To get started with loading and running OpenVLA models for inference, we provide a lightweight interface that leverages HuggingFace transformers AutoClasses, with minimal dependencies. For example, to ...