Project Tutorial

Projects allow you to group your Training Runs and AI Experiments when using the AI Module. This tutorial will help you create your first Project and configure your environment to use it.

Creating a Project

In order to use the AI Module, you must first create a Project. You can create a new project from the Projects page, using the CLI with sigopt create project --project=my_first_project, or directly in a Python interpreter with sigopt.create_project("my_first_project"). You may need to install the latest version of SigOpt to use these features.

Configuring your Environment

After you create a Project you have a few options to configure your environment to use it. The options are listed in ascending order of precedence, i.e., sigopt.set_project will configure the Project regardless of the other options and the current directory name will always be considered last.

Option 1: Directory Name

Create a directory with the same name as the Project's ID. When you run commands in this directory SigOpt will automatically associate your Training Runs and AI Experiments with the Project that matches the directory's name.

Option 2: SIGOPT_PROJECT Environment Variable

If you have access to a terminal, set the SIGOPT_PROJECT environment variable to configure your Project.

$ export SIGOPT_PROJECT=my_first_project

Option 3: --project CLI Argument

When using the CLI, you can configure the Project for each command with the --project argument.

$ sigopt run --project=my_first_project python ./my_run.py

Option 4: sigopt.set_project Function

In your code, use the sigopt.set_project function to configure the Project for a single session. This is especially useful in an interactive Python environment like IPython or Jupyter.

sigopt.set_project("my_first_project")