diff --git a/encoder/perceptual_model.py b/encoder/perceptual_model.py
index 5da09fb29..f9430527a 100644
--- a/encoder/perceptual_model.py
+++ b/encoder/perceptual_model.py
@@ -30,6 +30,12 @@ def __init__(self, img_size, layer=9, batch_size=1, sess=None):
         self.features_weight = None
         self.loss = None
 
+        self.features_weight_placeholder = None
+        self.img_features_placeholder = None
+
+        self.features_weight_op = None
+        self.img_features_op = None
+
     def build_perceptual_model(self, generated_image_tensor):
         vgg16 = VGG16(include_top=False, input_shape=(self.img_size, self.img_size, 3))
         self.perceptual_model = Model(vgg16.input, vgg16.layers[self.layer].output)
@@ -46,6 +52,12 @@ def build_perceptual_model(self, generated_image_tensor):
         self.loss = tf.losses.mean_squared_error(self.features_weight * self.ref_img_features,
                                                  self.features_weight * generated_img_features) / 82890.0
 
+        self.features_weight_placeholder = tf.placeholder(self.features_weight.dtype, shape=self.features_weight.get_shape())
+        self.img_features_placeholder = tf.placeholder(self.ref_img_features.dtype, shape=self.ref_img_features.get_shape())
+
+        self.features_weight_op = self.features_weight.assign(self.features_weight_placeholder)
+        self.img_features_op = self.ref_img_features.assign(self.img_features_placeholder)
+
     def set_reference_images(self, images_list):
         assert(len(images_list) != 0 and len(images_list) <= self.batch_size)
         loaded_image = load_images(images_list, self.img_size)
@@ -65,8 +77,8 @@ def set_reference_images(self, images_list):
 
             image_features = np.vstack([image_features, np.zeros(empty_features_shape)])
 
-        self.sess.run(tf.assign(self.features_weight, weight_mask))
-        self.sess.run(tf.assign(self.ref_img_features, image_features))
+        self.sess.run(self.features_weight_op, {self.features_weight_placeholder: weight_mask})
+        self.sess.run(self.img_features_op, {self.img_features_placeholder: image_features})
 
     def optimize(self, vars_to_optimize, iterations=500, learning_rate=1.):
         vars_to_optimize = vars_to_optimize if isinstance(vars_to_optimize, list) else [vars_to_optimize]