Build a Neural Network using PyTorch

Build a neural network using Linear Transformation

We will build a neural network for the following diagram. In the diagram, we have:

4 input features
2 hidden layers with 6 neurons each
1 output layer with 2 neurons

  # Neural Network Models
from torch import nn
 
class Net(nn.Module):
  # weight and bias are randomly initialised by PyTorch 
  def __init__(self):
    super.__init__()
    # num of in_features, num of out_features
    self.first_layer = nn.Linear(4, 6)
    self.second_layer = nn.Linear(6, 6)
    self.final_layer = nn.Linear(6, 2)
 
  # every subclass of nn.Module needs to implement a forward method
  def forward(self, x):
    # nn.Linear is a subclass of nn.Module
    # thus nn.Linear has a forward method that was written by Pytorch
    # self.first_layer.forward(x) is the same as self.first_layer(x)
    # output of first layer (x) is passed to second layer (x)
    # output of second layer (x) is passed to final layer (x
    x = self.first_layer(x)
    x = self.second_layer(x)
    x = self.final_layer(x)
    return x