Open yiskw713 opened 3 years ago
self-attentionなしのゲート付きMLP構造(gMLP)で,画像認識タスクやNLPタスクでTransformerに匹敵する精度を実現.
概要
self-attentionなしのゲート付きMLP構造(gMLP)で,画像認識タスクやNLPタスクでTransformerに匹敵する精度を実現.
提案手法
実験
Comment
Reference
link