make patch_mac_m1 fails

[rjalexa]

All previous steps work perfectly but then:

(env-tf-poet-3.10) bob@Roberts-Mac-mini py-poetry-m1 % make patch_mac_m1
conda install -y -c apple tensorflow-deps
Collecting package metadata (current_repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Collecting package metadata (repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.

PackagesNotFoundError: The following packages are not available from current channels:

  - tensorflow-deps

Current channels:

  - https://conda.anaconda.org/apple/osx-64
  - https://conda.anaconda.org/apple/noarch
  - https://conda.anaconda.org/conda-forge/osx-64
  - https://conda.anaconda.org/conda-forge/noarch
  - https://repo.anaconda.com/pkgs/main/osx-64
  - https://repo.anaconda.com/pkgs/main/noarch
  - https://repo.anaconda.com/pkgs/r/osx-64
  - https://repo.anaconda.com/pkgs/r/noarch

Environment:

(env-tf-poet-3.10) bob@Roberts-Mac-mini py-poetry-m1 % conda info

     active environment : env-tf-poet-3.10
    active env location : /Users/bob/opt/miniconda3/envs/env-tf-poet-3.10
            shell level : 2
       user config file : /Users/bob/.condarc
 populated config files : /Users/bob/.condarc
          conda version : 22.11.1
    conda-build version : not installed
         python version : 3.8.15.final.0
       virtual packages : __archspec=1=x86_64
                          __osx=10.16=0
                          __unix=0=0
       base environment : /Users/bob/opt/miniconda3  (writable)
      conda av data dir : /Users/bob/opt/miniconda3/etc/conda
  conda av metadata url : None
           channel URLs : https://conda.anaconda.org/conda-forge/osx-64
                          https://conda.anaconda.org/conda-forge/noarch
                          https://repo.anaconda.com/pkgs/main/osx-64
                          https://repo.anaconda.com/pkgs/main/noarch
                          https://repo.anaconda.com/pkgs/r/osx-64
                          https://repo.anaconda.com/pkgs/r/noarch
          package cache : /Users/bob/opt/miniconda3/pkgs
                          /Users/bob/.conda/pkgs
       envs directories : /Users/bob/opt/miniconda3/envs
                          /Users/bob/.conda/envs
               platform : osx-64
             user-agent : conda/22.11.1 requests/2.28.2 CPython/3.8.15 Darwin/22.3.0 OSX/10.16
                UID:GID : 501:20
             netrc file : None
           offline mode : False

Apple Mini M1 - Ventura 13.2.1

Maybe my problem is related to conda using osx-64 and not arm

May I also ask why is it you need to use conda and cannot do a poetry only install?

Thank you very much for this very interesting repo.

[dstalzjohn]

Thank you for your question.
I'm using conda because some packages required to be precompiled for the ARM architecture.
I followed a tutorial like here.

I think you are right with the assumption conda is using the wrong channel (osx-64). Maybe there something wrong with your installation of conda?

[rjalexa]

Might well be, will investigate.

However because I could not install with conda I switched to poetry (which as you know uses PyPI) and it works like a charm:

poetry new m1tfgpu
cd m1tfgpu
poetry add tensorflow-macos==2.9.2
poetry add tensorflow-metal==0.5.0
poetry shell

and then I ran a training benchmark which confirms the M1 GPU acceleration.

Take care.

[dstalzjohn]

Very interesting. Maybe that changed since I tried it maybe a year ago.
I will have a look at it.