Question About train.py Code

Hi,

Great work—I'm really impressed! However, I encountered some confusion while going through your code.

I noticed that you set Cfg.commands.inverse_IK_door_opening in the training code. Does this mean you need to train separate policies for different tasks? If so, can the code reproduce all the results presented in your paper?

Looking forward to your reply!

Best,
Shaofeng