diff --git a/colossalai/tensor/_ops/linear.py b/colossalai/tensor/_ops/linear.py
index e75f18609..d8bc338a5 100644
--- a/colossalai/tensor/_ops/linear.py
+++ b/colossalai/tensor/_ops/linear.py
@@ -19,8 +19,9 @@ def colo_linear(types, args, kwargs, pg):
             bias = None
     else:
         bias = kwargs.get('bias', None)
-        if isinstance(bias, ColoTensor):
-            bias = bias.torch_tensor()
+    
+    if isinstance(bias, ColoTensor):
+        bias = bias.torch_tensor()
 
     # Add communication logic before and after linear call.
     if isinstance(weight, ColoTensor):
diff --git a/tests/test_tensor/test_op.py b/tests/test_tensor/test_op.py
index c45dca8da..6cd45df44 100644
--- a/tests/test_tensor/test_op.py
+++ b/tests/test_tensor/test_op.py
@@ -3,7 +3,6 @@ import torch
 from colossalai.tensor import ColoTensor
 from copy import deepcopy
 
-
 def test_linear():
     in_dim = 4
     out_dim = 5
@@ -45,7 +44,6 @@ def test_linear():
 #     torch.nn.init.uniform_(t)
 #     print(t)
 
-
 def test_element_wise():
     t_ref = torch.randn(3, 5)
     t = ColoTensor.init_from_torch_tensor(t_ref.clone())
@@ -66,6 +64,11 @@ def test_lazy_init_tensor():
     assert lazy_t._torch_tensor == None
     assert lazy_t.torch_tensor().numel() == 6
 
-if __name__ == '__main__':
+def check_all():
+    test_linear()
+    test_element_wise()
     test_no_wrap_op()
-    # test_element_wise()
+    test_lazy_init_tensor()
+
+if __name__ == '__main__':
+    check_all()