-
The Hacker News launch thread reached 1,883 points and 492 comments, and most of its length went on one claim in the launch post: that Jev cannot hallucinate. This is the comment the thread organised around. Diogo Almeida answered it directly, that the same is true of all machine learning and a random forest is not said to hallucinate, and the thread's reply to that was that type safety is not factual correctness. He also called the reading of Jev as a zero-shot classifier 'very accurate'. The post's original title, naming it a frontier model 40 to 400x cheaper, was changed within the hour.
jacobgold, in Hacker News
First, congrats to the team on launching something genuinely interesting and new.
Seems like a more accurate title would be "Jev: Trading general purpose generation for fast typed inference" or something like that.
This is interesting, but the speed comparison seems misleading? A generative model that can output code in a Turing-complete language can do anything a computer can do.
Jev can only generate structured output, right? This is probably super useful for classification/routing/scoring, but it's nothing like the code generating models we're all using today for code and automation. [...]
eChai Partner Brands
The partnership →
