Intermediate·18 min·0/47 done

Python Modules

Import code from modules, guard entry points, and pin dependencies.

Why it matters: Pinned requirements and clean imports are what make a training run reproduce on another machine.

Basics

Import a Module

Every .py file is a module. Import with import math or from math import sqrt.

Keep imports at the top of the file. Group stdlib, third-party, then your own code.

Avoid from module import *. It hides where names come from.

Tiny example

Import a stdlib module and use one function.

import math
print(math.sqrt(9))

Basics

if __name__ == "__main__": runs code only when you execute the file, not when you import it.

Put CLI entry points there so tests can import helpers without starting a job.

Tiny example

Run main only when the file is executed directly.

def main():
    print("run entry")

if __name__ == "__main__":
    main()

Basics

A virtual environment gives each project its own packages.

Create one with python -m venv .venv, activate it, then install from requirements.txt.

Pin versions (pandas==2.2.3, not pandas) so rebuilds stay stable.

Tiny example

Print a pinned requirement line.

pin = "pandas==2.2.3"
print("pin:", pin)

Then AI / ML

Same idea, used in real AI work. Press Try in lab to run it.

AI / ML example

AI / ML: script entry point

Importing defines helpers. Running the file starts the demo.

In AI / ML: Preprocess and train scripts sit under if __name__ == "__main__".

example.py

Python

def preprocess(text):
    return text.strip().lower()

def main():
    print(preprocess("  Hello "))

if __name__ == "__main__":
    main()

AI / ML example 2

AI / ML: pin requirements

Exact versions, one per line.

In AI / ML: CI and Docker images lock torch, pandas, and fastapi versions.

example.py

Python

# requirements.txt shape
pins = [
    "numpy==2.1.3",
    "pandas==2.2.3",
    "scikit-learn==1.5.2",
]
for line in pins:
    print(line)