Skip to main content

SDK Examples

In this example, we'll run through a few basic commands using the SDK. We'll import the SDK, create a Project, and launch a Workflow.


The SDK is bundled with the Gradient CLI.


We'll be adding a Jupyter Notebook template that you can run on your own.

Getting Started

This assumes that you have the environment variable PAPERSPACE_API_KEY set to your api key. If you are using a Paperspace environment it is recommend to use a secret to pass it to your workload. If you are using an interactive environment use the gradient apiKey command to securely configure your api key.

#Import the SDK Client from the gradient package
from gradient import sdk_client
#Set some variables
project = "prkr4412y" # Set your project
cluster = "cl8pwu9qn" # Set your cluster
#Set your SDK client 
deployment_client = sdk_client.DeploymentsClient()
models_client = sdk_client.ModelsClient()
jobs_client = sdk_client.JobsClient()
projects_client = ProjectsClient()
experiment_client = sdk_client.ExperimentsClient()
workflows_client = sdk_client.WorkflowsClient()

#or access them all from a single client
#client = sdk_client.SdkClient()
projects_list = projects_client.list()
for project in projects_list:
print("project_id: " +" project_name: "+

Create project

#create a new project
project_id = projects_client.create("sample_tutorials")
print("project_id: "+proj_id)

Launch a Workflow

Grab your Workflow ID

workflows = workflows_client.list( project_id=project)

Run the Workflow

spec_path = "./sample.yaml"

yaml_spec = open(spec_path, 'r')
spec = yaml.safe_load(yaml_spec)

workflow_param = {
"workflow_id" : "08676d5c-9873-4d24-a5d4-1fcc92218379",
"spec": spec,
"cluster_id" : cluster,
"inputs": None

run_id = workflows_client.run_workflow(**workflow_param)