Note: Installing TensorFlow with GPU functionality requires a CUDA enabled card. Turn off the automatic activation with this command: conda config -set auto_activate_base false Delete an environment # pip cd /path/to/env_name rm -rf env_name # conda conda env remove -n env_nameĪ subset of models in the FiftyOne model zoo (image by author) It is recommended you create new environments yourself. Note: After installing anaconda, it automatically creates and activates a base environment. Create an environment # pip virtualenv env_name # conda conda create -name env_name Activate an environment # pip source /path/to/env_name/bin/activate # conda conda activate env_name Basic commands for things like creating, activating, and deleting environments are very similar between pip and conda. If you want to start using conda, follow the anaconda installation instructions in this link. Using both conda and pip together can be tricky at times, but I provide some tips to handle that later in this post. On the other hand, some packages do not exist in conda and you will have to install them through pip which is one reason that people might stray away from conda. Libraries like CUDA can be installed in your isolated environment. The primary difference between the two is that conda environments are not only for Python packages. Really just knowing that pip is the “official” Python package manager. I had previously only used pip due to a shoddy understanding of the difference between pip and conda.

Conda install pytorch