Hi author,
Great work! I have a question about the BP part. In the paper, you mention adopting BP to optimize the loss, but I didnt see an explicit BP code. It seems training proceeds via standard backprop. Could you clarify how BP is implemented in your code? I'm really interested in this part. Thank you!
Xin
Hi author,
Great work! I have a question about the BP part. In the paper, you mention adopting BP to optimize the loss, but I didnt see an explicit BP code. It seems training proceeds via standard backprop. Could you clarify how BP is implemented in your code? I'm really interested in this part. Thank you!
Xin