Conda Quick Guide

[ python  ]

We will be using Python in the Deep Learning Nanodegree, specifically the Anaconda distribution. Conda is a package management and virtual environment tool for creating and managing various Python environments.

Create a New Environment

conda create -n <env_name> python=<vNum> <pkg1> <pkg2> ... <pkgN>

NOTE: it’s a good practice to have two “default” environments on your system: one for Python 2.7.x and one for Python 3.y.z (as of now, 3.6.z). That said, any new project you begin in Python should be in Python3.

Packages

List all packages in your environment

conda list

Install new packages

conda install <pckgNm1> <pckNm2> ...

NOTE: dependencies are automatically installed.

Install specific version of new package

conda install <pckNm>=<vNum>

Uninstall a package

conda remove <pckNm>

Update/Upgrade all packages

Update and upgrade are synonyms as far as conda is concerned. The following two commands do the same thing.

conda update --all
conda upgrade --all

Search Conda’s Database for a Package

This is helpful if you’re unsure of a package name, but “kinda remember.”

conda search <search_term>

Conda Environments

Enter an environment

# OSX/Linux:
source activate <envNm>
# Windows
activate <envNm>     

Note that on Windows, this doesn’t work in PowerShell. Instead, I found that you must use old cmd prompt or the one provided by Anaconda.

Leave an environment

To “leave” an environment is to head back to default Python (called “root”).

# OSX/Linux
source deactivate
# Windows
deactivate

List All Your Anaconda Environments

Forgot what environments you have? Forgot an environment name? That’s Ok.

conda env list

Remove an environment you no longer need

conda env remove -n <envNm>

Create an Environment YAML File

This allows fo easily reusable and reproducible conda environments, e.g., for sharing code on GitHub .

conda env export > myEnvironment.yaml

It is recommend that, if sharing on GitHub, one also include the pip version of the environment as well for those not using Anaconda:

pip freeze

Create an Environment from Existing YAML Environment File

conda env create -f coolNewEnvironment.yaml
Written on February 4, 2017