Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
Hongzhi (Steve), Chen 5008af2210 | 1 year ago | |
---|---|---|
.. | ||
README.md | 2 years ago | |
configure.py | 1 year ago | |
gat.py | 1 year ago | |
main_proteins_full_dgl.py | 1 year ago | |
models.py | 1 year ago | |
utils.py | 1 year ago |
DGL implementation of GAT for ogbn-proteins. Using some of the techniques from Bag of Tricks for Node Classification with Graph Neural Networks (https://arxiv.org/abs/2103.13355).
Requires DGL 0.5 or later versions.
For the best score, run gat.py
and you should directly see the result.
python3 gat.py
For the score of GAT+labels
, run gat.py
with --use-labels
enabled and you should directly see the result.
python3 gat.py --use-labels
Here are the results over 10 runs.
Method | Validation ROC-AUC | Test ROC-AUC | #Parameters |
---|---|---|---|
GAT | 0.9276 ± 0.0007 | 0.8747 ± 0.0016 | 2,475,232 |
GAT+labels | 0.9280 ± 0.0008 | 0.8765 ± 0.0008 | 2,484,192 |
MWE-GCN and MWE-DGCN are GCN models designed for graphs whose edges contain multi-dimensional edge weights that indicate the strengths of the relations represented by the edges.
To use MWE-GCN:
python main_proteins_full_dgl.py --model MWE-GCN
To use MWE-DGCN:
python main_proteins_full_dgl.py --model MWE-DGCN
Additional optional arguments include 'rand_seed' (the random seed), 'cuda' (the cuda device number, if available), 'postfix' (a string appended to the saved-model file)
No Description
Python C++ Jupyter Notebook Cuda Text other
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》