ASPO: Adaptive Sentence-Level Preference Optimization for Fine-Grained Multimodal Reasoning

Open in new window