NVIDIA / caffe

Caffe: a fast open framework for deep learning.
http://caffe.berkeleyvision.org/
Other
672 stars 263 forks source link

Loss is more? #534

Closed rotorliu closed 6 years ago

rotorliu commented 6 years ago

Same model prototxt The loss of NVIDIA Caffe is more than the loss of BLVC Caffe when training. Why?