Tiny LLM Learning Framework

A simplified, educational LLM framework (~9M parameters) that trains quickly on free infrastructure to teach how language models work. Target users are AI students, educators, and hobbyists who want hands-on understanding without expensive hardware.

The explosion of AI curricula across universities and bootcamps has created genuine demand for hands-on LLM education tools that don't require A100 access, and Andrej Karpathy's nanoGPT has already validated that developers will engage deeply with stripped-down, educational model implementations. The closest substitute is exactly that — nanoGPT — which is free, well-documented, and already beloved by the community, making differentiation genuinely hard unless the framework offers structured curriculum, hosted environments, or interactive tooling that a GitHub repo doesn't. Revenue is the core unsolved problem here: students and hobbyists are notoriously resistant to paying, educators buy through institutional procurement cycles that take months, and "unknown" revenue band reflects that no one has clearly monetized this category yet. The most likely failure mode is building something technically competent that ends up as a free resource people star on GitHub but never pay for, leaving the builder with no sustainable business model.