Tencent/TurboTransformers

Refactor MultiheadAttention and other Layers

Ouverte

#140 ouverte le 22 juil. 2020

 (0 commentaire) (0 réaction) (0 personne assignée)C++ (206 forks)github user discovery
enhancementhelp wanted

Métriques du dépôt

Stars
 (1 546 étoiles)
Métriques de merge PR
 (Métriques PR en attente)

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.

Guide contributeur