How to install TensorFlow with Python 3.8

Question

Whenever I try to install TensorFlow with pip on Python 3.8, I get the error that TensorFlow is not found. I have realized later on that it is not supported by Python 3.8.

Answer 1

I tried this in Anaconda and faced problems when installing Tensorflow 1.14.0.

Steps to Resolve

1. Create a new environment

2. Check the python version

   conda search python
   

I found version 3.8.0. Tensorflow 1.14 is not supported with this version.

3. Next step is to downgrade python in new environment to 3.7.0

   conda install python=3.7.0

Once this is done, install TensorFlow 1.14.

Answer 2

Tensorflow does not support Python 3.8 at the moment. The latest supported Python version is 3.7. A solution is to install Python 3.7, this will not affect your codes since Python 3.7 and 3.8 are very similar. Right now Python 3.7 is supported by more frameworks like TensorFlow. Soon Python 3.8 will have more supported frameworks, and that´s when you can install TensorFlow for Python 3.8.

Answer 3

As of May 7, 2020, according to Tensorflow's Installation page with pip, Python 3.8 is now supported. Python 3.8 support requires TensorFlow 2.2 or later.

You should be able to install it normally via pip.

---

Prior to May 2020:

As you mentioned, it is currently not supported by Python 3.8, but is by Python 3.7.

You want to have virtualenv installed. You also need Python 3.7. Then you can just start a virtualenv with -p python3.7 and install it using pip like you did before:

virtualenv --system-site-packages -p python3.7 DEST_DIR

source ./DEST_DIR/bin/activate

pip install --upgrade pip

pip install --upgrade tensorflow