Intro to Neural Networks

Code

Create check_setup.py — an environment check

Before training networks, you need the tools. We use PyTorch — the most popular library for neural networks in Python.

Install it (if you do not have it yet) and create check_setup.py. This tiny script just checks everything is ready:

import torch — loads PyTorch.
torch.__version__ — shows the version (confirms it is installed).
torch.cuda.is_available() — checks for a GPU (a graphics card for speed). A GPU is not required — a CPU is perfectly fine for the whole course.

Verify. In a terminal:

python check_setup.py

You should see roughly (the version number may differ):

PyTorch version: 2.2.0
Using CPU (no GPU found — that is perfectly fine for this course).

If you have a CUDA GPU, the second line will be Using CUDA (GPU): ... — also good. Either case means: the environment works, you can keep learning.

Gotcha. You run the script and instead of a version you get:
Traceback (most recent call last):
  File "check_setup.py", line 3, in <module>
    import torch
ModuleNotFoundError: No module named 'torch'
This means PyTorch is not installed in this Python environment yet. The fix — install it:
pip install torch
(Find the exact command for your system on pytorch.org — the CPU version is fine.) After installing, run python check_setup.py again — the version should appear. This ModuleNotFoundError is the most common first hurdle; now you know what it means.

check_setup.py CREATE

Create this file in ./

"""check_setup.py — confirm PyTorch is installed and see which device it uses."""

import torch


def main() -> None:
    print(f"PyTorch version: {torch.__version__}")

    if torch.cuda.is_available():
        # A GPU speeds up training, but it is NOT required for this course.
        print(f"Using CUDA (GPU): {torch.cuda.get_device_name(0)}")
    else:
        print("Using CPU (no GPU found — that is perfectly fine for this course).")


if __name__ == "__main__":
    main()