Setting up Conda, TensorFlow, and PyTorch for your Deep Learning Experimentation
This post reviews how to build the experimenting environment with conda for deep learning. I compile the commands to ease the purpose of setting up the environment.
Step 1: Get miniconda installed
Conda is a package or environment management tool that eases the life of developers by eliminating the necessity of changing versions of packages for different applications. Using miniconda is recommended as we usually prefer the speed of working with the terminal over GUI. To install miniconda follow the steps here.
The primary ability in working with miniconda is to know how to make environments, activate them, and deactivate them. When environments are active, packages are installed are added to that environment.
Note: For a cleaner environment, building separate environments for TensorFlow and PyTorch is recommended.
Step 2: Get TensorFlow
First, run the following command to build the environment for TensorFlow, and then activate it.
$ conda env create --name tf # tf environment
$ conda activate tf
Then, go to the TensorFlow documentation and follow the instructions there :).
Step 3: Get PyTorch
Same for the PyTorch, create an environment, then activate it.
$ conda env create --name pt # PyTorch environment creation
$ conda activate pt_env # activated the created environment
PyTorch website eases its installation by providing the command for installing what is needed.
Now, everything is set up for experimenting with deep learning on GPUs.
References
https://docs.conda.io/projects/conda/en/latest/user-guide/install/linux.html