SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics Paper โข 2506.01844 โข Published Jun 2 โข 129