The
Daily Data

RedPajama, which creates fully open-source large language models, has released a 1.2 trillion token dataset following the LLaMA recipe.