Picture for Uladzislau Yorsh

Uladzislau Yorsh

On Difficulties of Attention Factorization through Shared Memory

Add code
Mar 31, 2024
Viaarxiv icon

Linear Self-Attention Approximation via Trainable Feedforward Kernel

Add code
Nov 08, 2022
Viaarxiv icon

SimpleTron: Eliminating Softmax from Attention Computation

Add code
Dec 02, 2021
Figure 1 for SimpleTron: Eliminating Softmax from Attention Computation
Figure 2 for SimpleTron: Eliminating Softmax from Attention Computation
Figure 3 for SimpleTron: Eliminating Softmax from Attention Computation
Figure 4 for SimpleTron: Eliminating Softmax from Attention Computation
Viaarxiv icon