Yes, I am running python 3.8.10. And for whl files, I am using MacBook Pro 16' with intel CPU, so, I do have to use the X86_64 version.

Still not work for me, that's so wired. Here is what I got, (base) hawkiyc@hawkiyc envs % source tf_metal/bin/activate (tf_metal) (base) hawkiyc@hawkiyc envs % python3 -m pip install --upgrade pip Requirement already satisfied: pip in ./tf_metal/lib/python3.9/site-packages (21.1.2) (tf_metal) (base) hawkiyc@hawkiyc envs % python3 -m pip install tensorflow-macos  ERROR: Could not find a version that satisfies the requirement tensorflow-macos (from versions: none) ERROR: No matching distribution found for tensorflow-macos Sincerely,

Thank you very much for your information. This problem resulted from Python 3.9. I succeeded in installing tf-macos and tf-metal in a python 3.8 env.

Dear Eugene, The feedback has been submitted last night, and the no. is FB9220496. I have also attached the requirements.txt file of my env and 2 py. files of NLP models in the feedback. Those files shall help you to reproduce this issue more easily. Sincerely, hawkiyc

Hi, Eugene. I made a new test, fine-tuned BERT without the TF-hub package, yet it still not work. It looks like that neither TF-hub nor TF_model_official gets the assignment of the device correctly. I also updated the feedback with the test .py file and hope this could help.

According to Apple, you shall update to Monterey.

I have tried to make a whole new backup with another HDD, but not work at all this time.

Hi  EugeneZhidkov. May I follow up on this bug? Will this issue be addressed in the new version of macOS or TF-macOS / TF-metal?

Hi, @Frameworks Engineer. I appreciate your time and looking forward to the new version of TF-metal / TF-macos.

Thank you very much for your time. I am looking forward to the new version.

Hi, @itsloudc. According to Apple, this GPU assignment issue happens while you are using TF-hub and/or huggingface for the NLP tasks and there is no known workaround right now. It looks like that we can only wait for the new version of TF-macos and TF-metal.

Hi, @SvenEck. No, the IDE kernel always crashes when I am trying to train any model.

That's weird. The previous version is fine for me. Here is my env file in the .yaml form. Maybe you can try it. https://www.icloud.com/iclouddrive/0Yhr444NAqu6oD8qxOfKriY-A#env%5Fmac%5Ftf

It's weird, I can run models with ADAM without any issue for the previous version.

Hi, @EugeneZhidkov. The latest version can't work at all. And I have already reported this issue at FB9691835. Could you please look into it and solve this issue? Thank you very much.