Multimodal models sometimes improve performance on all domains by providing additional context and understanding, though models trained on highly specific targeted datasets often perform well on narrow problems; for complicated or data-constrained problems, large multimodal base models with small amounts of fine-tuning data are preferable.

causalpending

Speaker

Justin Basilico

Evidence Quote

“if you have a narrow problem... very targeted data set... will often, you know, give you good performance... if you have a complicated problem... what you want is a model that has a huge amount of knowledge of lots of different things”

Source

Jeff Dean (Google): Exciting Trends in Machine Learning— Rice Ken Kennedy Institute
Created: 8/11/2026, 1:28:51 AM

Your notes

Loading notes...