Skip to main content

Package Commands

Package management has two paths: Conda packages and PyPI packages. Prefer conda install; fall back to pip install when absent.

Install packages​

Via Conda (searches configured channels):

Linux / macOS
conda install numpy
conda install -n myenv pandas=2.2   # specific env, specific version
conda install -c conda-forge pytorch  # specific channel

Via pip inside a Conda env (after conda activate):

Linux / macOS
pip install requests

⚠️ Mixing rule: use conda install for packages with binary deps (e.g. numpy, pytorch, opencv), then pip for pure-Python ones. Avoid conda and pip cross-uninstalling the same package.

List installed packages​

Linux / macOS
conda list                 # current env
conda list -n myenv        # specific env

Update / remove​

Linux / macOS
conda update numpy         # update one
conda update --all         # update all in current env
conda remove numpy         # remove (conda-installed)
pip uninstall requests     # remove (pip-installed)
Linux / macOS
conda search "numpy>=1.26"

Clean cache​

Conda caches downloaded packages. Clean periodically:

Linux / macOS
conda clean -a -y          # clean all caches

⚠️ conda clean -a deletes all download caches; next install re-downloads.

Join Us