KV-Efficient VLA: A Method to Speed up Vision Language Models with RNN-Gated Chunked KV Cache
返回详情
VLA / Vision-Language-Action 每日论文卡