LOSS of 0 when training denoising models

[dongttttttttt]

Hello! Thank you for the work you have done on denoising, it is very great. However, when I train the denoising model, I have a situation where the LOSS is nan, how can I solve this? Looking forward to your answer a million times!