Antonin RAFFIN

> @araffin has anything changed with regards to SB3 supporting the MPI or its still not supported? It is not (https://github.com/Stable-Baselines-Team/stable-baselines3-contrib/issues/11, https://github.com/Stable-Baselines-Team/stable-baselines3-contrib/issues/45), but contribution is welcomed ;) But with SB3,...

Hello, I think @AdamGleave tackled that problem in the [Adversarial policies](https://github.com/HumanCompatibleAI/adversarial-policies) repo, you should take a look ;)

I would rather keep a separate branch for minimal tf2 support, as this requires tf>=1.15. Regarding tf-contrib, all the issues are: https://github.com/Stable-Baselines-Team/stable-baselines/search?q=contrib I'm also afraid of breaking previously saved models....

> but we could inform users somehow that they should install the branch version yes, in the doc and readme. For the `setup.py`, there is already a version limit there....

> any progress on this? this should answer your question: https://github.com/hill-a/stable-baselines/issues/366 SB3 repo (pytorch): https://github.com/DLR-RM/stable-baselines3 SBX (jax, experimental): https://github.com/araffin/sbx SB2 tf2 (unnofficial): https://github.com/sophiagu/stable-baselines-tf2 SB tf2 (experimental, archive): https://github.com/Stable-Baselines-Team/stable-baselines-tf2/

Did you try putting your code in a `if __name__ == "__main__":` section (cf doc https://stable-baselines.readthedocs.io/en/master/guide/vec_envs.html) This is required to use multiprocessing on windows.

>Therefore my question is, is there any capability to achieve this within stable baselines? If not, how difficult would it be to modify stable baselines to work this way? As...

>Is there any example of the n_episodes_rollout parameter in SB3? I searched and have found it in the documentation but it is a bit too terse for me to understand...

>Can you confirm and check this? In short: yes it is a bug but luckily `target_update_interval=1`, so it won't change any results :sweat_smile:. Looking at the official implementation, there is...

Thanks @hartikainen for your answer. > to apply x/y gradient steps after each collected sample or Nx gradient steps after every Ny samples yes, I have also experienced that. My...