1 paper
Yijie Zheng, Bangjun Xiao, Lei Shi +7
Multimodal large language models (MLLMs), such as GPT-4o, are garnering significant attention. During the exploration of MLLM training, we identified Modality Composition Incoheren…