preference alignment
Preference alignment is the concept of ensuring that the behavior of an AI system aligns with the preferences and values of its users or stakeholders. This is critical in applications where ethical considerations and user satisfaction are paramount.
- Align-DA: Align Score-based Atmospheric Data Assimilation with Multiple Preferences
- ComPO: Preference Alignment via Comparison Oracles
- Enhancing Diffusion-based Unrestricted Adversarial Attacks via Adversary Preferences Alignment
- Human-assisted Robotic Policy Refinement via Action Preference Optimization
- InfiFPO: Implicit Model Fusion via Preference Optimization in Large Language Models
- Leveraging robust optimization for llm alignment under distribution shifts
- Reinforcement Learning Meets Masked Generative Models: Mask-GRPO for Text-to-Image Generation
- Reward-Instruct: A Reward-Centric Approach to Fast Photo-Realistic Image Generation
- SoPo: Text-to-Motion Generation Using Semi-Online Preference Optimization
- Walking the Tightrope: Autonomous Disentangling Beneficial and Detrimental Drifts in Non-Stationary Custom-Tuning
- Weak-to-Strong Generalization under Distribution Shifts