Developer builds GGUF inferencer from scratch in R

A developer wrote a GGUF model inferencer from scratch in R, achieving 60 seconds per token. The project is described as impractical but aimed at learning transformer architecture.