Finetune Vision-Language-Action Model

Finetune vision-language-model Pi0.5 on real-world robot dataset. Robot Task: grasp the salt box, grasp the black bag