Learnings
Small interactive explainers I build to understand things better — with animations and demos instead of dense equations.
🌊
Flow Matching
An interactive, beginner-friendly walkthrough of how Flow Matching turns random noise into real data — with animations, a live demo, and a full, step-by-step equation derivation.
Open → ⚖️DPO
Direct Preference Optimization for image models, explained for beginners. Follows the math step by step, from the per-image diffusion loss and shared noise through reward and margin to the loss, where it comes from, and what the gradient says about β. Every step is animated.
Open →
✨
More coming soon
New learnings get added here as I build them.