leonzfa/iResNet

Loss for ires_disp_loss0, ires_disp_loss1, ires_disp_loss2 looks same

Opened this issue · 0 comments

Hi,
While training the network on my own dataset,
the losses ires_disp_loss0, ires_disp_loss1 and ires_disp_loss2 looks same.

Is this seem reasonable ?

image