AI Toolkit and GitHub Copilot: Model Recommendations Workshop
April from Microsoft Developer walks through using Copilot and the AI Toolkit for model recommendations, deployments, and testing within the Microsoft Foundry platform in this practical workshop.
AI Toolkit and GitHub Copilot: Model Recommendations Workshop
Presenter
April (Microsoft Developer)
Overview
This video is part two of the AI Toolkit + GitHub Copilot Pet Planner workshop series. April demonstrates how developers can leverage Copilot in Agent mode to get intelligent model recommendations, use the AI Toolkit to deploy models, and effectively compare model performance for AI-powered agents.
Key Topics Covered
- Agent Mode in Copilot: April shows how to interact with Copilot in Agent mode to receive recommendations for optimal language models based on scenario requirements.
- Using the Model Catalog: Demonstrates deploying models directly from the AI Toolkit's Model Catalog.
- Model Playground: Guidance on testing and comparing the effectiveness of deployed models in a sandbox environment.
- Deleting Model Deployments: Best practices for managing model resources, including removal of unused deployments.
- Workshop Resources:
Chapter Markers
- 00:00 – Introduction
- 00:03 – Ask Copilot for a Model Recommendation
- 03:18 – Deploy Models from the Model Catalog
- 05:08 – Test and Compare Models in the Model Playground
- 12:25 – Delete a Model Deployment
Getting Started
- Install the AI Toolkit.
- Set up your project on Microsoft Foundry.
- Follow the workshop repo instructions to begin experimenting with model deployments and recommendations.
Practical Takeaways
- Hands-on demonstration of Copilot's agent capabilities for model selection.
- Practical deployment and lifecycle management tips for AI models on Azure.
- Resource links for continued learning and community engagement.
For a deeper dive, access the referenced resources to continue exploring agent model recommendation workflows in Microsoft’s AI ecosystem.