This DGL example implements the GNN model proposed in the paper DeeperGCN: All You Need to Train Deeper GCNs. For the original implementation, see here.
Contributor: xnuohz
The codebase is implemented in Python 3.7. For version requirement of packages, see below.
dgl 0.6.0.post1
torch 1.7.0
ogb 1.3.0
Open Graph Benchmark(OGB). Dataset summary:
| Dataset | #Graphs | #Node Feats | #Edge Feats | Metric |
|---|---|---|---|---|
| ogbg-molhiv | 41,127 | 9 | 3 | ROC-AUC |
Train a model which follows the original hyperparameters on different datasets.
# ogbg-molhiv
python main.py --gpu 0 --learn-beta- Table 6: Numbers associated with "Table 6" are the ones from table 6 in the paper.
- Author: Numbers associated with "Author" are the ones we got by running the original code.
- DGL: Numbers associated with "DGL" are the ones we got by running the DGL example.
| Dataset | ogbg-molhiv |
|---|---|
| Results(Table 6) | 0.786 |
| Results(Author) | 0.781 |
| Results(DGL) | 0.778 |
| Dataset | ogbg-molhiv |
|---|---|
| Results(Author) | 11.833 |
| Results(DGL) | 8.965 |