SESKStuff Everyone Should Know
SESK: Stuff Everyone Should Know · sesk.info/ai-mlops-version-everything

What must be versioned to reproduce an ML result?

The code, data, configuration, environment, and model artifact.

Start playing free

  1. The full training lineage ✔
  2. Only the final accuracy
  3. Only the model file name

Why: A model artifact alone does not reveal how it was made. Reproducibility requires the pipeline inputs and environment plus metadata linking them to the resulting model.

Huh, didn't know that: A model registry and an ML metadata store serve different but complementary tracking roles.

The dad joke
The experiment forgot its version and had an identity crisis.
Like this card?
Play the game

References

Answer: The full training lineage. A model artifact alone does not reveal how it was made. Reproducibility requires the pipeline inputs and environment plus metadata linking them to the resulting model.

  1. Google Cloud MLOps Guide docs.cloud.google.com/architecture/mlops-continuous-delivery-and-automation-pipelines-in-machine-learning
    ML metadata store records the following metadata: The pipeline and component versions that were executed.
    Checked 2026-10-10.

Think this answer is wrong? Tap "Challenge this answer" on the card in the game and tell us why.

Test yourself in the game

Artificial IntelligenceWhat is production data drift?Artificial IntelligenceWhy monitor model quality after deployment?Artificial IntelligenceHow do AI, machine learning and deep learning fit together?

That's stuff everyone should know.