V2PE Collection Improving Multimodal Long-Context Capability of Vision-Language Models with Variable Visual Position Encoding • 3 items • Updated 5 days ago • 3