Closed yuyijie1995 closed 5 years ago
My parameter is like these --arch rfcn \ --dataset pascal_voc --net res101 \ --bs 2 --nw 4 \ --lr 0.0001 --lr_decay_step 5\ --cuda But in the first iter the loss is already Nan ,what should I do?
My parameter is like these --arch rfcn \ --dataset pascal_voc --net res101 \ --bs 2 --nw 4 \ --lr 0.0001 --lr_decay_step 5\ --cuda But in the first iter the loss is already Nan ,what should I do?