/GNN_Workload_Characterization

Primary LanguagePythonMIT LicenseMIT

GNN_Workload_Characterization

PyGAT Usage

  1. Clone PyGAT repo
  2. Test install by training python train.py
  3. copy .pkl file and test.py into the PyGAT dir
  4. collect runtime by running test.py on trained-model.pkl using: python test.py --pkl_file gat.pkl --time_file time.txt
  5. collect metrics by running profiler on trained-model.pkl using: python test.py --pkl_file gat.pkl (note: do not specify time file)

PyGCN Usage

Same as PyGAT above, just with pygcn repo and pygcn folder

Profiler

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