We have defined a PyTorch network class DAN for you. You need to implement the forward pass for your deep averaging network. To start, first implement average that averages the words in a review and then implement forward that passes the "averaged" review to a linear layer to produce the model's belief.