comparisons

Two implementations, one spec tree

A specification is only half the idea. The other half is grading something real against it — and grading two things against the same tree is what makes them comparable.

NanoChat: Python vs .NET A real port measured against its reference — Karpathy's nanochat versus the .NET implementation chasing it. 72% parity · 8 missing · 7 partial NanoChat.NET: does the interface expose the engine? The same product graded against itself — one implementation's UI measured against its own engine's capabilities. 69% parity · 4 missing · 1 partial