Fix loss computation error on cpu

Author: bpesquet

pytorch_tutorial/logistic_regression/test_logistic_regression.py

@@ -44,7 +44,7 @@ def test_logistic_regression(show_plots=False):
 
     # Convert dataset to PyTorch tensors and put them on GPU memory (if available)
     x_train = torch.from_numpy(inputs).float().to(device)
-    y_train = torch.from_numpy(targets).int().to(device)
+    y_train = torch.from_numpy(targets).long().to(device)
 
     # Create data loader for loading data as batches
     blobs_dataloader = DataLoader(

pytorch_tutorial/utils.py

@@ -18,7 +18,12 @@ def get_device():
 
 
 def get_parameter_count(model):
-    """Return the number of trainable parameters for a PyTorch model"""
+    """
+    Return the number of trainable parameters for a PyTorch model
+
+    Args:
+        model (torch.nn.Module): a PyTorch model
+    """
 
     return sum(p.numel() for p in model.parameters() if p.requires_grad)
 
@@ -81,7 +86,7 @@ def plot_decision_boundaries(model, x, y, title, device):
     xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))
 
     # Convert mesh to PyTorch tensors and put it on device memory
-    x_mesh = torch.FloatTensor(np.c_[xx.ravel(), yy.ravel()]).to(device)
+    x_mesh = torch.tensor(np.c_[xx.ravel(), yy.ravel()], dtype=torch.float).to(device)
 
     # Get predictions for mesh points
     with torch.no_grad():