Denys Makoviichuk
Denys Makoviichuk
Main issue that IsaacGym doesn't support latest versions of the python
yes it is a critic network but with additional inputs. For example policy can have only a few parameters as obs but critic can have the whole world state because...
Hi, I experimented with it here https://github.com/Denys88/rl_games/blob/cba782ceb772795628e52a3da3d5dc8c20ecb779/rl_games/algos_torch/network_builder.py#L171 yes it is not in config. I tested TwoHot encoding. Feel free to try more options. I can add it to the yaml...
Awesome! I don't have much free time unfortunately :( Thank you.
Do you aware of any reference implementations? There are couple of ways how it can be done. Problem tht in PPO I am reusing old hidden state from previous step...
@sashwat-mahalingam could you try a regular ant. Does it work?
@sashwat-mahalingam this one could be related to the way how I do reporting. When you restart a couple of first total reward reports would be reports from the failed ants...
I think its fine. all other lines a related to the case where we need norm.
Thanks! But I am surprised now why it works :)
@xuanyaoming I've got what happened. Btw in the most IsaacGym configs there are like 256-128-64 MLP without normalization. need_norm is set to true by default and we set it to...