20+ high-performance LLMs with recipes to pretrain, finetune and deploy at scale. - View it on GitHub