InstructG2I: Synthesizing Images from Multimodal Attributed Graphs | Read Paper on Bytez