Emergent Abilities of Large Language Models
At a glance
Examined abilities that appeared in larger language models but were absent in smaller ones.
In one paragraph
Examined tasks where larger language models appeared to gain abilities absent in smaller models, such as arithmetic and answering questions. Called these patterns “emergent abilities” and explored what they might imply for predicting the effects of further scaling.
Authors 16
Organizations
Credited for this work at publication.