-

diff --git a/2.ASimpleNeuralNetwork/numpy_like_fizbuz.py b/2.ASimpleNeuralNetwork/numpy_like_fizbuz.py
index 62fa394..b6a5f5f 100644
--- a/2.ASimpleNeuralNetwork/numpy_like_fizbuz.py
+++ b/2.ASimpleNeuralNetwork/numpy_like_fizbuz.py
@@ -44,14 +44,18 @@ y = torch.from_numpy(trY).type(dtype)
 print(x.grad, x.grad_fn, x)
 # None, None, [torch.FloatTensor of size 900x10]
 
-w1 = torch.randn(input_size, hidden_units, requires_grad=True).type(dtype)
+w1 = torch.randn(input_size, hidden_units).type(dtype)
 w2 = torch.randn(hidden_units, output_size).type(dtype)
+w1.requires_grad = True
+w2.requires_grad = True
 
 print(w1.grad, w1.grad_fn, w1)
 # None, None, [torch.FloatTensor of size 10x100]
 
-b1 = torch.zeros(1, hidden_units, requires_grad=True).type(dtype)
-b2 = torch.zeros(1, output_size, requires_grad=True).type(dtype)
+b1 = torch.zeros(1, hidden_units).type(dtype)
+b2 = torch.zeros(1, output_size).type(dtype)
+b1.requires_grad = True
+b2.requires_grad = True
 
 no_of_batches = int(len(trX) / batches)
 for epoch in range(epochs):
@@ -86,7 +90,7 @@ for epoch in range(epochs):
 
         # Direct manipulation of data outside autograd is not allowed anymore
         # so this code snippet won't work with pytoch version 0.4+
-        try:
+        with torch.no_grad():
             w1 -= lr * w1.grad
             w2 -= lr * w2.grad
             b1 -= lr * b1.grad
@@ -95,9 +99,7 @@ for epoch in range(epochs):
             w2.grad.zero_()
             b1.grad.zero_()
             b2.grad.zero_()
-        except RuntimeError as e:
-            raise Exception('Direct manipulation of autograd Variable is not allowed in pytorch \
-version 0.4+. Error thrown by pytorch: {}'.format(e))
+
     if epoch % 10:
         print(epoch, output.item())
 # traversing the graph using .grad_fn