diff --git a/vae/model.py b/vae/model.py
index d5e6f2c..6b7915e 100644
--- a/vae/model.py
+++ b/vae/model.py
@@ -1,6 +1,6 @@
 from torch import nn
 import torch
-from networks.layers import ConvNormAct, ResBlock, SA_Conv2d
+from networks.layers import ConvNormAct, ResBlock, SA_Conv2d, SN_ConvTranspose2d
 
 
 class VAE(nn.Module):
@@ -29,8 +29,8 @@ def __init__(self, z_dim, model_dim, img_size, img_channels, n_res_blocks=0):
             ConvNormAct(model_dim * 8, model_dim * 4, 'basic', 'up', activation='lrelu'),
             ConvNormAct(model_dim * 4, model_dim * 2, 'basic', 'up', activation='lrelu'),
             ConvNormAct(model_dim * 2, model_dim, 'basic', 'up', activation='lrelu'),
-            ConvNormAct(model_dim, img_channels, 'basic', 'up', activation='lrelu'),
-            nn.Sigmoid()
+            nn.ConvTranspose2d(model_dim, img_channels, kernel_size=4, stride=2, padding=1),
+            nn.Tanh()
         )
 
     def reparameterize(self, mu, logvar):
@@ -79,7 +79,7 @@ def __init__(self, z_dim, model_dim, img_size, img_channels):
             ConvNormAct(model_dim * 4, model_dim * 2, 'sn', 'up', activation='lrelu'),
             ConvNormAct(model_dim * 2, model_dim, 'sn', 'up', activation='lrelu'),
             SA_Conv2d(model_dim),
-            ConvNormAct(model_dim, img_channels, 'sn', 'up', activation='lrelu'),
+            SN_ConvTranspose2d(in_channels=model_dim, out_channels=img_channels, kernel_size=4, stride=2, padding=1),
             nn.Tanh()
         )