Model Releases
SMRC: Aligning Large Language Models with Student Reasoning for Mathematical Error Correction
arXiv:2511.14684v2 Announce Type: replace Abstract: Large language models (LLMs) often make reasoning errors when solving mathematical problems, and how to automatically detect and correct these error
arXiv:2511.14684v2 Announce Type: replace Abstract: Large language models (LLMs) often make reasoning errors when solving mathematical problems, and how to automatically detect and correct these errors has become an important research direction. However, existing approaches extit{mainly focus on self-correction within the model}, which falls short of the "teacher-style" correction required in educational settings, extit{i.e.}, systematically guiding and revising a student' s problem-solving process. To address this gap, we propose exttt{SMRC} (extit{nderline{S}tudent nderline{M}athematical nderline{R}easoning nderline{C}orrection}), a novel method that aligns LLMs with student reasoning. Specifically, exttt{SMRC} formulates student reasoning as a multi-step sequential decision problem and introduces Monte Carlo Tree Search (MCTS) to explore optimal correction paths. To reduce the cost of the annotating process-level rewards, we leverage breadth-first search (BFS) guided by LLMs and final-answer evaluation to generate reward signals, which are then distributed across intermediate reasoning steps via a back-propagation mechanism, enabling fine-grained process supervision. Additionally, we construct a benchmark for high school mathematics, MSEB (Multi-Solution Error Benchmark), consisting of 158 instances that include problem statements, student solutions, and correct reasoning steps. We further propose a dual evaluation protocol centered on extbf{solution accuracy} and extbf{correct-step retention}, offering a comprehensive measure of educational applicability. Experiments demonstrate that exttt{SMRC} significantly outperforms existing methods on two public datasets (ProcessBench and MR-GSM8K) and our MSEB in terms of effectiveness and overall performance. The code are available at https://github.com/ECNU-RAIL/SMRC-EMNLP2026.
Related
- ErrorRadar: Benchmarking Complex Mathematical Reasoning of Multimodal Large Language Models Via Error Detection
- MathDebugger: Detecting and Diagnosing Errors in Synthetic Mathematical Data
- Toward Automated Robustness Evaluation of Mathematical Reasoning
- Not All Proofs Are Equal: Evaluating LLM Proof Quality Beyond Correctness
Source: arXiv cs.CL | 2026-08-31