Ah, right you are. I'm surprised they're able to get the kind of results described in the article out of GPT-J. I've tinkered with it a bit myself, and it's nowhere near GTP-3.5 in terms of "intelligence". Haven't tried it for programming though; might be that it's better at that than general chat.
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
replies: