News for hackers. Stuff that matters.

A comment on Vision Transformers (ViT) Explained. Parent comment

Re:Vision Transformers (ViT) Explained by thethimble on Tuesday November 29, 2022 @02:42PM (#33792229)
What’s the advantage of learning this positional embedding vs hardcoding one that has the desired properties of location similarity?
[ Parent | on HN ]

There are no comments on this comment yet.