Adam Gleave

Sounds like `v4l2loopback` is either not installed or not loaded as a kernel module. I believe this should work: ``` sudo modprobe v4l2loopback devices=1 exclusive_caps=1 ``` You can put it...

Interesting, what's the output of `v4l2-ctl --list-devices`? Do you see "Dummy video device" listed?

OK, well `platform:v4l2loopback-000` is the right thing but it seems to be called `'GoproLinux'` on your system for some reason -- I suspect there's a config file somewhere setting this?...

I don't think we have any examples beyond those in the `examples/` folder, although we'd be happy to see a PR adding these if you wanted to port qxcv's notebook.

Thanks for writing this up, Lev! # Modes We were a bit hesitant about relying on modes previously as it seemed less explicit than "call the right function" and so...

> Let's say I want to get the unshaped reward from a reward network. At the moment I have to check the net to see if it is an instance...

> The only other concrete use case I can think of at the moment would be reward ensembles. When doing training we might want to detect if we are using...

> IMO the cleanest fix to this would remove the two algorithm/policy fixes from SB3. Instead, the user would be responsible for wrapping environments in a `VecImageTranspose` wrapper if necessary....

I've added DRLHP. In general I'd like to not focus just on learning from demonstrations, given this has some severe limitations in terms of reward ambiguity.

We now have a DRLHP implementation thanks to https://github.com/HumanCompatibleAI/imitation/pull/320 Would welcome PRs on the others, too, but will be prioritizing making the existing algorithms solid and easy to use and...