Sam Toyer

This is done for BC, albeit in the `py3.6` branch: https://colab.research.google.com/github/qxcv/magical/blob/pyglet1.5/demo-notebook.ipynb Will probably have GAIL working with image-based observations this week.

Not on master, but we do have it on the `image-env-changes` branch. See this code in another repo for a BC example: https://github.com/HumanCompatibleAI/eirli/blob/master/src/il_representations/scripts/il_train.py#L281-L313 (although be warned that it's more complex...

I haven't touched this code in a while, so I'm not sure what would have to change in the example above. That said, [this fresh PR](https://github.com/HumanCompatibleAI/imitation/pull/519/) contains a helpful [example...

If we add a wrapper that transposes the images to be NCHW then SB3 will fail, because it will transpose them to NWCH. (SB3 does have a check that sometimes...

Also, I realised it would be easier to fix this upstream (very few lines of code need to change), so I'll try that first.

If it's necessary to change the SB API, it's probably cleaner to either change `learn()` so that it properly supports this use case (of only training for a few steps...

I'm pretty sure the SB3 fork is unnecessary now. IIRC something was just getting force-wrapped with vecenvs previously, but I think it's fixed. Can take another look at this later;...

I'm in the process of (belatedly) bringing our SB3 and imitation branches even with master for the IL representations project. Here is a diff between [our SB3 branch](https://github.com/HumanCompatibleAI/stable-baselines3/tree/ilr) and the...

Hmm, looks like a similar problem to #2 (although this time it's actually the code in this repo…). Converting the `width` and `height` in `handle_conv` to `int`s should fix the...

Unfortunately I'm not maintaining this any more. I've added this message to the README: > I don't use Caffe or Keras any longer and am not planning to update this...