Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
Guangyu Zhou eaa51ac4dc | 3 months ago | |
---|---|---|
.. | ||
GRACE_POT_trainer.py | 3 months ago | |
README.md | 3 months ago | |
config.yaml | 3 months ago | |
eval_gracepot.py | 3 months ago |
This GammaGL example implements the model proposed in the paper xxxx.
Author's code:
This example was implemented by Siyuan Zhang
'Cora', 'Citeseer' and 'Pubmed'
Dataset | # Nodes | # Edges | # Classes |
---|---|---|---|
Cora | 2,708 | 10,556 | 7 |
Pubmed | 19,717 | 88,651 | 3 |
Photo | 7,650 | 238,162 | 8 |
Fisrt, make the directories for datasets and bounds to save
mkdir ~/datasets
mkdir ~/datasets/bounds
Then, go into the directory of a model. If you want to set the parameters, you should modify the ocnfiguration files in the directory ("config.yaml" for GRACE). The following is the command line to run each model (dataset used is Cora for example):
# original GRACE
python GRACE_POT_trainer.py --dataset Cora --gpu_id 0
# GRACE + POT
python GRACE_POT_trainer.py --dataset Cora --gpu_id 0 --use_pot --kappa 0.4
The result will be appended to the file "res/{dataset_name}_base_temp.csv" and "res/{dataset_name}_pot_temp.csv" respectively. You can also set the parameter "save_file" to specify the file to save results. We use minibatch to reduce the memory occupation, you can modify it in the code.
| Author's Code | GAMMAGL's Code |
| Dataset | Metrics | GRACE |GRACE-POT| GRACE |GRACE-POT|
|:-------:|:-------:|:-------:|:-------:|:-------:|:-------:|
| Cora | Mi-F1 | 78.2 | 79.2 | 78.2 | 82.2 |
| Cora | Ma-F1 | 76.8 | 77.8 | 77.1 | 81.3 |
| PubMed | Mi-F1 | 81.6 | 82.0 | 81.6 | 82.0 |
| PubMed | Ma-F1 | 81.7 | 82.4 | 80.5 | 80.1 |
| Photo | Mi-F1 | 91.2 | 91.8 | 89.8 | 90.0 |
| Photo | Ma-F1 | 89.2 | 90.0 | 88.5 | 87.9 |
No Description
Python C++ Cuda Markdown Text
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》