Things I make
on my own time.
Side projects, experiments, and things I build just for fun.
Building is more fun than ever. Some of these are live, some in progress, but all have been a blast.
An interactive tool for exploring the inner mechanics of AI models. Currently features a live visualizer for transformer internals — attention heads, MLP layers, residual streams — using the mechanistic interpretability package nnsight and GPT-2 as a reference model. More experiences and components coming.
Visit ↗︎An accessible place to start exploring AI — to start learning about what it actually is and what it can do. The questions range from explaining the latest buzzy term to digging into the technology itself.
Visit ↗︎A modern interface for consuming the latest cutting-edge research, all from arXiv.
Coming Sooncoming soon.