Adam Gleave
Adam Gleave
> The remaining failing test is for `benchmark_and_table.sh`. It fails since it calls `bc_benchmark.sh` with the `--paper` option. It looks like it was skipped before, because it was in the...
The coverage seems fine to me -- the only check that's failing is an optional coverage check on the patch. There seems to be a bunch of indirect coverage changes...
I think you mean @AhmedARadwan?
Have you seen our README? This includes example commands that should be sufficient to replicate the paper results. You might also want to read some of the documentation for [sacred](https://github.com/idsia/sacred)...
Add support for this once https://github.com/ray-project/ray/issues/6162 is closed / https://github.com/ray-project/ray/pull/6192 is merged
The bug is still present in 0.7.5, I think we need to wait for Sacred 0.7.6 to come out.
Although Sacred 0.8.x is now out, unfortunately we still need this because of https://github.com/IDSIA/sacred/issues/499
With latest darglint release containing https://github.com/terrencepreilly/darglint/pull/160, I find that this issue is fixed with `@abstractmethod`, but not `@abc.abstractmethod`. Works, `test1.py`: ``` from abc import ABC, abstractmethod class Foo(ABC): @abstractmethod def...
Given `Self` is not supported, we should probably go with the TypeVar solution from https://peps.python.org/pep-0673/ ? WDYT, @Rocamonde ?
Can we just provide a named config to set reasonable defaults for CNN policies?