The Offical Code of Relating-Up: Advancing Graph Neural Networks through Inter-Graph Relationships
The code has been implemented and tested with Python 3.10.0. To install the required packages:
$ pip install -r requirements.txt
### Evaluation Commands
python main.py --dataset DATASET Name of dataset
--model {GCN,GIN,GCNRU,GINRU}
Name of model
--seed SEED Random seed (default: 2023)
--n_splits N_SPLITS Number of splits
--n_repeats N_REPEATS
Number of times cross-validation needs to be repeated
--batch_size BATCH_SIZE
Input batch size for training (default: 128)
--lr LR Learning rate (default: 0.001)
--weight_decay WEIGHT_DECAY
Weight decay (L2 penalty) (default: 5e-4)
--gradient_clip_val GRADIENT_CLIP_VAL
The value at which to clip gradients
--patience PATIENCE Number of validation epochs with no improvement after which training will be stopped
--min_epochs MIN_EPOCHS
Force training for at least `min_epochs` epochs
--max_epochs MAX_EPOCHS
Stop training once `max_epochs` is reached
--hidden_dim HIDDEN_DIM
Number of hidden units (default: 128)
--num_layers NUM_LAYERS
Number of layers (default: 5)
--alpha ALPHA The parameter controls the balance between the Cross Entropy loss and the distillation loss
--beta BETA Weight of representation hints loss
--temp TEMP Temperature to smooth the logits
--cuda