Reaction
Student releases TPU kernels for modern linear attention variants
The student's instructor calls the kernels “impressively fast” and asks Google DeepMind about collaborating on model architecture research.
TLDR
An instructor says one of their students released kernels for modern variants of linear attention on TPUs and calls them “impressively fast.” They say they still need TPUs to do model architecture research and ask Google DeepMind if it wants to collaborate.
Combined views
6
3 Sources, first seen ago
22 reposts
