Rethinking Visual Layer Selection in Multimodal LLMs

VLM
Re-assign community