Image de couverture de livre

Projecting Out the Malice: A Global Subspace Approach to LLM Detoxification

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Duan, Zenghao, Yin, Zhiyi, Shi, Zhichao, Pang, Liang, Jing, Shaoling, Huang, Zihe, Wu, Jiayi, Yan, Yu, Deng, Jingcheng, Shen, Huawei, Cheng, Xueqi
Format:	Preprint
Publié:	2026
Sujets:	Machine Learning Artificial Intelligence
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

GloSS over Toxicity: Understanding and Mitigating Toxicity in LLMs via Global Toxic Subspace
par: Duan, Zenghao, et autres
Publié: (2025)

Related Knowledge Perturbation Matters: Rethinking Multiple Pieces of Knowledge Editing in Same-Subject
par: Duan, Zenghao, et autres
Publié: (2025)

LLM Latent Reasoning as Chain of Superposition
par: Deng, Jingcheng, et autres
Publié: (2025)

LLM4MEA: Data-free Model Extraction Attacks on Sequential Recommenders via Large Language Models
par: Zhao, Shilong, et autres
Publié: (2025)

The Evolution of Thought: Tracking LLM Overthinking via Reasoning Dynamics Analysis
par: Wei, Zihao, et autres
Publié: (2025)

Circular Reasoning: Understanding Self-Reinforcing Loops in Large Reasoning Models
par: Duan, Zenghao, et autres
Publié: (2026)

SkillAttack: Automated Red Teaming of Agent Skills through Attack Path Refinement
par: Duan, Zenghao, et autres
Publié: (2026)

Latent-GRPO: Group Relative Policy Optimization for Latent Reasoning
par: Deng, Jingcheng, et autres
Publié: (2026)

Large Language Model Sourcing: A Survey
par: Pang, Liang, et autres
Publié: (2025)

MLaKE: Multilingual Knowledge Editing Benchmark for Large Language Models
par: Wei, Zihao, et autres
Publié: (2024)

Everything is Editable: Extend Knowledge Editing to Unstructured Data in Large Language Models
par: Deng, Jingcheng, et autres
Publié: (2024)

Stable Knowledge Editing in Large Language Models
par: Wei, Zihao, et autres
Publié: (2024)

Following the Autoregressive Nature of LLM Embeddings via Compression and Alignment
par: Deng, Jingcheng, et autres
Publié: (2025)

Invisible Relevance Bias: Text-Image Retrieval Models Prefer AI-Generated Images
par: Xu, Shicheng, et autres
Publié: (2023)

A Theory for Token-Level Harmonization in Retrieval-Augmented Generation
par: Xu, Shicheng, et autres
Publié: (2024)

Improving Video Corpus Moment Retrieval with Partial Relevance Enhancement
par: Hou, Danyang, et autres
Publié: (2024)

D-Models and E-Models: Diversity-Stability Trade-offs in the Sampling Behavior of Large Language Models
par: Gu, Jia, et autres
Publié: (2026)

Enhancing Training Data Attribution for Large Language Models with Fitting Error Consideration
par: Wu, Kangxi, et autres
Publié: (2024)

Do LLMs Play Dice? Exploring Probability Distribution Sampling in Large Language Models for Behavioral Simulation
par: Gu, Jia, et autres
Publié: (2024)

Think Before You Speak: Cultivating Communication Skills of Large Language Models via Inner Monologue
par: Zhou, Junkai, et autres
Publié: (2023)

Event-aware Video Corpus Moment Retrieval
par: Hou, Danyang, et autres
Publié: (2024)

RLKD: Distilling LLMs' Reasoning via Reinforcement Learning
par: Xu, Shicheng, et autres
Publié: (2025)

Detoxification for LLM: From Dataset Itself
par: Shao, Wei, et autres
Publié: (2026)

Knowledge Boundary and Persona Dynamic Shape A Better Social Media Agent
par: Zhou, Junkai, et autres
Publié: (2024)

Rowen: Adaptive Retrieval-Augmented Generation for Hallucination Mitigation in LLMs
par: Ding, Hanxing, et autres
Publié: (2024)

List-aware Reranking-Truncation Joint Model for Search and Retrieval-augmented Generation
par: Xu, Shicheng, et autres
Publié: (2024)

Qsnail: A Questionnaire Dataset for Sequential Question Generation
par: Lei, Yan, et autres
Publié: (2024)

Cross-Modal Safety Mechanism Transfer in Large Vision-Language Models
par: Xu, Shicheng, et autres
Publié: (2024)

Negative as Positive: Enhancing Out-of-distribution Generalization for Graph Contrastive Learning
par: Wang, Zixu, et autres
Publié: (2024)

Search-in-the-Chain: Interactively Enhancing Large Language Models with Search for Knowledge-intensive Tasks
par: Xu, Shicheng, et autres
Publié: (2023)

Reverse Physician-AI Relationship: Full-process Clinical Diagnosis Driven by a Large Language Model
par: Xu, Shicheng, et autres
Publié: (2025)

Cross-Model Comparative Loss for Enhancing Neuronal Utility in Language Understanding
par: Zhu, Yunchang, et autres
Publié: (2023)

BotTrans: A Multi-Source Graph Domain Adaptation Approach for Social Bot Detection
par: Shi, Boshen, et autres
Publié: (2025)

Unsupervised Information Refinement Training of Large Language Models for Retrieval-Augmented Generation
par: Xu, Shicheng, et autres
Publié: (2024)

Generative Ghost: Investigating Ranking Bias Hidden in AI-Generated Videos
par: Gao, Haowen, et autres
Publié: (2025)

Chain-of-Memory: Lightweight Memory Construction with Dynamic Evolution for LLM Agents
par: Xu, Xiucheng, et autres
Publié: (2026)

from Benign import Toxic: Jailbreaking the Language Model via Adversarial Metaphors
par: Yan, Yu, et autres
Publié: (2025)

ToolCoder: A Systematic Code-Empowered Tool Learning Framework for Large Language Models
par: Ding, Hanxing, et autres
Publié: (2025)

On the Diminishing Returns of Complex Robust RAG Training in the Era of Powerful LLMs
par: Ding, Hanxing, et autres
Publié: (2025)

Text Detoxification: Data Efficiency, Semantic Preservation and Model Generalization
par: Yu, Jing, et autres
Publié: (2025)