Vision Model Development

• Navigate to the “Deployment Section”: In the sidebar, locate and click on the Deployment section,
  • Click on the “Create Deployment +”,

• In the “Create Repository” section, provide a unique “Deployment Name” and set its Visibility to Private to ensure restricted access.

  

  • Click “Create”, and a popup message will appear confirming the successful

  

• You will now be redirected to the “Model Selection” page. “Select Model” to confirm your choice. If you wish to choose a different model, “Clear the selection” else press “Continue” to proceed.

• Now you will land on the configure page “Configure Repository”, Click update.

• a popup message will appear confirming the success.

• On the “Deploy Model” page, configure the following options as needed:

  • GPU: Default value is 0 (selectable range: 0–100).
  • CPU: Default value is 2 (selectable range: 1–16).
  • Memory (in GB): Default value is 2 (selectable range: 1–64).
  • Minimum Replicas: Default value is 1 (selectable range: 0–2).
  • Maximum Replicas: Default value is 2 (selectable range: 2–4).
  • Target QQPS per Replica: Default value is 4 (selectable range: 1–1000).

  

• After giving the required values, you will get a toast message with the values you gave for deployment.

• Monitor the “deployment status”,

  • Wait for replicas to initialize,
  • Status will change to “Ready” when deployment is complete,
  • Click “View Code” to get integration examples.

  Integrate the model:

  • Use the provided curl commands or Python code
  • Incorporate the endpoints into your application
  • Test the integration thoroughly

Example usage on the endpoint