Rishit-dagli/Nystromformer

An implementation of the Nyströmformer, using Nystrom method to approximate standard self attention

37 best points · 5 comments · 2 Hacker News posts · 57 GitHub stars · View on GitHub

Hacker News discussions

• I Implemented Nyströmformer (37 points, 5 comments, by dagli, 2022-08-21)
• I implemented Nystromformer, a transformer variant (1 points, 0 comments, by dagli, 2022-08-22)