02.01 · UNIT 01 · PyTorch in One Hour: foundations · Lesson
What is PyTorch
PyTorch is an open-source deep learning library. It has three parts: a tensor library, an autograd engine, and the deep learning utilities.
PLAIN-LANGUAGE INTRODUCTION
What is this?
PyTorch is an open-source deep learning library. It has three parts: a tensor library, an autograd engine, and the deep learning utilities.
One simple example
On a CPU-only machine, torch.__version__ prints 2.4.1, and torch.cuda.is_available() prints False.
What goes in?
A Python environment and the torch package.
What comes out?
A working PyTorch install that reports its version and GPU support.
Why does it matter?
The version and device checks prevent later failures from a wrong build.
What is it not?
The import name is torch, not pytorch.
WORK THROUGH THE IDEA
See the idea in more detail
- PyTorch is an open-source deep learning library that uses Python.
- It has three core parts: a tensor library, an automatic differentiation engine, and a deep learning library of layers, losses, and optimizers.
- AI contains machine learning, and machine learning contains deep learning. Deep learning trains deep neural networks.
- Install PyTorch with
pip install torch. Use a Python version one or two releases older than the newest. - Common mistake: a default install can be CPU-only. Check
torch.cuda.is_available()and name the CUDA version when a GPU is required.