Large Language Models are Surjective? Injective? Invertible?

Recently, there have been discussions on functional properties of Transformers, the basic building block of Large Language Models (LLM) and many other generative models. A paper ([1]) proves that Transformers can output anything given an …