Sam Toyer
Sam Toyer
Below are some algorithms that it would be nice to see in `imitation`, but which aren't urgently needed. Feel free to extend this list. Learning from demonstrations: - [ ]...
At the moment, all of our test environments have tabular observation spaces. It would be nice to include an example with a more complex observation space, like stacked images in...
This PR adds some features necessary to get clean support for image-based environments: - Switched to a (temporary) fork of SB3 that removes transpose magic. This makes it simpler for...
At the moment, GAIL and BC don't interoperate well with SB3 in environments with image-based observation spaces. The main problem is the channels axis: many environments put channels last, but...
The following script should time out after 1 second and raise a `pb.ProcessTimedOut` exception, as the `timeout` arg is set to 1s. Instead, it executes to completion, and prints 0....
I have a list of dicts annotated with as `List[T]` where `T` is a `TypedDict`. I've found that pytype complains when it has more than 63 elements. Here's a working...
GitHub was interpreting stuff between the \$ signs as math
This cleans up some of the existing tooling and adds some more. Specifically: - Removes `requirements.txt` and adds all deps to `setup.py` instead. - Adds `run_rep_learner`, `il_train` and `il_test` commands...
Currently we get a lot of errors like this in prod: ``` Unclosed client session client_session: ``` I believe these are coming from OpenAI. My reason for believing this is...
Currently we moderate the defense + access code together when saving a defense. However, when someone attacks, we moderate the attack in isolation. We should model the defense and the...