Guiding Cross-Modal Representations with MLLM Priors via Preference Alignment

Re-assign community