Adam Gleave

Assigning to @samuelarnesen to get PR https://github.com/HumanCompatibleAI/imitation/pull/524 over finish line once he starts

> We should probably have an ingredient for each algorithm. This makes constructing more complex experiments (such as warm-starting) easier. Note `sacred.Experiment` is-a `sacred.Ingredient` (parent class), so I think we...

> E.g. right now we use the `train_imitation` to train `BC` as well as `DAgger`. If I am only interested in training `BC`, then I don't want to be bothered...

Thanks for clarifying @thequilo -- good to know someone else in your lab is likely to pick it up. And good luck wrapping up your PhD!

I'll try to look at this next week. @qxcv any thoughts on above, I think you've used Sacred a fair bit?

I'm in favor of trying to port one script to Hydra to try it out. Eliminating `parallel.py` would be nice, I wrote that as a temporary hacky solution and somehow...

@ernestum `benchmarking/util.py` cleans up the configs generated by automatic hyperparameter tuning; the output of this is the JSON config files in `benchmarking/`. My understanding of the outstanding issues are: -...

> Is there interest in this or is that a lower priority thing? Cleaning up and documenting `utils.py` seems worthwhile. Documenting how to generate the summary table also worthwhile. Although...

Not sure how I missed this the first time around. I agree we should standardize, but some care needed here, as each of these names means a different thing or...

> So I propose: > > 1. Decide if we want to call it **episodes**, **trajectories** or **rollouts**? Maybe look at the naming conventions of SB3. OTOH I'd vote for...