Attention筆記

X ∈ R l s × d m ,   Y ∈ R l t × d m ,   W ∈ R d m × d h X\in \mathbb{R}^{l_s\times d_m},\ Y\in \mathbb{R}^{l_t \times d_m},\ W\in \mathbb{R}^{d_m\times d_h} X∈Rls​×dm​, Y∈Rlt​×dm​, W∈Rdm​×dh​ E = tanh

>>阅读原文<<