Retrieve, Don't Retrain: Extending Vision Language Action Models to New Tasks at Test Time figure
AlphaXiv 中文概览(可滚动查看)