Given that Google/DeepMind is one of the few labs actually trying to make smaller models that are just as performant (e.g. Chinchilla: https://arxiv.org/pdf/220

Given that Google/DeepMind is one of the few labs actually trying to make smaller models that are just as performant (e.g. Chinchilla: https://arxiv.org/pdf/2203.15556.pdf https://arxiv.org/pdf/2203.15556.pdf ), I'm optimistic.