We initialize the features of nodes of ogbg_ppa dataset through aggregating the features of their connected edges by a Sum (Add) aggregation, just like what we do for ogbn_proteins.
--use_gpu False
--batch_size 32
--aggr add #options: [mean, max, add]
--block res+ #options: [plain, res, res+]
--conv gen
--gcn_aggr max #options: [max, mean, add, softmax, softmax_sg, softmax_sum, power, power_sum]
--num_layers 3
--conv_encode_edge False
--mlp_layers 2
--norm layer
--hidden_channels 128
--epochs 200
--lr 0.01
--dropout 0.5
--graph_pooling mean #options: [mean, max, add]
python main.py --use_gpu --conv_encode_edge --num_layers 28 --gcn_aggr softmax_sg --t 0.01
Test (use pre-trained model, download from Google Drive)
python test.py --use_gpu --conv_encode_edge --num_layers 28 --gcn_aggr softmax_sg --t 0.01