Generate textbook-quality synthetic LLM pretraining data - View it on GitHub