contrastive loss results in 1.#QNAN

51 views

Skip to first unread message

unread,

Mar 15, 2016, 12:50:35 PM3/15/16

to Caffe Users

Anyone tried to train with contrastive loss?

I am getting the 1.#QNAN loss in contrastive layer after first iteration.

What made this situation?

Thanks.

Reply all

Reply to author

Forward

0 new messages