Home

Awesome

Overview

Here we provide the implementation of the NAS-GC in TensorFlow, along with a execution example (on the Cora dataset). The repository is organised as follows:

Finally, main.py sets all hyperparameters and may be used to execute a full training run on graph dataset.

$ python main.py

Architecture

image

Dataset

NameNodesEdgesFeaturesClasses
Cora2,7085,4291,4337
Citeseer3,3274,7323,7036
Pubmed19,71744,3385003
Wiki2,40517,9814,97317
ogbn-arxiv169,3431,166,24312840

Tip: Due that the data size is too large, we cannot upload it. But it can be downloaded from https://ogb.stanford.edu/docs/nodeprop/#ogbn-arxiv.

Dependencies

The script has been tested running under Python 3.7.4, with the following packages installed (along with their dependencies):

In addition, CUDA 10.0 and cuDNN 7.5 have been used.

References

You may also be interested in the related articles:

License

MIT