Enhancing Generalization in Vision-Language-Action Models by Preserving Pretrained Representations

Open in new window