Description
First of all, thank you for the incredible work you're doing—it's really exciting to see such groundbreaking developments in this area!
I've successfully integrated OpenVLA with my robot arm, which was a great first step. As expected, the actions aren’t yet working as intended, so I’d now like to fine-tune the model to perform simple pick tasks on my robotic system.
However, I’m fairly new to the field and would really appreciate some guidance on the following points:
-How should I record and save demonstration data correctly for fine-tuning?
-How do I perform the fine-tuning?
-Once I’ve fine-tuned the model, how do I properly run inference with the updated weights?
Any advice, best practices, or examples you can share would be extremely helpful.
Thanks again for your work and support!
Best regards,
Pete