Author: Visual Studio Code - Bewertung: 0x - Views:4
Shipping the right AI model for each task requires a lot of testing and evaluation. Get an inside look at how the VS Code and Copilot teams assess model quality, decide when to roll out updates, and balance capability with reliability.
To learn more, please check out these resources:
* https://aka.ms/VSCode/GHRepo
* https://aka.ms/VSCode/DBview
* https://aka.ms/VSCode/HarnessBlog
𝗦𝗽𝗲𝗮𝗸𝗲𝗿𝘀:
* Julia Kasper
* Seth Juarez
𝗦𝗲𝘀𝘀𝗶𝗼𝗻 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻:
This is one of many sessions from the Microsoft Build 2026 event. View even more sessions on-demand and learn about Microsoft Build at https://build.microsoft.com
LIVE161 | English (US)
Broadcast Stage
#MSBuild
Chapters:
0:00 - Introduction at Microsoft Build with Julia from VS Code team
00:00:22 - Julia shares background as VS Code Product Manager
00:00:35 - Discussion on generative coding era and model selection complexity
00:01:47 - Explaining differences between model families and personalities
00:03:00 - Introduction to the concept of AI harness and its role
00:04:39 - Demonstration of DJ Julia project and harness debugging tools
00:07:15 - Comparison between different models' behavior using same prompts
00:09:23 - Importance of collaborative model optimization and prompt tuning
00:11:00 - Process of model evaluation, benchmarks, and iteration
00:15:21 - Closing thoughts: continuous optimization and feedback in AI model deployment