Alex Thewsey

The only notable conflict in this merge was that somebody had tried to introduce `kms_key` argument explicitly through `ProcessingStep`, whereas this PR was already proposing to allow `**kwargs` to pass...

IMO the Docker Desktop licensing changes of August this year (per [docker.com](https://www.docker.com/blog/updating-product-subscriptions/), [theregister](https://www.theregister.com/2021/08/31/docker_desktop_no_longer_free/)) make this request even more relevant: Not just for Linux desktop users but anybody else that may...

Thanks both for your insights! Very useful as I try to wrap my head around it all too. To @Optimox 's first point, I think that's my bad: I used...

Just to point out, the [torch-scatter](https://github.com/rusty1s/pytorch_scatter#installation) dependency introduced here is making it a bit tricky for me to reproduce the Forest Cover Type tests from #217 because: - We seem...

I agree this library has been very quick & easy to get started with for me - but I don't think expanding the API to accept either array-like or generator-like...

Anecdotally, I recently noticed a drop in accuracy (or maybe convergence speed) on Forest Cover Type when upgrading version of PyTorch... Would be interested to see whether others experience same...

@Optimox those figures were I believe all using develop code as of my recent PR #164. Taking a random 80/10/10 Training/Validation/Test split of Forest Cover Type, and just trying the...

Completely agree with you about the for loops and my concerns about that were why I measured execution time initially! Having thought about it for a while, I realised `M_x`...

OK I tried several approaches but have not been able to get #92 to install on PyTorch v1.4 environments because of the `torch-scatter` dependency it introduces. Should theoretically be possible...

Sure thing, thanks! Happy to support on porting to another branch too if you have one that's looking stable & favoured - I'd expect this change should map fairly nicely...