Title: ML Concept Visualizer is live!
Body: Just wrapped up building ML Concept Visualizer for OpenAI Build Week — an AI pipeline that turns machine learning concepts, or your own trained models, into fully interactive, explained visualizations.
What's working:
- A real logistic regression classifier, visualized directly from its own prediction data — decision boundary, real data points, and an auto-generated explanation panel
- A gradient descent simulator showing convergence, local minima, and saddle points in real time
- An overfitting vs. underfitting demo with a live-adjustable polynomial fit
How it's built: GPT-5.6 designs the visualization spec (what to show, what the math is, what the explanation should say), and Codex writes the actual working HTML/JS module from that spec — including catching and fixing its own errors along the way.
Try it live:
Code's all on GitHub — feedback welcome!
Log in or sign up for Devpost to join the conversation.