kaishengtai/neuralart

SGD optimizer breaks after updates

gongenhao opened this issue · 1 comments

Optimization result turns to be all zero after second iteration.

Thanks for pointing out this issue. Fixed by reducing the learning rate for SGD with VGG.