Reproducibility on the AI Learning track. A reproducible run sets seeds, pins library versions, and uses a fixed data snapshot. You will not get bit-identical GPU results every time, but you should get close, and you must be able to rerun the recipe.
This lesson assumes you already worked through Experiment logs.
The idea in practice
Pin versions. Set seeds for Python, NumPy, and the framework. Record the hardware. Treat a one-time notebook as a draft, not a result.
A concrete check
goal = {
'track': 'AI Learning',
'lesson': 'Reproducibility',
}
checks = [
'input available at decision time',
'score matches the real decision',
'failure case written down',
]
print(goal['lesson'])
for item in checks:
print('-', item)
Run the sketch locally if you have Python. The printout is a reminder of the checks, not a trained model. Replace the strings with the real inputs from your own example before you treat it as a design.
What usually goes wrong
Reporting a score you cannot rerun next week. When this happens, stop adding parameters or tools. Fix the check, the data, or the permission, then run the same example again.
What to write down
- The input you are allowed to use at decision time.
- The output and the score or pass rule.
- One failure you will test on purpose.
- What you will not claim the system can do.
Practice
Write the three pins you would require before you trust a number.
Self-check
- Say Reproducibility in one sentence that mentions an input and an output.
- Name the failure mode in this lesson and the check that would catch it.
Done when: you can explain this lesson without the page open, and you have a written failure case.