Tag
Jasper Research has released a comprehensive cookbook and resources for building text-to-image models from scratch, featuring detailed explanations, a 100M-image dataset, and a codebase with a tiny model.
The author shares a habit of using an agent to document all training hacks and cheat codes, including hyperparameter changes and dataset upgrades, to maintain a factual log for future reference and tutorial creation.
A project-based roadmap for learning LLM engineering by building key components from tokenizers to serving stacks, including hardware foundations and post-training techniques.