search

Cartus / AGGCN

Attention Guided Graph Convolutional Networks for Relation Extraction (authors' PyTorch implementation for the ACL19 paper)
MIT License
433 stars 88 forks source link

Why did you configure the first densely connected layer with GraphConvLayer? #36

Closed jeongwonkwak closed 2 years ago

jeongwonkwak commented 2 years ago

hello :) I ran the code you gave but I would like to ask you why you constructed the first densely connected layer (layers[0]) with GraphConvLayer. The other layers(layers[1], ..., layers[N-1]) were configured with MultiGraphConvLayer, right? please let me know any mistake. Thank you !

Cartus commented 2 years ago

Hi, you are right. The reason is that the first densely connected layer takes the dependency tree as the input (we only have one graph). For the latter layer, we constructed multiple graphs based on the node representations. That is the reason why they were configured with MultiGraphConvLayer.