In the Asteroid framework, a Project is a container for tasks, which in turn contain runs. This hierarchy helps organize and track experiments and deployments.

  • Project: A container for tasks; it could represent an AI application or service.
  • Task: Represents a specific task or experiment within a project.
  • Run: A single execution instance of a task, which can be monitored and supervised.

​
Registering a Project, Task, and Run

To register a project, task, and run, you can use the asteroid_init function provided by the asteroid-sdk package.

Here’s how you can register all three in one Python snippet:

from asteroid_sdk.wrappers.openai import asteroid_openai_client, asteroid_init
# Initialize the OpenAI client
client = OpenAI()

# Important! For this to work, Asteroid server needs to be running, contact Asteroid to get access
run_id = asteroid_init(
    project_name="Project 1",
    task_name="Task 1",
    run_name="Run 1"
)

# When you wrap the client, all supervised functions will be registered
wrapped_client = asteroid_openai_client(client, run_id)

This code snippet demonstrates how to set up the project structure in Asteroid, allowing you to track and manage your AI application’s executions.

OverviewSupervise Decorator