SESKStuff Everyone Should Know
SESK: Stuff Everyone Should Know · sesk.info/ai-hyperparameter-vs-parameter

Which of these is a hyperparameter you set, not something the model learns?

The learning rate. Weights and biases are learned parameters.

Start playing free

  1. The model's weights
  2. The learning rate ✔
  3. The model's bias values

Why: Parameters, the weights and biases, are learned during training. Hyperparameters like learning rate, batch size and number of epochs are chosen by people (or a tuning service) between training runs.

Huh, didn't know that: If 0.01 turns out too high, you might try a learning rate of 0.003 on the next run.

The dad joke
Hyperparameters are the settings you argue about. Parameters are the ones the model quietly figures out.
Like this card?
Play the game

References

Answer: The learning rate. Parameters, the weights and biases, are learned during training. Hyperparameters like learning rate, batch size and number of epochs are chosen by people (or a tuning service) between training runs.

  1. Google Machine Learning Glossary: hyperparameter developers.google.com/machine-learning/glossary
    learning rate is a hyperparameter. You could set the learning rate to 0.01 before one training session.
    Checked 2026-10-10.
  2. Google Machine Learning Glossary: parameter developers.google.com/machine-learning/glossary
    The weights and biases that a model learns during training.
    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 happens if the learning rate is set too high?Artificial IntelligenceAbout how many parameters did AlexNet, the 2012 ImageNet breakthrough network, have?Artificial IntelligenceWhat can a neuron's bias change even when its inputs stay fixed?

That's stuff everyone should know.