Update test_core.py to check weights for 3.12

UnitTests/test_core.py

@@ -115,16 +115,11 @@ def __init__(self):
             self.conv1 = nn.Conv2d(1, 20, 5)
             self.bn1 = nn.BatchNorm2d(20)
 
-    def is_normal_distribution(self, tensor, mean, std, num_std=3, tolerance=1e-5):
+    def is_normal_distribution(self, tensor, mean, std, num_std=3):
         # Calculate Z-score
         z_scores = (tensor - mean) / std
         # Check if values are within num_std standard deviations
-        within_std = torch.abs(z_scores) < num_std
-        # Check if values are close to the mean with a given tolerance
-        close_to_mean = torch.allclose(tensor.mean(), torch.tensor(mean), atol=tolerance)
-        # Check if values are close to the standard deviation with a given tolerance
-        close_to_std = torch.allclose(tensor.std(), torch.tensor(std), atol=tolerance)
-        return torch.all(within_std).item() and close_to_mean and close_to_std
+        return torch.all(torch.abs(z_scores) < num_std).item()
 
     def setUp(self):
         self.net = test_NetworkWeights.MockNetwork()
@@ -136,7 +131,6 @@ def test_weights_init(self):
         # Check if weights and biases of BatchNorm layers are initialized correctly
         for module in self.net.modules():
             if isinstance(module, nn.BatchNorm2d):
-                # Log the actual weights for diagnostics
                 print("BatchNorm2d weights: ", module.weight.data)
                 print("BatchNorm2d weights mean: ", module.weight.data.mean().item())
                 print("BatchNorm2d weights std: ", module.weight.data.std().item())
@@ -148,6 +142,7 @@ def test_weights_init(self):
                 self.assertTrue(torch.all(module.bias.data == 0).item(), 
                                 "Biases of BatchNorm layer are not initialized to 0")
 
+
 class test_same_seeds(unittest.TestCase):
 
     def test_reproducibility(self):