CatalyticMLLM: A Graph-Text Multimodal Large Language Model for Catalytic Materials 事件

CatalyticMLLM: A Graph-Text Multimodal Large Language Model for Catalytic Materials arXiv:2605.17254v3 Announce Type: replace Abstract: Property prediction and inverse structural design of catalytic materials are typically modeled as two independent tasks: the former predicts target properties from given structures, whereas the latter generates candidate structures according to desired properties. Although the decoupled paradigm facilitates the implementation of a ``generation--evaluation--scre