Fix PyTorch example code in pytorch.md (#2976)

Two issues are fixed:
1. The example code previously could not run since `TorchModel.from_pytorch` does not accept a `loss` parameter.
2. Fix space.

Contributed by @fanzhuyifan

docs/docs/ProgrammingGuide/pytorch.md

@@ -14,24 +14,24 @@ Two wrappers are defined in Analytics Zoo for Pytorch:
 
 1. TorchModel: TorchModel is a wrapper class for Pytorch model.
 User may create a TorchModel by providing a Pytorch model, e.g.
-```python
+    ```python
     from zoo.pipeline.api.torch import TorchModel
     TorchModel.from_pytorch(torchvision.models.resnet18(pretrained=True))
-```
+    ```
 The above line creates TorchModel wrapping a ResNet model, and user can use the TorchModel for
 training or inference with Analytics Zoo.
 
 2. TorchLoss: TorchLoss is a wrapper for loss functions defined by Pytorch.
 User may create a TorchLoss from a Pytorch Criterion, 
-```python
+    ```python
     from torch import nn
     from zoo.pipeline.api.torch import TorchLoss
 
-    az_criterion = TorchLoss.from_pytorch(loss=nn.MSELoss())
-```
-or from a custom loss function, which takes input and label as parameters
+    az_criterion = TorchLoss.from_pytorch(nn.MSELoss())
+    ```
+    or from a custom loss function, which takes input and label as parameters
 
-```python
+    ```python
     from torch import nn
     from zoo.pipeline.api.torch import TorchLoss
 
@@ -44,8 +44,8 @@ or from a custom loss function, which takes input and label as parameters
         loss = loss1 + 0.4 * loss2
         return loss
 
-    az_criterion = TorchLoss.from_pytorch(loss=lossFunc)
-```
+    az_criterion = TorchLoss.from_pytorch(lossFunc)
+    ```
 
 # Examples
 Here we provide a simple end to end example, where we use TorchModel and TorchLoss to