Cover Image

Memorize Theorems, Not Instances: Probing SFT Generalization through Mathematical Reasoning

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Peng, Ruiying, Yang, Mengyu, Lei, Jing, Li, Xiaohui, Wu, Xueyu, Chen, Xinlei
Format:	Preprint
Published:	2026
Subjects:	Machine Learning Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

VAR-MATH: Probing True Mathematical Reasoning in LLMS via Symbolic Multi-Instance Benchmarks
by: Yao, Jian, et al.
Published: (2025)

SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training
by: Chu, Tianzhe, et al.
Published: (2025)

AceReason-Nemotron 1.1: Advancing Math and Code Reasoning through SFT and RL Synergy
by: Liu, Zihan, et al.
Published: (2025)

Deeper Thought, Weaker Aim: Understanding and Mitigating Perceptual Impairment during Reasoning in Multimodal Large Language Models
by: Peng, Ruiying, et al.
Published: (2026)

Debunk the Myth of SFT Generalization
by: Lin, Xiaofeng, et al.
Published: (2025)

The Pitfalls of Memorization: When Memorization Hurts Generalization
by: Bayat, Reza, et al.
Published: (2024)

mSFT: Addressing Dataset Mixtures Overfitting Heterogeneously in Multi-task SFT
by: Koh, Woosung, et al.
Published: (2026)

A Practical Two-Stage Recipe for Mathematical LLMs: Maximizing Accuracy with SFT and Efficiency with Reinforcement Learning
by: Yoshihara, Hiroshi, et al.
Published: (2025)

Step-KTO: Optimizing Mathematical Reasoning through Stepwise Binary Feedback
by: Lin, Yen-Ting, et al.
Published: (2025)

Complementing reinforcement learning with SFT through logit averaging in the post training of LLMs
by: Gan, Xingwei, et al.
Published: (2026)

Re-understanding Graph Unlearning through Memorization
by: Ding, Pengfei, et al.
Published: (2026)

SFT-then-RL Outperforms Mixed-Policy Methods for LLM Reasoning
by: Limozin, Alexis, et al.
Published: (2026)

Auditing Reasoning-Trace Memorization Claims after Unlearning with Head-Conditioned Canaries
by: Li, Yanhang, et al.
Published: (2026)

DynamixSFT: Dynamic Mixture Optimization of Instruction Tuning Collections
by: Shin, Haebin, et al.
Published: (2025)

Memorization vs. Reasoning: Updating LLMs with New Knowledge
by: Li, Aochong Oliver, et al.
Published: (2025)

Quagmires in SFT-RL Post-Training: When High SFT Scores Mislead and What to Use Instead
by: Kang, Feiyang, et al.
Published: (2025)

MathMixup: Boosting LLM Mathematical Reasoning with Difficulty-Controllable Data Synthesis and Curriculum Learning
by: Li, Xuchen, et al.
Published: (2026)

Climbing the Ladder of Reasoning: What LLMs Can-and Still Can't-Solve after SFT?
by: Sun, Yiyou, et al.
Published: (2025)

Beyond Frequency: The Role of Redundancy in Large Language Model Memorization
by: Zhang, Jie, et al.
Published: (2025)

Critical Windows of Complexity Control: When Transformers Decide to Reason or Memorize
by: Ali, Sarwan
Published: (2026)

Fast and Lightweight Backdoor Detection via Head Random Probing
by: Yu, Yinbo, et al.
Published: (2026)

Metis-RISE: RL Incentivizes and SFT Enhances Multimodal Reasoning Model Learning
by: Qiu, Haibo, et al.
Published: (2025)

Memorization Sinks: Isolating Memorization during LLM Training
by: Ghosal, Gaurav R., et al.
Published: (2025)

Memorization in deep learning: A survey
by: Wei, Jiaheng, et al.
Published: (2024)

Consolidation or Adaptation? PRISM: Disentangling SFT and RL Data via Gradient Concentration
by: Zhao, Yang, et al.
Published: (2026)

Quantization Meets Reasoning: Exploring and Mitigating Degradation of Low-Bit LLMs in Mathematical Reasoning
by: Li, Zhen, et al.
Published: (2025)

Gradients Must Earn Their Influence: Unifying SFT with Generalized Entropic Objectives
by: Wang, Zecheng, et al.
Published: (2026)

RL Fine-Tuning Heals OOD Forgetting in SFT
by: Jin, Hangzhan, et al.
Published: (2025)

How Do Flow Matching Models Memorize and Generalize in Sample Data Subspaces?
by: Gao, Weiguo, et al.
Published: (2024)

CAMA: Enhancing Mathematical Reasoning in Large Language Models with Causal Knowledge
by: Zan, Lei, et al.
Published: (2025)

Decoding Generalization from Memorization in Deep Neural Networks
by: Ketha, Simran, et al.
Published: (2025)

FormalMATH: Benchmarking Formal Mathematical Reasoning of Large Language Models
by: Yu, Zhouliang, et al.
Published: (2025)

Analyzing Memorization in Large Language Models through the Lens of Model Attribution
by: Menta, Tarun Ram, et al.
Published: (2025)

Self-Error-Instruct: Generalizing from Errors for LLMs Mathematical Reasoning
by: Yu, Erxin, et al.
Published: (2025)

REFACTOR: Learning to Extract Theorems from Proofs
by: Zhou, Jin Peng, et al.
Published: (2024)

Provably Mitigating Overoptimization in RLHF: Your SFT Loss is Implicitly an Adversarial Regularizer
by: Liu, Zhihan, et al.
Published: (2024)

Implicit Reward as the Bridge: A Unified View of SFT and DPO Connections
by: Wang, Bo, et al.
Published: (2025)

GAC: Noise-Aware Adaptive Mixing for Hybrid SFT-RL Post-Training
by: Hu, Yuelin, et al.
Published: (2026)

Beyond Memorization: Extending Reasoning Depth with Recurrence, Memory and Test-Time Compute Scaling
by: Rodkin, Ivan, et al.
Published: (2025)

Advanced Weakly-Supervised Formula Exploration for Neuro-Symbolic Mathematical Reasoning
by: Wu, Yuxuan, et al.
Published: (2025)