LLM_from_scratch. Building LLMs from scratch following the book from S. Raschka
JetBlock-Attention. Implementing the novel attention block design from the Nvidia's Jet-Nemotron paper
