Setting upTensorFlow on Your M1/M2 Apple Silicon MacBooks

2 min readAug 28, 2022

Setting up a virtual environment for Python and TensorFlow on a MacBook. This method worked for my M2 Mac Air with Monterey OS Version 12.5 and Rosetta 2.

TensorFlow, an extremely popular library for machine learning and AI

What you need:

Conda (Anaconda & Miniconda both suffice, see this post on which one better suits you)
Terminal

Step 1

Open Terminal and create a directory for the virtual environment

mkdir tensorflow_venv
cd tensorflow_venv

The directory will be created under /Users/YourName/tensorflow_venv

Step 2

Activate a Conda environment with a Python version compatible with TensorFlow (I was prompted that TensorFlow does not support Python 3.8, so I created a Python 3.7 environment instead)

conda create --prefix ./env python=3.7
conda activate ./env

Step 3

Now that python is installed successfully into this environment, use Conda to install TensorFlow

conda install tensorflow

The download could take a while. If you see something like this, your environment should be good to go.

Downloading and Extracting Packages
......
Preparing transaction: doneVerifying transaction: doneExecuting transaction: doneRetrieving notices: ...working... done

Step 4 (Setting up in PyCharm)

Open a project in which you want to set up TensorFlow. Go to Preferences >> Project: Your Project >> Python Interpreter.

Click on the current Python interpreter and select “Show All”

Step 5

Click on the “+” sign on the top left to add a new Python interpreter. Under “Base Interpreter”, select: /Users/YourName/tensorflow_venv/env/bin/python[some_version] Checkmark “Inherit global site-packages”, and click “Ok”.

And boom! Your Python interpreter should be correctly configured now. Run the following code to check if TensorFlow is installed.

import tensorflow as tf
print(tf.__version__)

if __name__ == '__main__':
    pass