Tencent/TurboTransformers

Refactor MultiheadAttention and other Layers

Aberta

#140 aberto em 22 de jul. de 2020

 (0 comentário) (0 reação) (0 responsável)C++ (206 forks)github user discovery
enhancementhelp wanted

Métricas do repositório

Stars
 (1.546 estrelas)
Métricas de merge de PR
 (Métricas PR pendentes)

Description

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.

Guia do colaborador