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

In three linesQE-Catalytic-V2 is a unified graph-text multimodal LLM for catalytic materials. It integrates property prediction and inverse design in a single shared representation space, eliminating distribution shifts between decoupled models. Demonstrates superior performance on relaxed-energy prediction and inverse design tasks.

Summary generated by Claude — human-verified