I have a binary classification neural network.
I want to compute the gradient of the score associated with a label with respect to the input. The following code compiles but returns 0 for the gradients on a non-zero input. ` conv_layer=network_dict['conv1'] conv_layer_params=lasagne.layers.get_all_params(conv_layer) weights_conv_layer=conv_layer_params[0] predictions=lasagne.layers.get_output(network_dict['prob']) top_pred_label=T.argmax(predictions,axis=1)[0] grad_label_conv=theano.grad(top_pred_label,weights_conv_layer) get_pred_and_grad=theano.function([input_var],[grad_label_conv]) ` Any suggestions on how to debug? 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.
