the network output does not match label dimensions

[listar0810]

Thx for the great job!
I debugg the train.py, the network output is [4,1,1000,513], but the groundtruth is [4,1,362,362] and I have not found the ray_tranform on the network output. When calculating the loss, it arise bug that the dimensions does not match.
Thx very much!