Exercise: Select, deploy, and evaluate models
Unit 6 of 7 · Module 2: Select, deploy, and evaluate Microsoft Foundry models
Read this lesson
Your turn
Time to put Module 2 into practice. In this lab, you'll work in the Microsoft Foundry portal: explore the model catalog, compare models with benchmarks, test them in the playground, and evaluate one with a synthetic dataset. For the real thing, you'll need an Azure subscription with permission to create AI resources, and about forty-five minutes. Or practice every step right here first, in our simulated version.
The crossing
Here's the route. Create a project. Open the model catalog, and review the model card and benchmarks for GPT five point two. Use the leaderboard's trade-off charts, and compare GPT five point two with GPT five mini side by side. Deploy both. In the playground, give them the same riddle: a fox, a chicken, and a bag of grain that have to cross a river, and ask each one to explain its reasoning. Then evaluate GPT five point two with forty-five synthetic travel questions, including some safety tests, and analyze any failures. Finally, delete the resource group, so you don't pay for anything you're not using.
Before you start
Some of these features are in preview, so the real portal may look a little different from what you see here. Let's go.
Facts from Microsoft's documentation: Microsoft Foundry Models overview; Model benchmarks and leaderboards in Microsoft Foundry; Understanding deployment types in Microsoft Foundry Models; Run evaluations from the Microsoft Foundry portal (CC BY 4.0), rewritten in Loop's own words and examples. Narration, animation, and quiz by Loop. Microsoft, Azure, and Microsoft Foundry are trademarks of Microsoft. Loop is independent and not affiliated with or endorsed by Microsoft.