The Missing Primitive: Diagnosing and Repairing Mathematical Reasoning in Large Language Models
cs.LG
Submitted: 2026-10-01
Updated: 2026-10-01
Project page: https://taco-group.github.io/Math-Primitive/ABSTRACT
Sources
- Large Language Models and Mathematical Reasoning Failures
- Early science acceleration experiments with GPT-5
- Humanity's Last Exam
- Training Verifiers to Solve Math Word Problems
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- Beyond Benchmarks: MathArena as an Evaluation Platform for Mathematics with LLMs
- Data for Mathematical Copilots: Better Ways of Presenting Proofs for Machine Learning
- FrontierMath: A Benchmark for Evaluating Advanced Mathematical Reasoning in AI
- Self-Distillation Zero: Self-Revision Turns Binary Rewards into Dense Supervision
- Measuring Mathematical Problem Solving With the MATH Dataset
- Point Convergence of Nesterov's Accelerated Gradient Method: An AI-Assisted Proof
- Rethinking On-Policy Self-Distillation for Thinking Models
- Executable Functional Abstractions: Inferring Generative Programs for Advanced Math Problems
- Learning to Reason with Insight for Informal Theorem Proving
- DemoPSD: Disagreement-Modulated Policy Self-Distillation
- AVSD: Adaptive-View Self-Distillation by Balancing Consensus and Teacher-Specific Privileged Signals
- gpt-oss-120b & gpt-oss-20b Model Card
- Privileged Information Distillation for Language Models
- Diagnosing and Mitigating Thinking Collapse in On-Policy Self-Distillation
- Decomposing Elements of Problem Solving: What "Math" Does RL Teach?
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks