Addison Klinke

I have the same issue as @Kangzf1996 and @MichaelCong - multiprocessing `ConnectionResetError` triggered by `RuntimeError: CUDAStream_setStreamOnDevice` from PyTorch. I included my full log below. My only changes to the default...

I like the idea of an `__as_config__` interface per class because it keeps the conversion code local to the rest of the source. That should make its usage more apparent...

@rsokl Another library class where I ran into issues ```python from hydra_zen import builds import torch cfg = builds(torch.optim.SGD, populate_full_signature=True) ``` This raises the unsupported primitive error for `lr=`. Is...

Based on the `torch.Generator()` example, my understanding is that `like` is restricted to patterns where instantiating the config immediately executes a linear sequence of actions. Is that correct? I have...

> The configuration for `Car` cannot be instantiated because there is no default value provided for `driver` To clarify, I understand why this fails from a technical standpoint. My question...

Another shortcoming of `Taxonomy` is that it doesn't support a `perRegion` parameter like `Choices`. That means it can only apply a global classification to an entire image or piece of...

@makseq How do you feel about re-naming this to focus on improvements to `Taxonomy` vs. creating a new `Select` control tag - perhaps that's easier to maintain? In either case,...

@PJEstrada I saw the `annotation_update` event listed in the [docs](https://diffgram.readme.io/docs/events) and am curious to understand more details about when it gets emitted. Does it include creation of new annotations and...

Would it take a lot of changes to support? Having that option is pretty critical for the labeling setup I need to create

I could give it a go if you have some pointers on the general design approach to take. I've just started exploring the project, so I don't have a natural...