Tencent/TurboTransformers

Refactor MultiheadAttention and other Layers

開放

#140 建立於 2020年7月22日

 (0 則留言) (0 個反應) (0 位負責人)C++ (206 個分叉)github user discovery
enhancementhelp wanted

倉庫指標

星標
 (1,546 顆星)
PR 合併指標
 (PR 指標待抓取)

描述

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.

貢獻者指南