search

XinyiYing / BasicIRSTD

BasicIRSTD toolbox
154 stars 24 forks source link

total_loss = NaN #16

Closed Reginaii closed 4 months ago

Reginaii commented 5 months ago

作者，您好，在复现代码时，输出total_loss = NaN model_name :DNA-Net dataset_name :SIRST3

Reginaii commented 5 months ago

train.py

XinyiYing commented 4 months ago

数据读取看看有没有问题？学习率降低一些？我目前没有遇到这个问题过，