Hacker News Reader

Comment by ram_rar | Hacker News Reader

> “Zero hallucination.”

This is the part that bothers me the most. How is it 0 hallucination, if the correct answer is not even the part of the options. There is no way to mark absentia or a way to know I absolutely cannot choose any of the options.

I am wondering, if anyones tried dead simple combinations of embedding with logistic regression to solve classifications problems?

View on HN

I believe the entire idea is that you always give it an explicit option for the "none of the above" cases and then handle the follow up branch outside of Jev like you usually would

Can’t produce text ergo can’t hallucinate I think is the marketing claim

Of course you can generate text. You just need to run it in an autoregressive loop as a sort of reverse of all the fun Jev-like papers that have come out in the last couple days. Ask it to predict the next letter in a string, then sample at your favorite temperature, then predict the next letter, etc. This will be quite expensive, and it may work terribly. I’m not personally inclined to try it. I am, however, curious whether it would work less horribly if you correctly guess what tokenizer the input uses and request a choice over next tokens consistent with the tokenizer in question.

It would be absolutely hilarious if you did this, asked it which model it was, and it gave a recognizable answer that wasn’t Jev.

See https://news.ycombinator.com/item?id=49778162.

Reply on news.ycombinator.com