Hello, I am trying to implement a GRBM in theano, but I can't make it work on MNIST. The Binary RBM works fine. I tried pretty much everything : normalizing, small learning rate, momentum / weight decay / sparsity, parallel tempering, learning rate decay, learning variances (couldn't make it work), no sampling when not learning variances, ... Can anyone share a code (or point me to a pure theano implementation) of a GRBM that works on MNIST ?
Ideally it would be in one file (commented or not), not a big library where it's difficult to understand what is going on. I know there is an implementation in pylearn2, but my goal is to implement a deep convolutional GRBM, and I saw that the code in pylearn2 for it does not work / is old / has bugs. So it would be a waste of time to learn to use pylearn2 if I have to patch it after. Thanks -- --- You received this message because you are subscribed to the Google Groups "theano-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
