The TensorFlow and TFX teams have struggled with supporting builds for M1 processors, since we don’t have internal build infrastructure for M1. It’s been a problem for quite some time, and it’s growing as more and more developers get shiny new M1 laptops.
One potential way to help developers in the meantime is to give them a build procedure so that they can build the projects themselves, using Bazel. One group of developers has had some success (although not a slam dunk) with that for MLMD:
I’m wondering if anyone in this group is interested in forming a project team to develop build scripts/procedures to enable developers to build for M1. Is anyone interested?