paper-with-me

홈 › Papers

Enhancing Large Language Model Reasoning via Selective Critical Token Fine-Tuning

2025-10-13 · Zhiwen Ruan, Yixia Li, He Zhu, Yun Chen, Peng Li, Yang Liu, Guanhua Chen arxiv

Large language models (LLMs) primarily rely on supervised fine-tuning (SFT) as a key method to adapt pre-trained models to domain-specific tasks such as mathematical reasoning. However, standard SFT uniformly penalizes all tokens, neglecting that only a small subset of critical tokens determines reasoning correctness. This uniform supervision often causes reduced output diversity and limited generalization. We propose Critical Token Fine-tuning (CFT), a simple yet effective approach that updates only tokens identified as functionally indispensable via counterfactual perturbations. By focusing gradient signals on these decisive reasoning steps while preserving the diversity of non-critical tokens, CFT can enhance both generation and diversity. Extensive experiments on five models across three families (Qwen, OLMo, LLaMA) and eleven mathematical reasoning benchmarks show that CFT, despite fine-tuning on less than 12% of tokens, consistently outperforms standard SFT. Moreover, CFT enables test-time scaling through improved sampling diversity and provides a stronger initialization for reinforcement learning, sustaining performance gains in later training stages while maintaining higher entropy for better exploration. These results highlight CFT as a practical and general framework for efficient and robust LLM fine-tuning.

📄 PDF Abstract BibTeX arXiv:2510.10974

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement LearningMathematical Reasoning

Similar Papers 제목 키워드 기반

Placing Puzzle Pieces Where They Matter: A Question Augmentation Framework for Reinforcement Learning

2026-04-17 · Yangyi Fang, Jiaye Lin, Xiaoliang Fu, Cong Qin 외 arxiv

Reinforcement learning has become a powerful approach for enhancing large language model reasoning, but faces a fundamental dilemma: training on easy problems can cause overfitting and pass@k degradation, while training …

Reinforcement LearningMathematical Reasoning

Identifying and Transferring Reasoning-Critical Neurons: Improving LLM Inference Reliability via Activation Steering

2026-01-27 · Fangan Dong, Zuming Yan, Xuri Ge, Zhiwei Xu 외 arxiv

Despite the strong reasoning capabilities of recent large language models (LLMs), achieving reliable performance on challenging tasks often requires post-training or computationally expensive sampling strategies, limitin…

Selective Latent Thinking: Adaptive Compression of LLM Reasoning Chains

2026-05-25 · Hui Xie, Jie Liu, Ziyue Qiao, Joaquin Vanschore arxiv

Explicit chain-of-thought (CoT) reasoning substantially improves the reasoning ability of large language models (LLMs), but incurs high inference cost due to lengthy autoregressive traces. Existing latent reasoning metho…

Reinforcement LearningMathematical Reasoning

Selective Expert Guidance for Effective and Diverse Exploration in Reinforcement Learning of LLMs

2025-10-05 · Zishang Jiang, Jinyi Han, Tingyun Li, Xinyi Wang 외 arxiv

Reinforcement Learning with Verifiable Rewards (RLVR) has become a widely adopted technique for enhancing the reasoning ability of Large Language Models (LLMs). However, the effectiveness of RLVR strongly depends on the …

Reinforcement Learning

Don't Look Only Once: Towards Multimodal Interactive Reasoning with Selective Visual Revisitation

2025-05-24 · Jiwan Chung, Junhyeok Kim, Siyeol Kim, Jaeyoung Lee 외

We present v1, a lightweight extension to Multimodal Large Language Models (MLLMs) that enables selective visual revisitation during inference. While current MLLMs typically consume visual input only once and reason pure…

Mathematical ReasoningMultimodal ReasoningVisual Grounding