| name | 09.7.verfiy-ai-models |
| description | This skill will guide you through verifying that your deployed AI models are accessible and functioning correctly. |
| argument-hint | Please follow the instructions to verify your AI model deployment. |
| compatibility | ["windows"] |
| license | MIT |
| user-invokable | true |
Verify AI Model Deployment
This skill will help you confirm that your AI models deployed to Azure are accessible and responding as expected. You'll test the model endpoints directly using a simple script.
Pre-requisites
- You have deployed AI models to Azure using Terraform as outlined in the previous skill.
- You have the Azure OpenAI API endpoint and API key for your deployed models.
- You have
curl installed on your machine for making HTTP requests.
Steps to Verify AI Models
-
Obtain API Endpoint and Key: Ensure you have the correct API endpoint URL and API key for your Azure OpenAI deployment. You can find these in the Azure portal under your Cognitive Services resource.
-
Run Verification Script: Use the provided PowerShell script to test the model endpoint. The script will prompt you for the API endpoint and key, then make a test request to the model.
Set-Location .github/skills/09.7.verfiy-ai-models/scripts
.\Verify-AI-Models.ps1
Note: Only a PowerShell script is currently available. A bash equivalent for Linux/macOS is not yet implemented.
-
Review Response: The script will output the response from the model. You should see a valid response indicating that the model is accessible and functioning correctly.
Validation
- Model endpoint returns a valid chat completion response.
- HTTP status code is
200.
- Response body contains
choices array with generated text.
Troubleshooting
