Update 12-3.md (#780)

* Update 12-3.md Self-Attention Paragraph Typos: Issues #779 * Update 12-3.md Correction in Spanish * Update 12-3.md * Update 12-3.md * Update 12-3.md * Update 12-3.md * Update 12-3.md * Update 12-3.md * Update docs/es/week12/12-3.md Co-authored-by: Alfredo Canziani <[email protected]> * Update French 12-3.md * Update english comment 12-3.md * Update korean 12-3.md * Update Russian 12-3.md * Update turkish 12-3.md Co-authored-by: Alfredo Canziani <[email protected]>
Atcold · May 7, 2021 · dafad98 · dafad98
1 parent d967d82
commit dafad98
Show file tree

Hide file tree

Showing 8 changed files with 15 additions and 15 deletions.
diff --git a/docs/en/week12/12-3.md b/docs/en/week12/12-3.md
@@ -36,7 +36,7 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-where $\boldsymbol{a} \in \mathbb{R}^n$ is a column vector with components $\alpha_i$.
+where $\boldsymbol{a} \in \mathbb{R}^t$ is a column vector with components $\alpha_i$.
 
 Note that this differs from the hidden representation we have seen so far, where the inputs are multiplied by a matrix of weights.
 

diff --git a/docs/es/week12/12-3.md b/docs/es/week12/12-3.md
@@ -46,8 +46,8 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-<!-- where $\boldsymbol{a} \in \mathbb{R}^n$ is a column vector with components $\alpha_i$. -->
-donde $\boldsymbol{a} \in \mathbb{R}^n$ es un vector columna con componentes $\alpha_i$.
+<!-- where $\boldsymbol{a} \in \mathbb{R}^t$ is a column vector with components $\alpha_i$. -->
+donde $\boldsymbol{a} \in \mathbb{R}^t$ es un vector columna con componentes $\alpha_i$.
 
 <!-- Note that this differs from the hidden representation we have seen so far, where the inputs are multiplied by a matrix of weights. -->
 Nótese que esto difiere de la representación oculta que hemos visto hasta ahora, donde las entradas son multiplicadas por una matriz de pesos.

diff --git a/docs/fr/week12/12-3.md b/docs/fr/week12/12-3.md
@@ -49,7 +49,7 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-where $\boldsymbol{a} \in \mathbb{R}^n$ is a column vector with components $\alpha_i$.
+where $\boldsymbol{a} \in \mathbb{R}^t$ is a column vector with components $\alpha_i$.
 
 Note that this differs from the hidden representation we have seen so far, where the inputs are multiplied by a matrix of weights.
 
@@ -78,7 +78,7 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-où $\boldsymbol{a} \in \mathbb{R}^n$ est un vecteur colonne avec les composantes $\alpha_i$.
+où $\boldsymbol{a} \in \mathbb{R}^t$ est un vecteur colonne avec les composantes $\alpha_i$.
 
 Notez que cela diffère de la représentation cachée que nous avons vue jusqu'à présent, où les entrées sont multipliées par une matrice de poids.
 

diff --git a/docs/ja/week12/12-3.md b/docs/ja/week12/12-3.md
@@ -46,8 +46,8 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-<!-- where $\boldsymbol{a} \in \mathbb{R}^n$ is a column vector with components $\alpha_i$. -->
-ただし$\boldsymbol{a} \in \mathbb{R}^n$は要素が$\alpha_i$の列ベクトルです。
+<!-- where $\boldsymbol{a} \in \mathbb{R}^t$ is a column vector with components $\alpha_i$. -->
+ただし$\boldsymbol{a} \in \mathbb{R}^t$は要素が$\alpha_i$の列ベクトルです。
 
 <!-- Note that this differs from the hidden representation we have seen so far, where the inputs are multiplied by a matrix of weights.
 

diff --git a/docs/ko/week12/12-3.md b/docs/ko/week12/12-3.md
@@ -48,7 +48,7 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-where $\boldsymbol{a} \in \mathbb{R}^n$ is a column vector with components $\alpha_i$.
+where $\boldsymbol{a} \in \mathbb{R}^t$ is a column vector with components $\alpha_i$.
 
 Note that this differs from the hidden representation we have seen so far, where the inputs are multiplied by a matrix of weights.
 
@@ -76,7 +76,7 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-$\boldsymbol{a} \in \mathbb{R}^n$는 요소 $\alpha_i$를 갖고있는 열 벡터이다.
+$\boldsymbol{a} \in \mathbb{R}^t$는 요소 $\alpha_i$를 갖고있는 열 벡터이다.
 
 이는 우리가 지금까지 보아온 은닉 표현, 입력값이 가중치 행렬로 곱해지는 것과는 다르다.
 

diff --git a/docs/ru/week12/12-3.md b/docs/ru/week12/12-3.md
@@ -46,7 +46,7 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-where $\boldsymbol{a} \in \mathbb{R}^n$ is a column vector with components $\alpha_i$.
+where $\boldsymbol{a} \in \mathbb{R}^t$ is a column vector with components $\alpha_i$.
 
 Note that this differs from the hidden representation we have seen so far, where the inputs are multiplied by a matrix of weights.
 
@@ -75,7 +75,7 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-где $\boldsymbol{a} \in \mathbb{R}^n$ вектор-столбец с компонентами $\alpha_i$.
+где $\boldsymbol{a} \in \mathbb{R}^t$ вектор-столбец с компонентами $\alpha_i$.
 
 Отметим, что это отличается от внутреннего представления, которое мы видели до сих пор, где входы умножались на матрицу весов.
 

diff --git a/docs/tr/week12/12-3.md b/docs/tr/week12/12-3.md
@@ -58,13 +58,13 @@ $$
 $$
 
 <!--
-where $\boldsymbol{a} \in \mathbb{R}^n$ is a column vector with components $\alpha_i$.
+where $\boldsymbol{a} \in \mathbb{R}^t$ is a column vector with components $\alpha_i$.
 
 Note that this differs from the hidden representation we have seen so far, where the inputs are multiplied by a matrix of weights.
 
 Depending on the constraints we impose on the vector $\vect{a}$, we can achieve hard or soft attention.
 -->
-Burada $\boldsymbol{a} \in \mathbb{R}^n$, bileşenleri  $\alpha_i$ olan bir sütun vektörüdür.
+Burada $\boldsymbol{a} \in \mathbb{R}^t$, bileşenleri  $\alpha_i$ olan bir sütun vektörüdür.
 
 Dikkat edin, bu daha önce gördüğümüz, girdilerin ağırlık matrisleriyle çarpıldığı gizli gösterimlerden farklı.
 
@@ -577,4 +577,4 @@ model = TransformerClassifier(num_layers=1, d_model=32, num_heads=2,
 ```
 <!--
 Where this model is trained in typical fashion.
--->
+-->
diff --git a/docs/zh/week12/12-3.md b/docs/zh/week12/12-3.md
@@ -38,7 +38,7 @@ $$
 \boldsymbol{h} = \boldsymbol{X} \boldsymbol{a}
 $$
 
-其中$\boldsymbol{a} \in \mathbb{R}^n$ 是具有$\alpha_i$分量的列向量.
+其中$\boldsymbol{a} \in \mathbb{R}^t$ 是具有$\alpha_i$分量的列向量.
 
 请注意，这与我们到目前为止看到的隐藏表示形式不同，在隐藏表示形式中，输入乘以权重矩阵。