guoswang

Sorry that I am not make it clear. I mean that the loss layer's output is the same as the loss layer's input when I train the net, I don't...

Hi, When I run the "C3D-v1.1/examples/c3d_ucf101_training/train_simple.sh", I get the high loss like this: ``` I0525 14:53:43.037530 28821 net.cpp:200] data does not need backward computation. I0525 14:53:43.037534 28821 net.cpp:242] This network...

me too