Login

Kye Gomez/Blockwise-Parallel-Transformer

32 times longer context window than vanilla Transformers and up to 4 times longer than memory efficient Transformers.

Issues looking for funding

use in pretrained model？
Fund

#1 opened in kyegomez/Blockwise-Parallel-Transformer

Fund
Explanation on the residual connection part?
Fund

#2 opened in kyegomez/Blockwise-Parallel-Transformer

Fund
kyegomez/Blockwise-Parallel-Transformer
Creatorkyegomez
Stars37
LicenseMIT License
RepositoryGitHub