undefined | Better HN

0 pointskromem1y ago0 comments

They were able to demonstrate conceptual vectors that were consistent across different languages and different mediums (text vs images) and that when manipulated were able to represent the abstract concept in the output regardless of prompt.

What kind of evidentiary threshold would you want if that's not sufficient?

0 comments

jwilber1y ago

My point is that you claimed this is a rebuff against those claiming models don’t understand themselves. Your interpretation seems to assign intelligence to the algorithms.

While this research allows us to interpret larger models in an amazing way, it doesn’t mean the models themselves ‘understand’ anything.

You can use this on much smaller scale models as well, as they showed 8 months ago. Does that research tell us about how models understand themselves? Or does it help us understand how the models work?

kromemOP1y ago

"Understand themselves" is a very different thing than "understand what they are saying."

Which exactly are we talking about here?

Because no, the research doesn't say much about the former, but yes, it says a lot about the latter, especially on top of the many, many earlier papers working in smaller toy models demonstrating world modeling.

j / k navigate · click thread line to collapse