Bo XU

While Vision-Language Models (VLMs) have demonstrated significant potential in chemical visual understanding, current models are predominantly optimized for direct visual question-answering tasks. This paradigm often results in “black-box” systems that fail to utilize the inherent capability of L...