arXiv preprint / 令牌优化

Focusing on What Matters: Object-Agent-centric Tokenization for Vision Language Action models

作者：Rokas Bendikas, Daniel Dijkman, Markus Peschl, Sanjay Haresh, Pietro Mazzaglia · 单位：Centre for Artificial Intelligence, UCL, Qualcomm AI Research · 会议/期刊：arXiv preprint · 方向：令牌优化 · 发布日期：2025-09-28

视觉语言动作令牌优化模型压缩
新标签打开 AlphaXiv
Focusing on What Matters: Object-Agent-centric Tokenization for Vision Language Action models figure
AlphaXiv 中文概览（可滚动查看）