about bi-level warp function "align_feature" and "align_feature_v"

[JAYatBUAA]

dear author,
I cannot understand the bi-level warp function "align_feature" and "align_feature_v". The align_feature_v seems like the inverse function of the align_feature. Can you explain more about them, especially about details of bi-level warp and its inverse function. Thank you very much!

[JAYatBUAA]

how to get coarse level corr matrix R whose shape is [b,u,v,k]

RABIT/models/networks/ranking_attention.py

Line 154 in 658c6af

b_v_r = torch.einsum('buvk,bvtd->butkd', R, b_v)