Install required dependencies
- Follow the instructions on https://pytorch.org/get-started/previous-versions/ to install pyTorch 1.2.0
- Install pytorch-lightning 0.5.0
$ pip install pytorch-lightning==0.5.0
- Install pytorch_geometric (make sure Cuda is correctly installed)
$ pip install torch-scatter torch-sparse torch-cluster torch-spline-conv torch-geometric
Download the dataset of the chosen size into the src/data folder, go to src/ and then preprocess the data with
$ python data_maker.py
Finally, start training with
$ python train.py --exp_name <experiment name>
Note: In order to use GPU for training, we suggest you install Apex by
$ git clone https://github.com/NVIDIA/apex
$ cd apex
$ pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./
In the case of training on CPU or not wanting to use Apex, please comment out the line 45 and 46 in train.py.
Please cite our paper if you use the code.
@inproceedings{shao2020paper2repo,
title={paper2repo: GitHub Repository Recommendation for Academic Papers},
author={Shao, Huajie and Sun, Dachun and Wu, Jiahao and Zhang, Zecheng and Zhang, Aston and Yao, Shuochao and Liu, Shengzhong and Wang, Tianshi and Zhang, Chao and Abdelzaher, Tarek},
booktitle={Proceedings of The Web Conference 2020},
pages={629--639},
year={2020}
}