nd convolution and pooling revision#3515

Open

lfz wants to merge 25 commits intoBVLC:masterfrom

Commits on Jul 7, 2015

Added Nd Convolution and Pooling layers using cuDNN
Guillaume Bono
committed
Fix std::vector undefined reference + Various debug
Guillaume Bono
committed
Added factories for Nd operations
Guillaume Bono
committed
Corrected missing REGISTER_LAYER_CLASS
Guillaume Bono
committed
Revert "Corrected missing REGISTER_LAYER_CLASS"
Guillaume Bono
committed
Instantiate virtual forward and backward cpu (Not Implemented)
Guillaume Bono
committed
Fixed typo in class name
Guillaume Bono
committed
Remove Blob dimension if equals to 1
Guillaume Bono
committed
Made pad_shape and stride_shape optionnal, default to 0 and 1 respectively
Guillaume Bono
committed
Corrected missing parentheses in output shape computation
Guillaume Bono
committed
Minor modification to debug usage of cudnn primitives
Guillaume Bono
committed

Commits on Aug 3, 2015

Merge branch 'master' into nd-conv-pool
Guillaume Bono
committed

Commits on Sep 14, 2015

Merge branch 'master' into nd-conv-pool
Guillaume Bono
committed

Commits on Sep 18, 2015

Merge remote-tracking branch 'upstream/master' into nd-conv-pool
Guillaume Bono
committed

Commits on Jan 5, 2016

Commits on Jan 9, 2016

Commits on Jan 10, 2016

double const fix
lfz
committed