Self-attention parallel b2 b3 b a2 a3 Q￥ Can be either input or a hidden layer 19Self-attention 𝒂 𝟒 𝒂 𝟑 𝒂 𝟐 𝒂 𝟏 𝒃 𝟒 𝒃 𝟑 𝒃 𝟐 𝒃 𝟏 Can be either input or a hidden layer parallel 19