News for hackers. Stuff that matters.
Re:Jev: New frontier model 40-400x cheaper and 20-200x faster by skerit on Tuesday September 15, 2026 @04:17PM (#49718291)
So in theory you could feed it incomplete text, and then ask it for the probabilities of what the next character could be?
Re:Jev: New frontier model 40-400x cheaper and 20-200x faster by mckngbrd on Tuesday September 15, 2026 @04:54PM (#49718779)
I think the joke here is getting missed
Re:Jev: New frontier model 40-400x cheaper and 20-200x faster by zenlikethat on Tuesday September 15, 2026 @05:43PM (#49719271)
Strings trigger us
Re:Jev: New frontier model 40-400x cheaper and 20-200x faster by CompleteSkeptic on Tuesday September 15, 2026 @04:25PM (#49718416)
you could, but it the model is not optimized for text

this is complex, but generating text is highly complicated and requires mode dropping to make long cohesive text

Re:Jev: New frontier model 40-400x cheaper and 20-200x faster by vatsachak on Tuesday September 15, 2026 @04:19PM (#49718323)
If you provide it an AST of the english language, yes.
Re:Jev: New frontier model 40-400x cheaper and 20-200x faster by aghilmort on Tuesday September 15, 2026 @08:47PM (#49720819)
was wondering same
Re:Jev: New frontier model 40-400x cheaper and 20-200x faster by iforgotmypasswo on Tuesday September 15, 2026 @10:47PM (#49721553)
Or a partially completed song, asking for the next note. I’m not sure if you’re joking, but using it for space constrained next token generation within a grammar sounds like a really neat use case.
Re:Jev: New frontier model 40-400x cheaper and 20-200x faster by copperx on Wednesday September 16, 2026 @12:40AM (#49722165)
Feed the generated note back into the input for the next query and you have ... autoregression?