Today’s episode isn’t specifically about GPT-3, however. It’s about the issue of how massive these large language models are and how we can prune these models to compress them.