Hands-On Reinforcement Learning for Games

Building neural networks with Torch

In the last section, we explored building computational graphs that resemble neural networks. This is a fairly common task as you may expect. So much so that PyTorch, as well as most DL frameworks, provides helper methods, classes, and functions to build DL graphs. Keras is essentially a wrapper around TensorFlow that does just that. Therefore, in this section, we are going to recreate the last exercise's example using the neural network helper functions in PyTorch. Open the Chapter_6_2.py code example and follow the next exercise:

1. The source code for the entire sample is as follows:

import torchbatch_size, inputs, hidden, outputs = 64, 1000, 100, 10x = torch.randn(batch_size, inputs)y = torch.randn(batch_size, ...