Cover Image

MPO: Multilingual Safety Alignment via Reward Gap Optimization

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Zhao, Weixiang, Hu, Yulin, Deng, Yang, Wu, Tongtong, Zhang, Wenxuan, Guo, Jiahe, Zhang, An, Zhao, Yanyan, Qin, Bing, Chua, Tat-Seng, Liu, Ting
Format:	Preprint
Published:	2025
Subjects:	Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Towards Comprehensive Post Safety Alignment of Large Language Models via Safety Patching
by: Zhao, Weixiang, et al.
Published: (2024)

When Less Language is More: Language-Reasoning Disentanglement Makes LLMs Better Multilingual Reasoners
by: Zhao, Weixiang, et al.
Published: (2025)

Beware of Your Po! Measuring and Mitigating AI Safety Risks in Role-Play Fine-Tuning of LLMs
by: Zhao, Weixiang, et al.
Published: (2025)

Lens: Rethinking Multilingual Enhancement for Large Language Models
by: Zhao, Weixiang, et al.
Published: (2024)

AdaSteer: Your Aligned LLM is Inherently an Adaptive Jailbreak Defender
by: Zhao, Weixiang, et al.
Published: (2025)

Exploring and Exploiting the Inherent Efficiency within Large Reasoning Models for Self-Guided Efficiency Enhancement
by: Zhao, Weixiang, et al.
Published: (2025)

TEA-Bench: A Systematic Benchmarking of Tool-enhanced Emotional Support Dialogue Agent
by: Sui, Xingyu, et al.
Published: (2026)

Teaching Language Models to Evolve with Users: Dynamic Profile Modeling for Personalized Alignment
by: Zhao, Weixiang, et al.
Published: (2025)

Chain of Strategy Optimization Makes Large Language Models Better Emotional Supporter
by: Zhao, Weixiang, et al.
Published: (2025)

OP-Bench: Benchmarking Over-Personalization for Memory-Augmented Personalized Conversational Agents
by: Hu, Yulin, et al.
Published: (2026)

When Personalization Legitimizes Risks: Uncovering Safety Vulnerabilities in Personalized Dialogue Agents
by: Guo, Jiahe, et al.
Published: (2026)

Understanding Multilingualism in Mixture-of-Experts LLMs: Routing Mechanism, Expert Specialization, and Layerwise Steering
by: Chen, Yuxin, et al.
Published: (2026)

SafeNeuron: Neuron-Level Safety Alignment for Large Language Models
by: Wang, Zhaoxin, et al.
Published: (2026)

Safety Geometry Collapse in Multimodal LLMs and Adaptive Drift Correction
by: Guo, Jiahe, et al.
Published: (2026)

Trade-offs in Large Reasoning Models: An Empirical Analysis of Deliberative and Adaptive Reasoning over Foundational Capabilities
by: Zhao, Weixiang, et al.
Published: (2025)

Who Transfers Safety? Identifying and Targeting Cross-Lingual Shared Safety Neurons
by: Zhang, Xianhui, et al.
Published: (2026)

On Safety Risks in Experience-Driven Self-Evolving Agents
by: Zhao, Weixiang, et al.
Published: (2026)

STAR-S: Improving Safety Alignment through Self-Taught Reasoning on Safety Rules
by: Wu, Di, et al.
Published: (2026)

Lingua-SafetyBench: A Benchmark for Safety Evaluation of Multilingual Vision-Language Models
by: Shi, Enyi, et al.
Published: (2026)

Large Language Model Agents Are Not Always Faithful Self-Evolvers
by: Zhao, Weixiang, et al.
Published: (2026)

Plug-and-Play Policy Planner for Large Language Model Powered Dialogue Agents
by: Deng, Yang, et al.
Published: (2023)

ENPMR-Bench: Benchmarking Proactive Memory Retrieval for Emotional Support Agents
by: Fu, Xing, et al.
Published: (2026)

Both Matter: Enhancing the Emotional Intelligence of Large Language Models without Compromising the General Intelligence
by: Zhao, Weixiang, et al.
Published: (2024)

SAPT: A Shared Attention Framework for Parameter-Efficient Continual Learning of Large Language Models
by: Zhao, Weixiang, et al.
Published: (2024)

Towards Goal-oriented Intelligent Tutoring Systems in Online Education
by: Deng, Yang, et al.
Published: (2023)

On Reasoning Strength Planning in Large Reasoning Models
by: Sheng, Leheng, et al.
Published: (2025)

Separate the Wheat from the Chaff: A Post-Hoc Approach to Safety Re-Alignment for Fine-Tuned Language Models
by: Wu, Di, et al.
Published: (2024)

On the Multi-turn Instruction Following for Conversational Web Agents
by: Deng, Yang, et al.
Published: (2024)

RSafe: Incentivizing proactive reasoning to build robust and adaptive LLM safeguards
by: Zheng, Jingnan, et al.
Published: (2025)

Aligning Large Language Models for Faithful Integrity Against Opposing Argument
by: Zhao, Yong, et al.
Published: (2025)

ConflictBench: Evaluating Human-AI Conflict via Interactive and Visually Grounded Environments
by: Zhao, Weixiang, et al.
Published: (2026)

MPO: An Efficient Post-Processing Framework for Mixing Diverse Preference Alignment
by: Wang, Tianze, et al.
Published: (2025)

Bridging Jensen Gap for Max-Min Group Fairness Optimization in Recommendation
by: Xu, Chen, et al.
Published: (2025)

ALI-Agent: Assessing LLMs' Alignment with Human Values via Agent-based Evaluation
by: Zheng, Jingnan, et al.
Published: (2024)

Compose Your Aesthetics: Empowering Text-to-Image Models with the Principles of Art
by: Jin, Zhe, et al.
Published: (2025)

Fine-Grained Verifiers: Preference Modeling as Next-token Prediction in Vision-Language Alignment
by: Cui, Chenhang, et al.
Published: (2024)

Reinforced Strategy Optimization for Conversational Recommender Systems via Network-of-Experts
by: Zhao, Xiaoyan, et al.
Published: (2025)

Walk the Talk: Bridging the Reasoning-Action Gap for Thinking with Images via Multimodal Agentic Policy Optimization
by: Yang, Wenhao, et al.
Published: (2026)

Enhancing Video-Language Representations with Structural Spatio-Temporal Alignment
by: Fei, Hao, et al.
Published: (2024)

Rethinking Experience Utilization in Self-Evolving Language Model Agents
by: Zhao, Weixiang, et al.
Published: (2026)