Sergey Levine argues LLMs develop emergent capabilities by composing simpler skills in novel ways instead of imitating training data
Anirudh Goyal's co-authored paper provides a mathematical framework.
Check out the full interview with one of the top robotics researchers: https://www.dwarkesh.com/p/sergey-levine
Dwarkesh Patel@dwarkesh_spWe pre-train LLMs on the whole of the internet. You might think this explains how they learn so many emergent capabilities: the knowledge is implicit in the training data. But in fact models can do things that were never demonstrated anywhere in training! @svlevine argues that the real source of emergent capabilities is compositionality:
#901Anirudh Goyal@ANIRUDHG9119@dwarkesh_sp
This is the phenomenon our paper (with @prfsanjeevarora) tried to formalize: as models scale, basic skills can compose into complex skills.
That gives a theory for emergence beyond direct imitation of training data.
Dwarkesh Patel@dwarkesh_spWe pre-train LLMs on the whole of the internet. You might think this explains how they learn so many emergent capabilities: the knowledge is implicit in the training data. But in fact models can do things that were never demonstrated anywhere in training! @svlevine argues that the real source of emergent capabilities is compositionality:
#999Cody Blakeney@CODE_STARI'm willing to believe this is true, but I also think people who make statements like this haven't seen the amount of crazy shit that is actually on the internet. It might be very hard for you to track down, but there is almost always an example for anything.
Dwarkesh Patel@dwarkesh_spWe pre-train LLMs on the whole of the internet. You might think this explains how they learn so many emergent capabilities: the knowledge is implicit in the training data. But in fact models can do things that were never demonstrated anywhere in training! @svlevine argues that the real source of emergent capabilities is compositionality: