diff --git a/README.md b/README.md
index bde85d7..d2f3705 100644
--- a/README.md
+++ b/README.md
@@ -2,8 +2,8 @@
 A PyTorch implementation for [Adversarial Discriminative Domain Adaptation](https://arxiv.org/abs/1702.05464).
 
 ## Environment
-- Python 3.6
-- PyTorch 0.2.0
+- Python >= 3.6 (Tested on Python 3.8)
+- PyTorch >= 1.0.0 (Tested on PyTorch 1.11.0)
 
 ## Usage
 
diff --git a/core/adapt.py b/core/adapt.py
index 63d11c8..bc4e26d 100644
--- a/core/adapt.py
+++ b/core/adapt.py
@@ -107,9 +107,9 @@ def train_tgt(src_encoder, tgt_encoder, critic,
                               params.num_epochs,
                               step + 1,
                               len_data_loader,
-                              loss_critic.data[0],
-                              loss_tgt.data[0],
-                              acc.data[0]))
+                              loss_critic.item(),
+                              loss_tgt.item(),
+                              acc.item()))
 
         #############################
         # 2.4 save model parameters #
diff --git a/core/pretrain.py b/core/pretrain.py
index 4038c0f..40f6c37 100644
--- a/core/pretrain.py
+++ b/core/pretrain.py
@@ -52,7 +52,7 @@ def train_src(encoder, classifier, data_loader):
                               params.num_epochs_pre,
                               step + 1,
                               len(data_loader),
-                              loss.data[0]))
+                              loss.item()))
 
         # eval model on test set
         if ((epoch + 1) % params.eval_step_pre == 0):
@@ -78,8 +78,8 @@ def eval_src(encoder, classifier, data_loader):
     classifier.eval()
 
     # init loss and accuracy
-    loss = 0
-    acc = 0
+    loss = 0.
+    acc = 0.
 
     # set loss function
     criterion = nn.CrossEntropyLoss()
@@ -90,7 +90,7 @@ def eval_src(encoder, classifier, data_loader):
         labels = make_variable(labels)
 
         preds = classifier(encoder(images))
-        loss += criterion(preds, labels).data[0]
+        loss += criterion(preds, labels).item()
 
         pred_cls = preds.data.max(1)[1]
         acc += pred_cls.eq(labels.data).cpu().sum()
diff --git a/core/test.py b/core/test.py
index d414961..8c99960 100644
--- a/core/test.py
+++ b/core/test.py
@@ -13,8 +13,8 @@ def eval_tgt(encoder, classifier, data_loader):
     classifier.eval()
 
     # init loss and accuracy
-    loss = 0
-    acc = 0
+    loss = 0.
+    acc = 0.
 
     # set loss function
     criterion = nn.CrossEntropyLoss()
@@ -25,7 +25,7 @@ def eval_tgt(encoder, classifier, data_loader):
         labels = make_variable(labels).squeeze_()
 
         preds = classifier(encoder(images))
-        loss += criterion(preds, labels).data[0]
+        loss += criterion(preds, labels).item()
 
         pred_cls = preds.data.max(1)[1]
         acc += pred_cls.eq(labels.data).cpu().sum()
diff --git a/datasets/usps.py b/datasets/usps.py
index 62c4629..f6d906e 100644
--- a/datasets/usps.py
+++ b/datasets/usps.py
@@ -57,7 +57,7 @@ def __init__(self, root, train=True, transform=None, download=False):
             np.random.shuffle(indices)
             self.train_data = self.train_data[indices[0:self.dataset_size], ::]
             self.train_labels = self.train_labels[indices[0:self.dataset_size]]
-        self.train_data *= 255.0
+
         self.train_data = self.train_data.transpose(
             (0, 2, 3, 1))  # convert to HWC
 
diff --git a/models/discriminator.py b/models/discriminator.py
index 3e4a782..3740765 100644
--- a/models/discriminator.py
+++ b/models/discriminator.py
@@ -18,7 +18,6 @@ def __init__(self, input_dims, hidden_dims, output_dims):
             nn.Linear(hidden_dims, hidden_dims),
             nn.ReLU(),
             nn.Linear(hidden_dims, output_dims),
-            nn.LogSoftmax()
         )
 
     def forward(self, input):
diff --git a/params.py b/params.py
index 707967d..14374b5 100644
--- a/params.py
+++ b/params.py
@@ -4,8 +4,8 @@
 data_root = "data"
 dataset_mean_value = 0.5
 dataset_std_value = 0.5
-dataset_mean = (dataset_mean_value, dataset_mean_value, dataset_mean_value)
-dataset_std = (dataset_std_value, dataset_std_value, dataset_std_value)
+dataset_mean = dataset_mean_value
+dataset_std = dataset_std_value
 batch_size = 50
 image_size = 64
 
diff --git a/utils.py b/utils.py
index 7951c6d..944e4d1 100644
--- a/utils.py
+++ b/utils.py
@@ -15,7 +15,7 @@ def make_variable(tensor, volatile=False):
     """Convert Tensor to Variable."""
     if torch.cuda.is_available():
         tensor = tensor.cuda()
-    return Variable(tensor, volatile=volatile)
+    return tensor
 
 
 def make_cuda(tensor):