Antonin RAFFIN

Hello, No, this is not normal, it should take only some seconds (once tensorflow is loaded). as in #29 what simulator are you using? and what is your OS?

>.. just Windows 10 it has issues (tried it on two different Windows 10 computers) This was only tested on linux. I cannot give any guarantees for windows.

Hello, There is one important info that is missing: what simulator are you using? And did you launch it before launching the script?

>Using donkey sim "SDSandbox v 20.5.16". I've tried several things: This is a newer version compared to the one tested. Did you try using the one from the README?

I've got an internal pytorch version working with "SDSandbox v 20.5.16" and SB3, I planned to open source it at some point. In the meantime, you can take a look...

You can also try older binaries from the release page: https://github.com/tawnkramer/gym-donkeycar/releases September 2018 may work: https://github.com/tawnkramer/gym-donkeycar/releases/tag/v18.9 There are binaries for Linux, Mac and Windows.

Hello, The answer is the documentation of stable-baselines (cf saving/loading in stable-baselines.readthedocs.io/), you need to use a callback to make checkpoints. We have already a callback saving the best model:...

Hello, You need to use a custom version of stable-baselines for that.

Might be a duplicate of https://github.com/Stable-Baselines-Team/stable-baselines3-contrib/issues/81 or https://github.com/Stable-Baselines-Team/stable-baselines3-contrib/issues/195 Probably a combination from your env/hyperparameters. Please note that we do not offer tech support, see https://github.com/Stable-Baselines-Team/stable-baselines3-contrib/issues/81#issuecomment-1175302831

> have you been able to take a look at this yet? I am very curious what you think about it. no, not yet, still on my stack... and going...