A tiny language model writes one letter at a time. In its training text, here’s how often each of a, e, h, i, o, u came right after a t. Write next_chances(counts): the chance of each letter coming next after a t, for a model that will be scored on text it has never seen (“How confused is the model?”).