Antonin RAFFIN

Hello, I will try to have a look, probably related to https://github.com/DLR-RM/stable-baselines3/issues/834, see my comment for potential solutions: https://github.com/DLR-RM/stable-baselines3/issues/834#issuecomment-1077642829

> have done the math for the space taken by the batch of observations in Float32: ((101*51) + (9) + (1) +(1)) * 4bytes * 10000 = 0.2 GB of...

> I guess I could only normalize the rewards and not the observations? yes, and you can exclude specific observation key too (recommended here). could you please give the output...

> have done the math for the space taken by the batch of observations in Float32: ((101*51) + (9) + (1) +(1)) * 4bytes * 10000 = 0.2 GB of...

should be closed by #842

Hello, that would be a valuable extension to SB3 but should be done in the [RL Zoo](https://github.com/DLR-RM/rl-baselines3-zoo) I think (or in an external repo). > Here is how I would...

> Making it some contrib to RL Zoo indeed looks like the way to go, since the export can be achieve "from the outside" or SB3. Feel free to open...

Thanks for the PR =) > The export is initiated through enjoy.py, since I didn't wanted to duplicate or factor out the environment loading logic, to start the export pass...

> , dropping the dependency with CMRC or making it optional Less dependencies is usually better ;) > Enabling the Python binding is more like a test than a real...

For the reason why, you can read the migration guide: https://stable-baselines3.readthedocs.io/en/master/guide/migration.html Also, each algorithm page explains which spaces and policies are supported, for instance, for ppo: https://stable-baselines3.readthedocs.io/en/master/modules/ppo.html