Mikoto10032/AutomaticWeightedLoss

The weight doesn't change. It's always 1.

zeehuilin opened this issue · 2 comments

zeehuilin commented

Hi, my network has 5 tasks, 2 kinds of loss-- nn.mseloss（point_loss） and Dice (seg_loss)
when I use AutomaticWeightedLoss, the self.params[i] always 1。
image

Mikoto10032 commented

After communicating with you privately, the problem has been resolved. Hahaha, the params of AutomaticWeightedLoss should also be put in the optimizer. Good luck!:stuck_out_tongue_winking_eye:

deryaguler95 commented

Can you help us for me I couldnt not resolved that this problem