TY  - RPRT
TI  - Token Expand-Merge: Training-Free Token Compression for Vision-Language-Action Models
AU  - Yifan Ye
AU  - Jiaqi Ma
AU  - Jun Cen
AU  - Zhihe Lu
PY  - 2025
UR  - https://arxiv.org/abs/2512.09927
ID  - 2512.09927
ER  -