Will, there be any update for M1 mac. I just bought one and couldn’t make RASA work anyway.
Hi @arupkrdas76, welcome to the forum
Regarding M1 macs, there’s very little we can do – many popular libraries used by Rasa (in particular TensorFlow) don’t support those new chips yet (see 1, 2, 3, 4), and therefore Rasa can’t run on M1 either. It’s possible that TensorFlow 2.4 will partly solve this, but I think we don’t know for sure (2.4 is also pretty young and hence it’s not the version Rasa uses for now, but we’re working on it).
Saw that Apple have released this TensorFlow M1 compatible package on GitHub. Can you do something with this to make all of us with M1 processors run Rasa?
Hey Magnus, I know it’s annoying but as far as I know this is about issues with other packages that we can’t really influence. We’re aware of the TensorFlow fork you’ve linked, but it seems to still be very young and buggy (just looking at the list of fresh issues created on the repo). I think that my answer above still holds. Rasa will start using a higher version of TensorFlow at some point, which might or might not solve this.
Out of curiosity, have you been able to run something other than Rasa on your Mac, that would use at least TensorFlow and NumPy?
@SamS thanks for the answer. Just got the M1 the other day and my prio was to get Rasa running. So haven’t tried anything else yet.
Hey, is there any new update with this issue?
Hi is there any solution to this problem at all? for example does building tensorflow 2.1.0 from source work for m1? Would really appreciate if someone can provide any help
@HarmanDotpy I think this is more of a general question not specific to Rasa… Indeed, if you find somewhere some way of getting the dependencies (especially Tensorflow) installed and running, then Rasa should run okay on the machine. This being said, I personally am not fully up to date with possible workarounds for installing things like Tensorflow on M1 machines…
@SamS thank you. I kind of write the above feeling helpless. Any way thank you for your prompt response, I believe I will get some workaround soon.