SA-VLA: State-aware tokenizer for improving Vision-Language-Action Models' performance figure
AlphaXiv 中文概览(可滚动查看)