#PyTorch

Briefing: Show HN: I built a tiny LLM to demystify how language models work

Strategic angle: Built a ~9M param LLM from scratch to understand how they actually work.