Get our free extension to see links to code for papers anywhere online!

Chrome logo  Add to Chrome

Firefox logo Add to Firefox
Picture for Felix Chern

Felix Chern
Alert button

Large Models are Parsimonious Learners: Activation Sparsity in Trained Transformers


Oct 12, 2022
Zonglin Li, Chong You, Srinadh Bhojanapalli, Daliang Li, Ankit Singh Rawat, Sashank J. Reddi, Ke Ye, Felix Chern, Felix Yu, Ruiqi Guo, Sanjiv Kumar

Add code
Alert button

Figure 1 for Large Models are Parsimonious Learners: Activation Sparsity in Trained Transformers
Figure 2 for Large Models are Parsimonious Learners: Activation Sparsity in Trained Transformers
Figure 3 for Large Models are Parsimonious Learners: Activation Sparsity in Trained Transformers
Figure 4 for Large Models are Parsimonious Learners: Activation Sparsity in Trained Transformers

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

TPU-KNN: K Nearest Neighbor Search at Peak FLOP/s


Jun 30, 2022
Felix Chern, Blake Hechtman, Andy Davis, Ruiqi Guo, David Majnemer, Sanjiv Kumar

Add code
Alert button

Figure 1 for TPU-KNN: K Nearest Neighbor Search at Peak FLOP/s
Figure 2 for TPU-KNN: K Nearest Neighbor Search at Peak FLOP/s
Figure 3 for TPU-KNN: K Nearest Neighbor Search at Peak FLOP/s
Figure 4 for TPU-KNN: K Nearest Neighbor Search at Peak FLOP/s

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

New Loss Functions for Fast Maximum Inner Product Search


Sep 11, 2019
Ruiqi Guo, Quan Geng, David Simcha, Felix Chern, Sanjiv Kumar, Xiang Wu

Add code
Alert button

Figure 1 for New Loss Functions for Fast Maximum Inner Product Search
Figure 2 for New Loss Functions for Fast Maximum Inner Product Search

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email