Sam Levang

I guess my main point with this PR is that #8877 already introduced the majority of the "magic" to drop all indexes. If we're doing that, I don't see any...

Thanks @benbovy, this seems good, but still doesn't fix my original issue in #8809. See [comment](https://github.com/pydata/xarray/pull/8809#issuecomment-2027289650) there for more detail.

You were asked to close the issue on keras, because it is not a keras issue. This is tensorflow, where proper detection of the cuda libs is definitely still an...

Again, **you are on the tensorflow repo**. You closed the wrong issue.