search

EvelynFan / FaceFormer

[CVPR 2022] FaceFormer: Speech-Driven 3D Facial Animation with Transformers
MIT License
761 stars 135 forks source link

The code of the alignment bias is not consistent with Equation 9. #86

Open FacePoluke opened 1 year ago

FacePoluke commented 1 year ago

image The values outside of the diagonal are 1, not -inf.

ChengyuanYan commented 9 months ago

the function returns a boolean mask so the placeholder value doesn't really matter. When the transformer computes attention, places that have non-zero value (i.e. True in the returned matrix) will be masked out, which is somewhat equivalent to adding -inf to the attention value