- Clone PyGAT repo
- Test install by training
python train.py
- copy .pkl file and test.py into the PyGAT dir
- collect runtime by running test.py on trained-model.pkl using:
python test.py --pkl_file gat.pkl --time_file time.txt
- collect metrics by running profiler on trained-model.pkl using:
python test.py --pkl_file gat.pkl
(note: do not specify time file)
Same as PyGAT above, just with pygcn repo and pygcn folder
collect profiler metrics by running: 'python3 profiler.py --pkl_file gat.pkl'
mkdir gnns && cd gnns
git clone https://github.com/tkipf/pygcn.git
git clone https://github.com/HazyResearch/hgcn.git
git clone https://github.com/Diego999/pyGAT.git