Everybody is doing distillation to teach cheaper smaller model using output of more expensive one to save some money and speed up response times. The Western labs are using word: “distillation”, to blame China’s companies that they are stealing Western expensive and difficult to get for users Western AI. Western labs must somehow explain why they are wasting so much money for such mediocre results. The fault is not with the intelligence of Western leaders of the labs. The fault is “distillation”. Or isn’t it?
Užsisakykite:
Rašyti komentarus (Atom)
Komentarų nėra:
Rašyti komentarą