logoalt Hacker News

hbrn • yesterday at 9:03 PM • 2 replies • view on HN

And what do you think Jev confidence score is?

Here's a hint: confidence is not generated by a model.


Replies

adrian17 • yesterday at 10:17 PM

Maybe I'm missing something, but why couldn't it be generated by the model? In older classification tasks with transformers like BERT, you could absolutely obtain a confidence score.

➕ show 1 reply
jgilias • yesterday at 9:33 PM

Thanks, fixed my understanding!

Do you think though that Luna being a model post-trained for chat produces over-confidence in logprobs?

➕ show 1 reply