Re-Align: Aligning Vision Language Models via Retrieval-Augmented Direct Preference Optimization

Re-assign community