Docker reproducibility anecdotes

Beginner Accelerating numerics & developer hygiene

Created by Pavel · 29.04.2026 at 19:11 UTC

Containers pin OS-level dependencies and, with lock files, Python package versions so collaborators and CI rebuild similar environments. They do not replace correctness testing—bad code reproduces faithfully too.

For reproducible ML, pair images with requirements.txt/uv.lock/poetry.lock, fixed seeds, and data versioning (DVC, lakeFS). Docker overhead is engineering trade-off: slower iteration locally vs fewer “works on my machine” incidents.

Docker docs: [1].

Sources

[1]https://docs.docker.com/get-started/Return to text

University approvals: 0

Tasks

Question 1

Pinned dependencies inside a Docker image mainly help with:

Hint

Same environment, same behaviour.

Guaranteed absence of security issues

Reproducible installs across machines, time, and teammates

Automatic GPU kernel fusion

Disabling unit tests

Question 2

A colleague says “Docker makes code correct.” What is the accurate response?

Hint

Reproducibility ≠ correctness.

Correct—containers prove statistical validity

Containers reproduce environments; they do not validate models or data

Docker replaces git

Docker forbids numpy

Question 3

freeze_tuple(parts) normalises each part with str(int(str(p))), joins by ., raises ValueError on bad conversion.

Hint

int normalises leading zeros.

def freeze_tuple(parts):
    pass

Starter code is prefilled; replace TODO blocks with your solution.

Runtime output (stdout/stderr)

1 test case will be used for grading

Run checks runtime behavior only. Final correctness is evaluated when you submit.

Card Info

Topic: Accelerating numerics & developer hygiene
Difficulty: Beginner
Completed: 0 users

Creator

Pavel