I think that there are many people who are surprised that Large Language Models (LLMs) that predict the next word act intelligently and are capable of solving many challenging tasks. In my opinion, the success of LLMs is not surprising. In this blog post, I will explain why I think LLM’s success was obvious. So, around two years ago, during a discussion with some colleges, I coined the term “GPT-\(\infty\)” to represent language modeling in the limit of infinite power and infinite data. let it ...