diff --git a/src/mlpack/methods/ann/layer/fast_lstm_impl.hpp b/src/mlpack/methods/ann/layer/fast_lstm_impl.hpp
index 2f90963674..752b132ae4 100644
--- a/src/mlpack/methods/ann/layer/fast_lstm_impl.hpp
+++ b/src/mlpack/methods/ann/layer/fast_lstm_impl.hpp
@@ -50,7 +50,16 @@ FastLSTM<InputDataType, OutputDataType>::FastLSTM(const FastLSTM& layer) :
     inSize(layer.inSize),
     outSize(layer.outSize),
     rho(layer.rho),
-    weights(layer.weights)
+    forwardStep(layer.forwardStep),
+    backwardStep(layer.backwardStep),
+    gradientStep(layer.gradientStep),
+    weights(layer.weights),
+    batchSize(layer.batchSize),
+    batchStep(layer.batchStep),
+    gradientStepIdx(layer.gradientStepIdx),
+    grad(layer.grad),
+    rhoSize(layer.rho),
+    bpttSteps(layer.bpttSteps)
 {
   // Nothing to do here.
 }
@@ -60,7 +69,16 @@ FastLSTM<InputDataType, OutputDataType>::FastLSTM(FastLSTM&& layer) :
     inSize(std::move(layer.inSize)),
     outSize(std::move(layer.outSize)),
     rho(std::move(layer.rho)),
-    weights(std::move(layer.weights))
+    forwardStep(std::move(layer.forwardStep)),
+    backwardStep(std::move(layer.backwardStep)),
+    gradientStep(std::move(layer.gradientStep)),
+    weights(std::move(layer.weights)),
+    batchSize(std::move(layer.batchSize)),
+    batchStep(std::move(layer.batchStep)),
+    gradientStepIdx(std::move(layer.gradientStepIdx)),
+    grad(std::move(layer.grad)),
+    rhoSize(std::move(layer.rho)),
+    bpttSteps(std::move(layer.bpttSteps))
 {
   // Nothing to do here.
 }
@@ -74,7 +92,16 @@ FastLSTM<InputDataType, OutputDataType>::operator=(const FastLSTM& layer)
     inSize = layer.inSize;
     outSize = layer.outSize;
     rho = layer.rho;
+    forwardStep = layer.forwardStep;
+    backwardStep = layer.backwardStep;
+    gradientStep = layer.gradientStep;
     weights = layer.weights;
+    batchSize = layer.batchSize;
+    batchStep = layer.batchStep;
+    gradientStepIdx = layer.gradientStepIdx;
+    grad = layer.grad;
+    rhoSize = layer.rho;
+    bpttSteps = layer.bpttSteps;
   }
   return *this;
 }
@@ -88,7 +115,16 @@ FastLSTM<InputDataType, OutputDataType>::operator=(FastLSTM&& layer)
     inSize = std::move(layer.inSize);
     outSize = std::move(layer.outSize);
     rho = std::move(layer.rho);
+    forwardStep = std::move(layer.forwardStep);
+    backwardStep = std::move(layer.backwardStep);
+    gradientStep = std::move(layer.gradientStep);
     weights = std::move(layer.weights);
+    batchSize = std::move(layer.batchSize);
+    batchStep = std::move(layer.batchStep);
+    gradientStepIdx = std::move(layer.gradientStepIdx);
+    grad = std::move(layer.grad);
+    rhoSize = std::move(layer.rho);
+    bpttSteps = std::move(layer.bpttSteps);
   }
   return *this;
 }
diff --git a/src/mlpack/methods/ann/rnn_impl.hpp b/src/mlpack/methods/ann/rnn_impl.hpp
index 65ab29ad4f..6c909585c7 100644
--- a/src/mlpack/methods/ann/rnn_impl.hpp
+++ b/src/mlpack/methods/ann/rnn_impl.hpp
@@ -71,11 +71,6 @@ RNN<OutputLayerType, InitializationRuleType, CustomLayers...>::RNN(
         network.network[i]));
     boost::apply_visitor(resetVisitor, this->network.back());
   }
-  ResetCells();
-  if (parameter.is_empty())
-  {
-    ResetParameters();
-  }
 }
 
 template<typename OutputLayerType, typename InitializationRuleType,