Christopher Dupuis

As it so happens, there is a bug with `xr.DataSet.expand_dims`. You can see the bug report here: https://github.com/pydata/xarray/issues/7456

It seems the xarray documentation is partially at fault, at least, according to them. Apparently, there is no combination of `xr.DataSet.expand_dims` and `xr.DataSet.transpose` that will put the sample dimension of...

TBH I actually blame Keras/PyTorch for caring about axis order. So passé!

Thinking about this some more, the current behavior does make sense if we're not considering an ML context. Like, if you wanted to sample a bunch of patches and average...

pre-commit.ci autofix

pre-commit.ci autofix

/run eddy/low_res.zarr

Hey @cisaacstern, the issue blocking this was resolved a couple of weeks ago, with no changes needed on our end, as I understand it. Do you think we can get...

Hi @aaronspring, is this something you're already working on, or just proposing? I'm looking for recipe projects to join, and I have some experience working with Koppen climate zones, as...

Actually, if we go with the proposal of default `False` and rely on `xr.DataArray.squeeze`, this option becomes completely unnecessary.