DGL v0.2 release [News] (3)

DGL v0.2 just released! Many updates including new features, bugfix and performance improvement. Up to 4.32x speedup of GCN and 3.59x speedup of RGCN models! New APIs on graph sampling. Checkout our release note and stay…

Mode readout output [Questions] (1)
Any suggestions on accelerating the mailbox and node.data of DGL? [Questions] (9)
Support initial node features in RGCN [Questions] (2)
Problem to sample the graph [Questions] (2)
Some questions about the code of GCN [Questions] (5)
How to use dynamic edges per attention head? [Questions] (5)
Problem of the tutorial of LGNN [Questions] (5)
How to build a weighted graph in dgl? [Questions] (3)
Low-level graph representations (C++ stuffs) [Questions] (5)
Where is the dgl.nn.pytorch.GraphConv()? [Questions] (5)
How to pass message inversely through an edge? [Questions] (3)
How to combine DGL with torch.nn.DataParallel? [Questions] (5)
How to deploy dgl based on docker? [Questions] (2)
Operations on edges group by weight [Questions] (7)
TreeLSTM -- nodes' hidden representations [Questions] (5)
How to Implement GCN layers above on LSTM in BatchedDGLGraph [Questions] (2)
Node types and edge types [Questions] (3)
Speed up DGLGraph.subgraphs (or dgl.unbatch)? [Questions] (4)
Broadcast readout output to each node in BatchedDGLGraph? [Questions] (3)
How can I sample the neighborhood dynamically? [Questions] (3)
How to fix random seeds in dgl in training? [Questions] (4)
Minibatch training on graph [Questions] (2)
Delete/remove or add edges [Questions] (3)
How many layers do you use for GCN based on DGL? [Questions] (2)
Can Reduce-function return Tensors with different dimension? [Questions] (9)
[Blog] Understand Graph Attention Network [News] (2)
I am confused about the calculation order of GAT [Questions] (2)
Question about update_all [Questions] (2)
Why DGL is fast than the Pytorch or MXNet? [Questions] (7)