diff --git a/src/mlpack/methods/ann/gan.hpp b/src/mlpack/methods/ann/gan.hpp
index f0cf7a743..b76e6b613 100644
--- a/src/mlpack/methods/ann/gan.hpp
+++ b/src/mlpack/methods/ann/gan.hpp
@@ -30,9 +30,9 @@ using namespace mlpack::distribution;
 namespace mlpack {
 namespace ann /** artifical neural network **/ {
 template<
-typename Model = FFN<CrossEntropyError<>>,
-typename InitializationRuleType = GaussianInitialization,
-class Noise = std::normal_distribution<>>
+typename Model,
+typename InitializationRuleType,
+class Noise>
 class GAN
 {
  public:
diff --git a/src/mlpack/methods/ann/gan_impl.hpp b/src/mlpack/methods/ann/gan_impl.hpp
index 0b787443c..97f945979 100644
--- a/src/mlpack/methods/ann/gan_impl.hpp
+++ b/src/mlpack/methods/ann/gan_impl.hpp
@@ -136,7 +136,7 @@ double GAN<Model, InitializationRuleType, Noise>::Evaluate(
       std::move(boost::apply_visitor(
           outputParameterVisitor, 
           discriminator.network.back())), std::move(currentTarget));
-  noise.imbue( [&]() { return noiseFunction(randGen);} );
+  noise.imbue( [&]() { return noiseFunction();} );
   generator.Forward(std::move(noise));
   arma::mat temp = boost::apply_visitor(
       outputParameterVisitor, generator.network.back());
@@ -193,7 +193,7 @@ Gradient(const arma::mat& /*parameters*/, const size_t i, arma::mat& gradient)
   // get the gradients of the discriminator
   discriminator.Gradient(discriminator.parameter, i, gradientDiscriminator);
 
-  noise.imbue( [&]() { return noiseFunction(randGen);} );
+  noise.imbue( [&]() { return noiseFunction();} );
   generator.Forward(std::move(noise));
   discriminator.predictors.col(numFunctions) = boost::apply_visitor(
       outputParameterVisitor, generator.network.back());
@@ -204,7 +204,7 @@ Gradient(const arma::mat& /*parameters*/, const size_t i, arma::mat& gradient)
 
   gradientDiscriminator += noiseGradientDiscriminator;
 
-  if (currentBatch % generatorUpdateStep == 0 && preTrainSize != 0)
+  if (currentBatch % generatorUpdateStep == 0 && preTrainSize == 0)
   {
     // Minimize log(1 - D(G(noise)))
     // pass the error from discriminator to generator
@@ -220,12 +220,13 @@ Gradient(const arma::mat& /*parameters*/, const size_t i, arma::mat& gradient)
 
     gradientGenerator = -gradientGenerator;
     gradientGenerator *= multiplier;
-
+/*
     if (counter % batchSize == 0)
     {
       Log::Info << "gradientDiscriminator = " << std::max(std::fabs(gradientDiscriminator.min()), std::fabs(gradientDiscriminator.max())) << std::endl;
       Log::Info << "gradientGenerator = " << std::max(std::fabs(gradientGenerator.min()), std::fabs(gradientGenerator.max())) << std::endl;
     }
+*/
   }
   counter++;
   if (counter >= numFunctions)
diff --git a/src/mlpack/tests/gan_test.cpp b/src/mlpack/tests/gan_test.cpp
index d969ef0fb..0453041d8 100644
--- a/src/mlpack/tests/gan_test.cpp
+++ b/src/mlpack/tests/gan_test.cpp
@@ -36,6 +36,7 @@ BOOST_AUTO_TEST_CASE(GanTest)
   size_t dOutputSize = 1;
   size_t batchSize = 100;
   size_t noiseDim  = 100;
+  size_t numSamples = 10;
 
   // Load the dataset
   arma::mat trainData, dataset, noiseData;
@@ -70,9 +71,9 @@ BOOST_AUTO_TEST_CASE(GanTest)
   // Optimizer
   MiniBatchSGD optimizer(batchSize, 1e-4, 100 * trainData.n_cols, 1e-5, true);
 
-  std::normal_distribution<> noiseFunction(0.0, 1.0);
+  std::function<double ()> noiseFunction = [] () { return math::RandNormal(0, 1); };
   // GAN model
-  GAN<> gan(trainData, generator, discriminator, gaussian, noiseFunction,
+  GAN<FFN<CrossEntropyError<>>, GaussianInitialization, std::function<double ()> > gan(trainData, generator, discriminator, gaussian, noiseFunction,
       trainData.n_rows, batchSize, 10, 10);
   gan.Train(optimizer);
 
@@ -99,7 +100,7 @@ BOOST_AUTO_TEST_CASE(GanTest)
 
     generatedData.submat(dim, i * dim, 2 * dim - 1, i * dim + dim - 1) = samples;
   }
-  std::string output_dataset = "./output_gan_ffn"
+  std::string output_dataset = "./output_gan_ffn";
   Log::Info << "Saving output to " << output_dataset << "..." << std::endl;
   generatedData.save(output_dataset, arma::raw_ascii);
   Log::Info << "Output saved!" << std::endl;