How do you solve the u-net "lazy" problem?
fengchexz opened this issue · comments
I notice that you don't use guide decoder,so how do you solve the u-net "lazy" problem?Can you help me? Thank you!
Sorry for the late answer. I noticed the Unet lazy problem in this paper. However, I succeeded to train my model without considering Unet lazy problem. In other worlds, I don't know how to solve the problem. I'm sorry not to answer your question fully. But, it may produce better quality to use guide decoder.