diff --git a/lpips/__init__.py b/lpips/__init__.py
index 6bad0d39..62b9a079 100755
--- a/lpips/__init__.py
+++ b/lpips/__init__.py
@@ -4,9 +4,8 @@
 from __future__ import print_function
 
 import numpy as np
-from skimage.measure import compare_ssim
 import torch
-from torch.autograd import Variable
+# from torch.autograd import Variable
 
 from lpips.trainer import *
 from lpips.lpips import *
@@ -51,6 +50,7 @@ def psnr(p0, p1, peak=255.):
     return 10*np.log10(peak**2/np.mean((1.*p0-1.*p1)**2))
 
 def dssim(p0, p1, range=255.):
+    from skimage.measure import compare_ssim
     return (1 - compare_ssim(p0, p1, data_range=range, multichannel=True)) / 2.
 
 def rgb2lab(in_img,mean_cent=False):
diff --git a/lpips/lpips.py b/lpips/lpips.py
index b1e64ba2..0deef85a 100755
--- a/lpips/lpips.py
+++ b/lpips/lpips.py
@@ -6,7 +6,6 @@
 import torch.nn.init as init
 from torch.autograd import Variable
 import numpy as np
-from IPython import embed
 from . import pretrained_networks as pn
 import torch.nn
 
diff --git a/lpips/pretrained_networks.py b/lpips/pretrained_networks.py
index 077a2441..a70ebbea 100644
--- a/lpips/pretrained_networks.py
+++ b/lpips/pretrained_networks.py
@@ -1,7 +1,6 @@
 from collections import namedtuple
 import torch
 from torchvision import models as tv
-from IPython import embed
 
 class squeezenet(torch.nn.Module):
     def __init__(self, requires_grad=False, pretrained=True):
diff --git a/lpips/trainer.py b/lpips/trainer.py
index 4c77c7be..c994a1cf 100755
--- a/lpips/trainer.py
+++ b/lpips/trainer.py
@@ -8,7 +8,6 @@
 from torch.autograd import Variable
 from scipy.ndimage import zoom
 from tqdm import tqdm
-from IPython import embed
 import lpips
 
 class Trainer():
diff --git a/lpips_loss.py b/lpips_loss.py
index ed16e34d..ed2f9921 100644
--- a/lpips_loss.py
+++ b/lpips_loss.py
@@ -19,8 +19,9 @@
 ref = lpips.im2tensor(lpips.load_image(opt.ref_path))
 pred = Variable(lpips.im2tensor(lpips.load_image(opt.pred_path)), requires_grad=True)
 if(opt.use_gpu):
-    ref = ref.cuda()
-    pred = pred.cuda()
+    with torch.no_grad():
+        ref = ref.cuda()
+        pred = pred.cuda()
 
 optimizer = torch.optim.Adam([pred,], lr=1e-3, betas=(0.9, 0.999))