Adam Gleave

## Bug description Some files are failing mypy type checks (but passing on pytype) and being skipped by `EXCLUDE_MYPY` in `.circleci/config.yml` introduced in https://github.com/HumanCompatibleAI/imitation/pull/534 We should fix these errors, by...

`mdp.ValueIteration` uses `ValueIteration._boundIter` to set `self.max_iter` for discounted MDPs. This is based on the span: the difference between the minimum and maximum values after one step of the Bellman operator....

## Problem GAIL and AIRL can be slow to train due to RL in the inner loop, which is both computationally expensive and can require many environment interactions. Behavioral cloning...

## Problem `imitation`'s testing is currently limited to static analysis (type checking, linting, etc) and unit testing. There are no automated, end-to-end tests of algorithm training performance. This is problematic...

We are pinning `setuptools` and `wheel` in `ci/build_and_activate_venv.ps1` and `setuptools` in `docs/requirements.txt` / `readthedocs.yml` in order to avoid the issue described in https://github.com/freqtrade/freqtrade/issues/8376 This issue should be resolved in recent...

Once regenerated models, remove: - Wildcard import in `src/evaluating_rewards/rewards/__init__.py` - `src/evaluating_rewards/analysis/resulst.py` -- path rewriting

Interpretability in realistic environments

1

Apply sparsification techniques to higher-dimensional, more realistic environments than the simple `evaulating_rewards/PointMassLine-v0` used so far.

It currently takes around 2 seconds to load the pirl module. Most of the time this isn't a big deal, but due to a combination of: 1. We restart each...