PALM2 Is a Terrible Poet

Why would you use Bison instead of Unicorn if the goal is to say something about “PALM2” (where Unicorn is the large model) and not Bison specifically? And why judge poetry primarily on rhyme (which is <i>not</i> a necessary element of poetry, and so is not unambiguously requested by “write a short poem…”) without rhyme being part of the prompt?

We compared AI models on simple task and google was terrible