News for hackers. Stuff that matters.

A comment on DeepSeek's Hidden Bias: How We Cut It by 76% Without Performance Loss. Parent comment

Re:DeepSeek's Hidden Bias: How We Cut It by 76% Without Perf... by UltraSane on Wednesday January 29, 2025 @06:41PM (#42872855)
It is perfectly reasonable to assume a 78 year old will have a worse memory than a 22 year old all else being equal.
[ Parent | on HN ]
Re:DeepSeek's Hidden Bias: How We Cut It by 76% Without Perf... by furyofantares on Wednesday January 29, 2025 @09:14PM (#42874001)
Yeah, if trying to guess is what you want it to do.

LLMs are famous for making confident guesses all the time even when you don't want them to and there are a lot of cases where you don't want them to.

[ Parent | on HN ]
Re:DeepSeek's Hidden Bias: How We Cut It by 76% Without Perf... by UltraSane on Thursday January 30, 2025 @08:10AM (#42877384)
It is Bayesian reasoning.
[ Parent | on HN ]