SESKStuff Everyone Should Know
SESK: Stuff Everyone Should Know · sesk.info/ai-test-set-final-check

Which data split is saved for the final check after tuning on the validation set?

The test set.

Start playing free

  1. The test set ✔
  2. The training set
  3. The validation set again

Why: You train on the training set and compare versions on the validation set. When the model looks good, the test set gives a last, fresh double-check on data it has not been tuned to.

Huh, didn't know that: Validation and test sets can "wear out": the more you reuse them, the more the model quietly fits them.

The dad joke
The test set is like the final exam. No peeking, even if the model asks nicely.
Like this card?
Play the game

References

Answer: The test set. You train on the training set and compare versions on the validation set. When the model looks good, the test set gives a last, fresh double-check on data it has not been tuned to.

  1. Google ML Crash Course: Dividing the original dataset developers.google.com/machine-learning/crash-course/overfitting/dividing-datasets
    After repeated use of the validation set suggests that your model is making good predictions, use the test set to double-check your model.
    Checked 2026-10-10.
  2. Google ML Crash Course: Dividing the original dataset developers.google.com/machine-learning/crash-course/overfitting/dividing-datasets
    contains no duplicates from the training set
    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 IntelligenceResearchers wrote GSM1k, fresh math problems like GSM8K. What happened to some models' scores?Artificial IntelligenceHow do AI, machine learning and deep learning fit together?Artificial IntelligenceWhat did Alan Turing call his 1950 test of machine intelligence?

That's stuff everyone should know.