1

Recently, all my notebooks have been updated to python 3.10, and I can't run my old code. I was using jax 0.2.17 and numpyro 0.7.1, but these no longer work with my version, so I tried changing to python 3.9. I used the following code:

!sudo apt-get update -y

!sudo apt-get install python3.9

!sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.9 1

!sudo update-alternatives --config python3

!apt-get install python3-pip

However, after downloading this and using !pip install jax==0.2.17, when I check the jax version using jax.__version__, I get 0.4.8, the version of jax already installed in Google Colab. Are there any ways to fix this? I want to be able to use the old versions of jax and numpyro, and pip installs them, but when I import I appear to be importing from the 3.10 version instead. Thanks in advance!

Amir reza Riahi
  • 1,540
  • 2
  • 8
  • 34
Cristian Calin
  • 11
  • 2

1 Answers1

1

Colab runtimes only support a single Python version, and in general it's not possible to install a different one. After Colab does a Python version upgrade, they usually provide a fallback for a short period, and currently a Python 3.9 fallback will be available for the next few weeks (see Colab Updated to Python 3.10 for details).

After a few weeks, however, this fallback will no longer be available and it will not be possible to use Python 3.9 with Colab. Given that, I'd suggest upgrading your code to be compatible with a more recent package releases (I believe JAX v0.3.11 or newer supports Python 3.10). If or some reason that is not possible, then you might think about running your code in a local Jupyter notebook runtime where you have more fine-grained control over the compute environment.

jakevdp
  • 77,104
  • 11
  • 125
  • 160