All about AI Toolkit for Visual Studio Code
summary
AI Toolkit for VS Code shows how to acquire curated small models, run them locally on a laptop GPU, evaluate changes and fine tune for specific tasks. It demos downloading Phi‑3 mini, using Azure Studio for evaluation, and local fine tuning (or Azure Container Apps in private preview). A 2,300 Q&A dataset raises category prediction from 14% to 91% accuracy and outperforms a GPT-based baseline at this task. local model download fine-tune with 2,300 Q&A 91% category accuracy
AI-written summary · about
chapters (7)
- 0:00AI Toolkit intro
- 1:34AI Toolkit interface & models
- 3:02Run Phi-3 mini locally
- 4:10Fine-tuning for predictions
- 6:41Model evaluation & off-topic handling
- 8:063 steps to use AI Toolkit
- 9:32Call to action & resources
related talks
23mFine Tune Your AI Toolbox with GitHub Copilot, Azure AI Foundry...
25mBuilding AI Agents
57mEasier AI allows everyone to build smart apps
34mBuild an AI-powered Pet Detector in Visual Studio Code
20mAgentic Excellence: Mastering AI Agent Evals w/ Azure AI Evaluation SDK
01h 10mSolutions prototyping using Azure AI Studio
33mExplore Windows AI Dev Options 🔥 AI Dev Gallery Explained (2026)
20mAI and Machine Learning for developers
