[ES] 13-3 Graph Convolutional Network III #595
Conversation
| @@ -0,0 +1,386 @@ | |||
| --- | |||
| lang-ref: ch.13-3 | |||
There was a problem hiding this comment.
| lang-ref: ch.13-3 | |
| lang: es | |
| lang-ref: ch.13-3 |
docs/es/week13/13-3.md
Outdated
| donde $\boldsymbol{x}$ es la representación de la entrada, $\boldsymbol{Ax}$ representa una rotación aplicada a la entrada $\boldsymbol{x}$ y $\sum_{v_j→v}{\eta(\boldsymbol{e_{j}})\odot \boldsymbol{Bx_{j}}}$ denota el sumatorio de la multiplicación elemento a elemento de las características de entrada rotadas y una compuerta (gate) $\eta(\boldsymbol{e_{j}})$. A diferencia de la antedicha GCN, donde promediamos las representaciones entrantes, el término compuerta es crítico para la implementación de las Residual Gated GCN, ya que nos permite modular las representaciones de entrada en base a las representaciones de los ejes. | ||
|
|
||
| <!-- Note that the summation is only over vertices ${v_j}$ that have incoming edges to vertex ${v}$. The term residual (in Residual Gated GCN) comes from the fact that in order to calculate the hidden representation $\boldsymbol{h}$, we add the input representation $\boldsymbol{x}$. The gate term $\eta(\boldsymbol{e_{j}})$ is calculated as shown below: --> | ||
| Nótese que el sumatorio es solo sobre los vértices ${v_j}$ que tienen ejes entrantes en el vértice ${v}$. El término residual (en las Residual Gated GCN) viene del hecho de que para calcular la representación oculta $\boldsymbol{h}$, sumamos la representación de entrada $\boldsymbol{x}$. El término compuerta $\eta(\boldsymbol{e_{j}})$ se calcula como se muestra a continuación: |
There was a problem hiding this comment.
Suggestion: change "sumatorio" to "suma" (wiki)
| (0): GatedGCN_layer(...) | ||
| (1): GatedGCN_layer(... )) | ||
| (MLP_layer): MLP_layer( | ||
| (FC_layers): ModuleList(...)) |
There was a problem hiding this comment.
From here on the comments are marked with red. Were you able to compile it?
There was a problem hiding this comment.
Yeah I noticed, my blind guess is that because the snippet above is labeled as Python but it's not real Python syntax it's breaking somehow the color highlighting below. The page rendered correctly anyway.
You can see the same issue happening in this PR for the chinese translation at exactly the same point:
https://github.com/Atcold/pytorch-Deep-Learning/pull/571/files
There was a problem hiding this comment.
Just for reference:
There was a problem hiding this comment.
Great! thank you for your time
…R suggestion about the translation.
No description provided.