Final loss when you train your model
hding2455 opened this issue · comments
Hao Ding commented
Hi I am trying to train a SSD using xxyy and KL loss. May I ask the final loss when you train your model using xxyy and KL loss? It seems location_loss can be a negative number, is that normal?
Ethan He commented
The loss can be a negative number. BTW, please have a look at our CVPR version code: https://github.com/yihui-he/KL-Loss
The performance is improved.
Ethan He commented
have a look at YOLOv3 + KL-loss https://github.com/yihui-he/KL-Loss#pytorch-re-implementations