search

LifeIsTwoShort / Nuclear-norm-based-loss

Nuclear norm-based loss with Pytorch.
MIT License
4 stars 0 forks source link

compute loss #1

Open mzprose opened 3 years ago

mzprose commented 3 years ago

Hello, is your code suitable for calculating the nuclear norm loss between tensor (batch, channel, height, width)?

AndrewTal commented 3 years ago

Hello, is your code suitable for calculating the nuclear norm loss between tensor (batch, channel, height, width)?

是的，核范数交叉熵损失输入与交叉熵损失shape与type相同。另外，训练时你需要仔细调节学习率。