Skip to content

nn Basics

Jump to bottom
Joel Schlosser edited this page Jul 1, 2021 · 3 revisions
Clone this wiki locally

Scope

  • Understand what torch.nn is
  • Understand what a module is
  • Understand how modules are used to build and train neural networks
  • Understand how to author a module in PyTorch
  • Understand how to test modules in PyTorch

What is torch.nn?

torch.nn is the component of PyTorch that provides building blocks for neural networks. Its core abstraction is nn.Module, which encapsulates stateful computation with learnable parameters. Modules integrate with the autograd system and are generally trained using optimizers provided in torch.optim.

What is a module?

Read through the following links:

How to author and test a module in PyTorch

Work through the lab.

PyTorch presented to you with love by the PyTorch Team of contributors