NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:2394
Title:Efficient Graph Generation with Graph Recurrent Attention Networks

The reviewers agree that this paper has a number of interesting ideas, and makes a useful contribution to the literature. Please take the reviewer comments into account while preparing the camera ready version. A concern raised in the discussion was about the new ablation study, showing that the RNN is not needed and training a GNN longer can achieve similar performance. This muddies the message of the paper and weakens some of the claims. This should definitely be taken into account for the final version