Image de couverture de livre

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

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Dong, Fangan, Yan, Zuming, Ge, Xuri, Xu, Zhiwei, Zhang, Mengqi, Chen, Xuanang, He, Ben, Xin, Xin, Chen, Zhumin, Zhou, Ying
Format:	Preprint
Publié:	2026
Sujets:	Computation and Language
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

MCoT-MVS: Multi-level Vision Selection by Multi-modal Chain-of-Thought Reasoning for Composed Image Retrieval
par: Ge, Xuri, et autres
Publié: (2026)

Reinforced Efficient Reasoning via Semantically Diverse Exploration
par: Zhao, Ziqi, et autres
Publié: (2026)

Improving Sequential Recommenders through Counterfactual Augmentation of System Exposure
par: Zhao, Ziqi, et autres
Publié: (2025)

Are Emotion and Rhetoric Neurons in LLM? Neuron Recognition and Adaptive Masking for Emotion-Rhetoric Prediction Steering
par: Zheng, Li, et autres
Publié: (2026)

R^3AG: First Workshop on Refined and Reliable Retrieval Augmented Generation
par: Wang, Zihan, et autres
Publié: (2024)

RankSteer: Activation Steering for Pointwise LLM Ranking
par: Wang, Yumeng, et autres
Publié: (2026)

UIPE: Enhancing LLM Unlearning by Removing Knowledge Related to Forgetting Targets
par: Wang, Wenyu, et autres
Publié: (2025)

Fractional Reasoning via Latent Steering Vectors Improves Inference Time Compute
par: Liu, Sheng, et autres
Publié: (2025)

Improving LLM Reasoning through Interpretable Role-Playing Steering
par: Wang, Anyi, et autres
Publié: (2025)

FairSteer: Inference Time Debiasing for LLMs with Dynamic Activation Steering
par: Li, Yichen, et autres
Publié: (2025)

Beyond Local Edits: Embedding-Virtualized Knowledge for Broader Evaluation and Preservation of Model Editing
par: Liu, Shuainan, et autres
Publié: (2026)

Multimodal Sentiment Analysis Based on Causal Reasoning
par: Chen, Fuhai, et autres
Publié: (2024)

Constrained Auto-Regressive Decoding Constrains Generative Retrieval
par: Wu, Shiguang, et autres
Publié: (2025)

Less is More: Improving LLM Reasoning with Minimal Test-Time Intervention
par: Yang, Zhen, et autres
Publié: (2025)

Beyond Local Views: Global State Inference with Diffusion Models for Cooperative Multi-Agent Reinforcement Learning
par: Xu, Zhiwei, et autres
Publié: (2024)

Enhancing Multi-hop Reasoning through Knowledge Erasure in Large Language Model Editing
par: Zhang, Mengqi, et autres
Publié: (2024)

Neuralink: Fast LLM Inference on Smartphones with Neuron Co-Activation Linking
par: Wang, Tuowei, et autres
Publié: (2024)

Enhancing Cross-task Transfer of Large Language Models via Activation Steering
par: Tang, Xinyu, et autres
Publié: (2025)

WASD: Locating Critical Neurons as Sufficient Conditions for Explaining and Controlling LLM Behavior
par: Yu, Haonan, et autres
Publié: (2026)

Steer Like the LLM: Activation Steering that Mimics Prompting
par: Heyman, Geert, et autres
Publié: (2026)

DiffuGR: Generative Document Retrieval with Diffusion Language Models
par: Zhao, Xinpeng, et autres
Publié: (2025)

Style-Specific Neurons for Steering LLMs in Text Style Transfer
par: Lai, Wen, et autres
Publié: (2024)

Self-Supervised Position Debiasing for Large Language Models
par: Liu, Zhongkun, et autres
Publié: (2024)

Steered LLM Activations are Non-Surjective
par: Mishra, Aayush, et autres
Publié: (2026)

WINA: Weight Informed Neuron Activation for Accelerating Large Language Model Inference
par: Chen, Sihan, et autres
Publié: (2025)

Steering Externalities: Benign Activation Steering Unintentionally Increases Jailbreak Risk for Large Language Models
par: Xiong, Chen, et autres
Publié: (2026)

Beyond Steering Vector: Flow-based Activation Steering for Inference-Time Intervention
par: Jin, Zehao, et autres
Publié: (2026)

P-Guide: Parameter-Efficient Prior Steering for Single-Pass CFG Inference
par: Peng, Xin, et autres
Publié: (2026)

Steer2Edit: From Activation Steering to Component-Level Editing
par: Sun, Chung-En, et autres
Publié: (2026)

TACT: Mitigating Overthinking and Overacting in Coding Agents via Activation Steering
par: Sui, Yuan, et autres
Publié: (2026)

LLMSteer: Improving Long-Context LLM Inference by Steering Attention on Reused Contexts
par: Gu, Zhuohan, et autres
Publié: (2024)

Knowledge Graph Enhanced Large Language Model Editing
par: Zhang, Mengqi, et autres
Publié: (2024)

Uncovering Overfitting in Large Language Model Editing
par: Zhang, Mengqi, et autres
Publié: (2024)

AI-Salesman: Towards Reliable Large Language Model Driven Telemarketing
par: Zhang, Qingyu, et autres
Publié: (2025)

Adaptive Probe-based Steering for Robust LLM Jailbreaking
par: Chen, Junxi, et autres
Publié: (2026)

ASLoRA: Adaptive Sharing Low-Rank Adaptation Across Layers
par: Hu, Junyan, et autres
Publié: (2024)

Uncovering Selective State Space Model's Capabilities in Lifelong Sequential Recommendation
par: Yang, Jiyuan, et autres
Publié: (2024)

CoCoNUTS: Concentrating on Content while Neglecting Uninformative Textual Styles for AI-Generated Peer Review Detection
par: Chen, Yihan, et autres
Publié: (2025)

Spectral Characterization and Mitigation of Sequential Knowledge Editing Collapse
par: Zhang, Chi, et autres
Publié: (2026)

Reasoning Runtime Behavior of a Program with LLM: How Far Are We?
par: Chen, Junkai, et autres
Publié: (2024)