Claude 3 catches researchers testing it

Re:Claude 3 catches researchers testing it by omneity on Saturday March 09, 2024 @04:52AM (#39650621)

I can’t find the tweet handy but a researcher showed this behavior was actually present in several datasets such as The Pile. So it’s less likely to be emergent behavior and just the model doing its thing.

[ Parent | on HN ]

Re:Claude 3 catches researchers testing it by omneity on Saturday March 09, 2024 @07:27AM (#39651221)

Found it. The claim is not as strong as I remembered though:

https://twitter.com/yanaiela/status/1764816123898565020

[ Parent | on HN ]

Re:Claude 3 catches researchers testing it by Culonavirus on Saturday March 09, 2024 @05:39AM (#39650807)

But but but arllms gen ai is the futureeeee, agi here we gooo. Seriously though, the last Lex Fridman podcast with Yann Lecun is hilarious at times. Lex is desperately trying to be the llm poster boy and Yann is always like "um, no."

[ Parent | on HN ]

Re:Claude 3 catches researchers testing it by Satam on Saturday March 09, 2024 @06:07AM (#39650932)

Haven't watched the interview but on Twitter Yann never admits being wrong, always bashes LLMs, and pushes his own architectures as the future but never shows big wins with them in practice.

[ Parent | on HN ]

Re:Claude 3 catches researchers testing it by Duxon on Saturday March 09, 2024 @06:13AM (#39650964)

Yup, it was quite like that in the Lex podcast when Yann went on about joint embeddings.

[ Parent | on HN ]

Re:Claude 3 catches researchers testing it by gjvc on Saturday March 09, 2024 @05:43AM (#39650822)

"Fraudman"

[ Parent | on HN ]

Re:Claude 3 catches researchers testing it by petters on Saturday March 09, 2024 @05:41AM (#39650815)

This is nothing new in Claude 3, either. But the model seems quite impressive in other ways

[ Parent | on HN ]