evolving Reasoning LLM from Zero Data
72. Exploration v.s. Exploitation: Rethinking RLVR through Clipping, Entropy, and Spurious Reward
73. Controllable Mathematical Reasoning via Self-Optimizing Thought Vectors
74. RLVR vs. Distillation: Understanding Accuracy and Capability in LLM Mathematical Reasoning
75. Combining Textual and Structural Information for Premise Selection in Lean
76. Improving autoformalization via cycle consistency and incremental type-checking using language-model probabilistic programs
77. Beyond Correctness: Harmonizing Process and Outcome Rewards through RL Training
78. Adaptive Control for Test-time Scaling
79. Amortized Latent Steering: Low-Cost Alternative to Test-Time Optimization
80. AntiderivBench: Evaluating language models on indefinite integration
81. Improving ML attacks on LWE with data repetition and stepwise regression
82. Credit Cards, Confusion, Computation, and Consequences: How Well Do LLMs Reason About Financial Literacy?
83. Kimina Lean Server: A High-Performance Lean Server for Large-Scale Verification
84. Restructuring the Corpus Makes RAG Work for Math
85. Numbers Already Carry Their Own Embeddings
86. Simultaneous Multi-objective Alignment Across Verifiable and Non-verifiable Rewards
87. OMEGA: Can LLMs Reason Outside the Box in Math? Evaluating Exploratory, Compositional, and Transformative Generalization
88. Stoic Reasoner: Dual-Mode Transformers that Compress to Think and Decompress to Speak
89. FoCus: Improving Faithfulness in Chain-of-Thoughts by Training on Structured Reasoning Data
90. DELTA: How Does RL Unlock and Transfer New Algorithms in LLMs?
91. Scaling Generative Verifiers For Natural Language Mathematical Proof Verification And Selection
92. MathBode: Understanding LLM Reasoning with Dynamical Systems
93. STAT: Skill-Targeted Adaptive Training
94. SAND-Math: Using LLMs to Generate Novel, Difficult and Useful Mathematics Questions and Answers
95. Learning How to Use Tools, Not Just When: Pattern-Aware Tool-Integrated Reasoning
96. Babel-formal: Translation of Proofs between Lean and Rocq
97. HeuriGym: An Agentic Benchmark for LLM-Crafted Heuristics in Combinatorial Optimization
98. Usefulness-Driven Learning of Formal Mathematics
99. Stepwise Guided Policy Optimization: Coloring your Incorrect Reasoning in GRPO
100. Bridging Vision, Language, and Mathematics: Pictographic Character Reconstruction with Bézier Curves
101. Scheherazade: Evaluating Chain-of-Thought Math Reasoning in LLMs with Chain-of-Problems
102. VeriBench-FTP: A Formal Theorem Proving Benchmark in Lean 4 for Code Verification
103. STELAR-VISION: Self-Topology-Aware Efficient Learning for Aligned Reasoning in Vision
104. HYBRIDMIND: Meta Selection of Natural Language and Symbolic Language for Enhanced LLM Reasoning
105. StreetMath: Study of LLMs’ Approximation Behaviors
106. Climbing the Ladder of Reasoning: What LLMs Can—and Still Can’t—Solve after SFT?
107. Inpainting-Guided Policy Optimization for Diffusion Large Language Models
108. How does RL induce skill composition? A Case Study using Countdown
109. Blind Spot Navigation in Large Language Model Reasoning with Thought Space Explorer
110. CombiGraph-Vis: A Curated Multimodal Olympiad Benchmark for Discrete Mathematical Reasoning
111. Pretraining Scaling Laws for Generative Evaluations of Language Models
112. RAISE: Enhancing Scientific Reasoning in LLMs via Step-by-Step Retrieval
113. Reliable Fine-Grained Evaluation of Natural Language Math Proofs
114. SPG: Sandwiched Policy Gradient for Mask Diffusion Language Models
115. On the Evolution of Language Models without Labels: Majority Drives Selection, Novelty Promotes Variation
116. Measuring Off-Trajectory Math Reasoning of LLMs
117. AlphaOne: Reasoning Models Thinking Slow and Fast at Test Time
118. Towards Understanding Self-play for LLM Reasoning
119. Systematic Diagnosis of Brittle Reasoning in Large Language Models
120. Scaling up Multi-Turn Off-Policy RL and Multi-Agent Tree Search for LLM Step-Provers
121. IMProofBench: Benchmarking AI on Research-Level Mathematical Proof Generation
122. Landscape of Thoughts: Visualizing the Reasoning Process of Large Language Models
123. Revisiting the Uniform Information Density Hypothesis in LLM Reasoning Traces
124. Automated Discovery of Conservation Laws via Hybrid Neural ODE-Transformers
125. MathNet: a Global Multimodal Benchmark for Mathematical Reasoning and Retrieval
126. BrokenMath: A Benchmark for Sycophancy in Theorem Proving with LLMs
127. Best-of-L: Cross-Lingual Reward Modeling for Mathematical Reasoning
128. Understanding Tool-Integrated Reasoning
129. CoDaPO: Confidence and Difficulty-Adaptive Policy Optimization for Language Models
130. ProofGym: Unifying LLM-Based Theorem Proving Across Formal Systems
131. Unspoken Logic: Understanding and bridging the gap between free-form and LLM-interpretable natural language mathematical proofs
132. Evaluating Spatial Reasoning in Language Models
133. Faults in our Formal Benchmarks
134. CayleyPy Growth: Efficient growth computations and hundreds of new conjectures on Cayley graphs
135. In Good GRACES: Principled Teacher Selection for Knowledge Distillation
136. Why GRPO Needs Normalization: A Local-Curvature Perspective on Adaptive Gradients
137. Layer Importance for Mathematical Reasoning is Forged in Pre-Training and Invariant after Post-Training
138. DiagramIR: An Automatic Pipeline for Educational Math Diagram Evaluation
139. ARM: Discovering Agentic Reasoning Modules for Mathematical Problem-Solving
140. Skill-Aware Data Selection and Fine-Tuning for Data-Efficient Reasoning Distillation
141. Limits of Generalization in RLVR: Two Case Studies in Mathematical Reasoning
142. Reinforcement Learning for Hierarchical Proof Generation in Lean 4
143. Exact Learning of Arithmetic with Differentiable Agents
144. Think, Align, Select: Query–Key Scores for LLM Reasoning
145. FractalBench: Diagnosing Visual-Mathematical Reasoning Through Recursive Program Synthesis
146. Fractional Reasoning via Latent Steering Vectors Improves Inference Time Compute
147. Local Coherence or Global Validity? Investigating RLVR Traces in Math Domains
148. A NUMA Aware Compiler Framework for Large Scale Mathematical Reasoning Inference on PCIe Based Multi Accelerator Systems
149. Winning Gold at IMO 2025 with a Model-Agnostic Verification-and-Refinement Pipeline
150. Learning to Reason on Hard Problems with Privileged On-Policy Exploration
151. Patching Gaps In LLM Reasoning With Interventional Training
152. A Toolbox, Not a Hammer -- Multi-TAG: Scaling Math Reasoning with Multi-Tool Aggregation
153. RefGrader: Automated Grading of Mathematical Competition Proofs using Agentic Workflows

The list of accepted papers can be found on OpenReview here.

Reviewers

We are grateful to our fantastic reviewers for making our workshop reviewing process run smoothly: