upcarta
  • Sign In
  • Sign Up
  • Explore
  • Search
Mentions
Arthur B. 🌮 @ArthurB · Oct 29, 2022
  • From Twitter

Great read, long convolution with decaying weights finitely parametrized do well on long range benchmarks. What I love about this is that it uses a math cheat code, the FFT. When you use a cheat code, there's a real possibility of doing much better than expected.

Tweet Oct 29, 2022
"What Makes Convolutional Models Great on Long Sequence Modeling?" CNNs—not transformers—now dominate the hardest sequence modeling benchmark. Here's how this happened: [1/14] twitter.com/davisblalock/status/1586274303565189121/photo/1
by Davis Blalock
Post Add to Collection Mark as Completed
Recommended by 1 person
1 mention
Share on Twitter Repost
  • upcarta ©2025
  • Home
  • About
  • Terms
  • Privacy
  • Cookies
  • @upcarta