Thank you so much for such a good paper and idea for IRTS.
I am looking in the details of the implementation. but the edge weights seems no updates after attention operation in here line 343. The value of edge weight and alpha are all 1 during training. Is this on purpose due to some configuration of model?
Hi there,
Thank you so much for such a good paper and idea for IRTS.
I am looking in the details of the implementation. but the edge weights seems no updates after attention operation in here line 343. The value of edge weight and alpha are all 1 during training. Is this on purpose due to some configuration of model?
All the best Nathan