Multimodal Instruction Tuning with Hybrid State Space Models

Jianing Zhou, Han Li, Shuai Zhang +5

Handling lengthy context is crucial for enhancing the recognition and understanding capabilities of multimodal large language models (MLLMs) in applications such as processing high…