Installing new librariesΒΆ

SherlockML servers come with over 100 of the most useful Python data science libraries pre-installed, including NumPy, pandas, matplotlib, scikit-learn and TensorFlow. To see all pre-installed packages open a terminal and type:

$ conda list

You can use conda or pip to install new libraries in SherlockML. The easiest way of doing so is through a terminal. To open a terminal, click NEW and then Terminal in the Workspace tab in your project.


Within SherlockML you can launch notebooks in either a Python 2 or Python 3 environment, so the first thing needed is to activate the environment you will use so that the libraries are available from the notebook. This is done with the source activate command. For the Python2 environment you would do:

$ source activate Python2

Once you have the terminal open and the environment activated, find out if the package exists in Anaconda by looking through this list. Once you have the package name, type in the terminal:

$ conda install <package-name>

If the package does not exist in Anaconda, try and install it with pip:

$ pip search scikit-image

Once you have found it, install it with:

$ pip install -U scikit-image

You can now import the newly installed package from you Jupyter notebook. If a package can be installed with both conda and pip, we recommend using conda.