Comment by turingbook

I am curious whether this theory can explain some phenomena of large language models.