John Lawson

@lu4 I have absolutely no idea if this will work, but when you create a tensorflow session try setting the SYCL device count in the session config options: ```python import...

This is a great idea. It looks like other actions like setup-node do something like this too. I'm not sure when I'll have time to look at it, but it...

That's interesting. It seems the node `semver` module compresses all versions down to the three main versions. In this case it means `2.8.12`, `2.8.12.1` and `2.8.12.2` all get converted to...

Yeah it would definitely be faster to guess the version info when the version is fully specified rather than pulling all versions from github. That probably requires a fair bit...

I guess that this is down to the two Mac packages: * cmake-3.19.4-macos-universal.dmg * cmake-3.19.4-macos10.10-universal.dmg https://github.com/Kitware/CMake/releases/tag/v3.19.4 Looks like this comes from https://github.com/Kitware/CMake/commit/6410425e5b036dc643ea40343899c04cf8b2fc85 so both packages should work with the MacOS...

Yeah I guess a warning that happens all the time isn't super useful. #56 should resolve this, available in release v1.14.0

Hmm yeah I can see how that might be helpful. Unfortunately this action is currently fairly tightly coupled to pulling the releases from github rather than other sources so I...

The `ScatterAddNdim` op is an inplace version of `ScatterNdAdd` from Tensorflow core. There is no GPU version of the `ScatterAddNdim` kernel, and I'm not too sure what is being run...