This project provides a minimal, easy-to-understand codebase for fine-tuning Large Language Models. Our core philosophy is to explain complex optimization techniques with the simplest possible code.