There has been an M1 fork of Tensorflow almost since the chip launched last year. I believe Apple did the leg work. It’s a hoop to jump through, yes, and no ones training big image models or transformers with this, but I imagine students or someone sandboxing a problem offline would benefit from the increased performance over CPU only.
I was in the hole about 3 years ago, I bought a book, some tools, and a couple of broken movements off of eBay with the thought that I’d just “ dive in and learn how they work “, but I stalled out. Without a set direction I found the space of things that could be wrong with my sample broken movement to be too large. Hell, I didn’t even know if all the components were there.
It’s something I’d like to get into, hopefully with some hand holding. I’m curious what resources you’re using, and if there’s some kind of loose curriculum to ease the beginning efforts.
https://blog.tensorflow.org/2020/11/accelerating-tensorflow-...