NanoEuler: GPT-2 scale model from scratch in C/CUDA

A personal project implements a GPT-2-scale transformer entirely in C and CUDA, with no external ML libraries. The code is open-source on GitHub, aimed at understanding LLM internals.