That's a reasonable question! The model is related to this trend but is a bit different - this article only touches on text-based models like GPT-3, whereas M6 is multi-modal (it deals with both images and text). Huge multi-modal models are a topic of their own I did not touch on in this piece.
I have a stupid, uninformed question: what about M6 from Alibaba? Where does the 10T (pre-trained) model fit in this discussion? https://www.infoq.com/news/2021/04/alibaba-m6-model/
That's a reasonable question! The model is related to this trend but is a bit different - this article only touches on text-based models like GPT-3, whereas M6 is multi-modal (it deals with both images and text). Huge multi-modal models are a topic of their own I did not touch on in this piece.
Golly, that's a big number. I'm sure you meant 181 billion.