Creating new Conda environment messes with old environments?

Question

I have 2 different projects with different dependencies. One requires Tensorflow1.15 and another needs 1.14. I first created an environment env1 and pip installed Tf1.14, ran my code, all went well. Then I created a new environment env2 and pip installed Tf1.15, during which I could see it was uninstalling Tf1.14, I assumed it knew what it was doing. However now when I run my code in env1 it throws errors because tf1.14 is removed and env1 also tries to use tf1.15!

What am I doing wrong? I thought we use Conda to create completely separate environments for specifically this kind of situation but I'm confused.

Answer 1

You should not have used pip to install the packages. If you have Anaconda you should use conda to install them. conda install numpy will install the numpy module.

You need

conda install your-module