Tencent/TurboTransformers

Refactor MultiheadAttention and other Layers

Aperta

#140 aperta il 22 lug 2020

 (0 commenti) (0 reazioni) (0 assegnatari)C++ (206 fork)github user discovery
enhancementhelp wanted

Metriche repository

Star
 (1546 stelle)
Metriche merge PR
 (Metriche PR in attesa)

Descrizione

Now the logic inside MultiheadAttention Layer is too complex for development. Moreover, some bugs exist in intermediate management. It is the first priority to rewrite these codes to make others easily understand what Turbo is doing.

Guida contributor