Can you please add some sim eval code just as "gym_aloha"?
I have no idea what the batch data structure is like. Can you please write a example for detail? Thank you!
Adapt this tutorial (https://github.com/huggingface/lerobot/blob/main/examples/2_evaluate_pretrained_policy.py) by replacing the environment from gym_pusht to gym_aloha and replacing the diffusion policy by pi0-fast
For the batch data structure LeRobot basically has no documentation (as far as I know) except the tutorials. Luckily the code is still fairly small and simple so you can just read it. Look at the policy implementation and look at what features it tries to read from the input dictionary, you will see it uses standard names to get the different kind of inputs: https://github.com/huggingface/lerobot/blob/main/lerobot/common/constants.py