the ce_loss = inf when I changed the dataset to cora

bdy9527 / SDCN

Structural Deep Clustering Network

Apache License 2.0

262 stars 73 forks source link

the ce_loss = inf when I changed the dataset to cora #32

Open beiTiaoLongJie opened 1 year ago

beiTiaoLongJie commented 1 year ago

Hi，and as a result of the situation，the performance of model is terrible（nmi=0.17）. What should I do to avoid it？ PS: I have excuted the pretrain.py to obtain pretrain weight.