Lulu Wen

MPPO: Multi Pair-wise Preference Optimization for LLMs with Arbitrary Negative Samples

Aligning Large Language Models (LLMs) with human feedback is crucial for their development. Existing preference optimization methods such as DPO and KTO, while improved based on Reinforcement Learning from Human Feedback (RLHF), are inherently derived from PPO, requiring a reference model that adds…