MegaTrain Full Precision Training of 100B+ Parameter LLMs on a Single GPUgithub.com1 point·adulau··0 commentsOpen articleSaveView on HN