There are four ways to add dependencies to your project:

  1. Installing packages one-by-one, for example tensorflow==2.10.0 (most common)
  2. Installing packages in bulk from a requirements.txt
  3. Importing a Github Repo in your sandbox
  4. Downloading a package in setup.sh

We’ll walk through each of these scenarios below.

​
Installing Public Python Dependencies

To install packages, click the “Add Dependency” button at the bottom of the page. You can either enter the Python package name to install, or you can upload an existing requirements.txt file:

Sandbox dependencies

Add Python Dependency

​
Installing Dependencies from Github

You may want to install dependencies from a Github repo. To do so, you can use our Github Integration to import your repo as a submodule dependency in the sandbox.

​
Connect your Github account to Slai

First, you’ll need to connect your Github to Slai. You can do that on the Integrations page.

​
Importing a repo into a sandbox

After connecting your Github account to Slai, you can import repos directly into a sandbox.

Click the + Add File icon in the files section, and select Import from Github. You can also enter the URL of any Github repo you have read access to.

Add GitHub

You’ll be prompted to select from a list of repos in your account.

List of repos

​
Using imported repos in your projects

After importing a repo, you’ll see all the files in the repo added to the file tree.

GitHub files

After importing your repo, you can reference any of the modules like any ordinary Python module.

Using GitHub files

​
Installing Custom Dependencies

You may want to use dependencies that aren’t public on pip. For this use case, we recommend using the setup.sh script. setup.sh can be customized with anything that you’d apt-get install in a Docker container.

Select the setup.sh option in the config section of the sandbox.

Select the setup.sh

Here’s an example of how you might use it:

Example

RevisionsDeployments