What is the best practice for 1. installing packages & 2. returning the system to having lambda stack properly working?

Hello, I want to have tree so I can use it at the command line. Should I use apt? apt-get? snap? What is the best practice for keeping one’s tensorbook in ship shape?

Along the same lines, how can I return my tensorbook to having a properly working lambda stack, and then what is the best practice after I’ve attained that for maintaining it in said condition? I have recently finally upgraded to Ubuntu 20.04 successfully (I bought this tensorbook 4 years ago but haven’t used it much, but I’m planning to start using it a lot), but I’m pretty sure I have messed up some elements of Lambda Stack & I want to 1. fix it, and then 2. do things right this time so I don’t mess it up again.

The biggest issue was related to a major upgrade from Ubuntu 18.04 to 20.04 and some packages removed. Also over time people change what they have also installed, like
installing from Ubuntu versions from lambda stack.

Or another is having anaconda or pip software installs, which conflict at time with versions or do not have complete or have out dated installs.

You can check what you have installed with pip (whether in anaconda or out of anaconda with):

$ pip -v list

I use to find what is in the local directly:

$ pip -v list | grep local 

And to clean up either use ‘pip uninstall ’ or if it easier:

$ mv ~/.local ~/.local.bak
  • This will move it to a backup file in case you really want to see what was there.

For testing the system versus in anaconda also:

(anaconda) $ conda deactivate
$ python -c ‘import torch; print(torch.version,torch.file)’
1.11.0 /usr/lib/python3/dist-packages/torch/init.py