News for hackers. Stuff that matters.

boxcake

Ok - but what was it trained on? How much was copied from source material and how much was novel.

I could train a model that creates Shakespearean plays with a 10 word prompt - a 219 word spec is pretty meaningless for a model specifically trained on prior architectures, and specifically for this task.

I think this sounds more impressive than it is.

Re:S1: The $6 R1 Competitor? on Thursday February 06, 2025 @02:42PM (#42965772)
Isn't any answer to a question which hasn't been previously answered a derivative work? Or when a human write a parody of a song, or when a new type of music is influenced by something which came before.