DiffPO: Diffusion-styled Preference Optimization for Efficient Inference-Time Alignment of Large Language Models

Re-assign community