项目文件夹

文件

T

文件历史

Quan (Andy) Gan 6c23fba8fa [Doc][Org] move edge_softmax to functional (#2442 )

* move edge_softmax to functional

* change examples

* fixes

* revert

* fix

* fix

* remove

* fix

* test

* add init

Co-authored-by: Minjie Wang <wmjlyjemaine@gmail.com>

2021-01-26 12:46:37 +08:00

gat.py

[Doc][Org] move edge_softmax to functional (#2442 )

2021-01-26 12:46:37 +08:00

README.md

Update (#1534 )

2020-05-17 21:17:03 +08:00

train_ppi.py

[Doc] Update GraphDataLoader in our docs (#2504 )

2021-01-11 13:11:19 +08:00

train.py

[bugfix] Fix a bunch of examples to be compatible with dgl 0.5 (#1957 )

2020-08-08 01:31:23 +08:00

utils.py

[Model] Early stop GAT (#750 )

2019-08-21 14:58:56 +08:00

README.md

Graph Attention Networks (GAT)

Paper link: https://arxiv.org/abs/1710.10903
Author's code repo (in Tensorflow): https://github.com/PetarV-/GAT.
Popular pytorch implementation: https://github.com/Diego999/pyGAT.

Dependencies

torch v1.0: the autograd support for sparse mm is only available in v1.0.
requests
sklearn

pip install torch==1.0.0 requests

How to run

Run with following:

python3 train.py --dataset=cora --gpu=0

python3 train.py --dataset=citeseer --gpu=0 --early-stop

python3 train.py --dataset=pubmed --gpu=0 --num-out-heads=8 --weight-decay=0.001 --early-stop

python3 train_ppi.py --gpu=0

Results

Dataset	Test Accuracy	Time(s)	Baseline#1 times(s)	Baseline#2 times(s)
Cora	84.02(0.40)	0.0113	0.0982 (8.7x)	0.0424 (3.8x)
Citeseer	70.91(0.79)	0.0111	n/a	n/a
Pubmed	78.57(0.75)	0.0115	n/a	n/a
PPI	0.9836	n/a	n/a	n/a

All the accuracy numbers are obtained after 300 epochs.
The time measures how long it takes to train one epoch.
All time is measured on EC2 p3.2xlarge instance w/ V100 GPU.
Baseline#1: https://github.com/PetarV-/GAT.
Baseline#2: https://github.com/Diego999/pyGAT.