This is a Pytorch implementation of the model described in the paper "GMAN: A Graph Multi-Attention Network for Traffic Prediction". This is not an official implementation. The official implementation by the authors(TensorFlow) can be found here.
The datasets can be downloaded from Google Drive or Baidu Yun, provided by DCRNN. Place the traffic data files in the data folder and rename these as PeMS.h5 and METR.h5 for the respective datasets
Train and evaluate the model by executing
python train.py --dataset PeMS --K 2 --max_epoch 150 --batch_size 10
python train.py --dataset METR --K 4 --max_epoch 100 --batch_size 10