Just to save time, I'd like to point out that a transformer with infinite parameters would still have no real intelligence.
-
So, the Church thesis is incorrect?
That's not the problem here; it's accurate generalization to new situations.
-
-
Sufficiently large transformers are approximately universal Turing machines.
-
But somehow they plateau after a certain size. When I trained a large Transformer on github dump for code completion in
@pycharm it was great at generating comments/counting braces, but completely failed to learn rules like syntactic scope. - Näytä vastaukset
Uusi keskustelu -
Lataaminen näyttää kestävän hetken.
Twitter saattaa olla ruuhkautunut tai ongelma on muuten hetkellinen. Yritä uudelleen tai käy Twitterin tilasivulla saadaksesi lisätietoja.