Lucas Colley
Lucas Colley
Might be nice to have an example using `pytorch-gpu` from conda-forge if that is supported?
> maybe good to handle this in the scikit or in a compatibility layer. It would be useful to take a look at wrapping MLX in array-api-compat (see e.g. https://github.com/data-apis/array-api-compat/pull/76...
> Exposing a fake xp.float64 symbol in a compat layer might work but feels weird. It might be a pragmatic first step into exploring what else would break in practice...
> Noting that more information can be found in the notes goes without saying for short docstrings, IMO. I agree that this looks a little unusual - we don't typically...
> Why is it unusual? I just meant that relative to the rest of the codebase currently. Points about directing to notes from the extended summary and adding a ToC...
(to clarify, I have seen this, but I would like to see @DietBru's comments addressed before merging myself - another maintainer may merge as is, though)
this looks related to https://github.com/scipy/scipy/issues/23458#issuecomment-3208010508
@HaoZeke with mamba 2.0 it seems like some of the things we need have been made public. gh-1438 reports that the current mamba plugin is incompatible with mamba 2.0. Just...
> > Error handling: Since value-based checks are no longer possible, we cannot emit certain errors. Instead, we return NaN values. > > At first glance, this is a compromise...
right, and this PR addresses "an array API agnostic variant is not feasible (or delayed to a future date)" for `spatial.transform`