search

qitianwu / NodeFormer

The official implementation of NeurIPS22 spotlight paper "NodeFormer: A Scalable Graph Structure Learning Transformer for Node Classification"
286 stars 27 forks source link

Is the relatinoal bias still calculated in w/o graph circumstances? #4

Closed AaronNing closed 1 year ago

AaronNing commented 1 year ago

Hi Qitian,

Thanks for this fantastic work!

In nodeformer.py line 284:

        # compute update by relational bias of input adjacency, requires O(E)
        for i in range(self.rb_order):
            z_next += add_conv_relational_bias(value, adjs[i], self.b[i], self.rb_trans)

I am a little bit confused that, in w/o graph circumstances (e.g. #2 ,) is the relatinoal bias still calculated? Thanks.

qitianwu commented 1 year ago

Hi Weixi,

In the w/o graph circumstances you mentioned, the relational bias is not used since we set rb_order=0 (see the run.sh for detailed hyper-parameter settings)

AaronNing commented 1 year ago

I see. Thanks for your quick reply : )