This is an implementation of HPN
for heterogeneous graphs.
python hpn_trainer.py
Note: this scripts only support
IMDB
, which means commandpython hpn_trainer.py --dataset ACM
will not run onACM
. If you want to test the performance of other datasets, you are suggested to make some modification of the trainer script.
Reference performance numbers for the IMDB dataset: (0.01, 200, 0.0001, 8, 0.8, 0.58178, 0.002811689883326394)
train test val = 400, 3478, 400, about 9% for trianing
Dataset | Our(tf) | Our(th) | Our(pd) |
---|---|---|---|
IMDB | 58.05(±0.38) | 57.23(±0.47) | 57.75(±0.34) |
TL_BACKEND=tensorflow python3 hpn_trainer.py --lr 0.01 --hidden_dim 512 --iter_K 1 --l2_coef 0.001 --drop_rate 0.4 --alpha 0.3
TL_BACKEND=torch python3 hpn_trainer.py --lr 0.01 --hidden_dim 512 --iter_K 1 --l2_coef 0.001 --drop_rate 0.4 --alpha 0.3
TL_BACKEND=paddle python3 hpn_trainer.py --lr 0.01 --hidden_dim 512 --iter_K 1 --l2_coef 0.001 --drop_rate 0.4 --alpha 0.3