search

sdemyanov / ConvNet

Convolutional Neural Networks for Matlab for classification and segmentation, including Invariang Backpropagation (IBP) and Adversarial Training (AT) algorithms. Trained on GPU, require cuDNN v5.
240 stars 141 forks source link

Dropout for training was always disabled #2

Closed rmanor closed 10 years ago

sdemyanov commented 10 years ago

@rmanor yeah, I forgot to test it and mention in the readme. Moreover, there was a bug in the weight updates implementation, that made it work only for momentum = 0. I'm fixing these things now and going to make a commit within a couple of hours.

rmanor commented 10 years ago

Cool, thanks. On Dec 5, 2013 1:12 PM, "Sergey Demyanov" notifications@github.com wrote:

@rmanor https://github.com/rmanor yeah, I forgot to test it and mention in the readme. Moreover, there was a bug in the weight updates implementation, that made it work only for momentum = 0. I'm fixing these things now and going to make a commit within a couple of hours.

— Reply to this email directly or view it on GitHubhttps://github.com/sdemyanov/ConvNet/pull/2#issuecomment-29889429 .