AI Glossary Term
What does “Hyperparameter” mean?
A configuration value set before training, such as learning rate, batch size, or depth.
How Hyperparameter works
Example
Before training, an engineer picks a learning rate of 0.001 and a batch size of 32. If results are poor, she tries other values and compares scores on validation data.
Why it matters
Hyperparameters can decide whether training works at all. Because they are set by people, results depend on how carefully they are tuned and recorded.
Common questions
How is a hyperparameter different from a parameter?
Hyperparameters are chosen before training. Parameters, such as weights, are learned during training.
How are they chosen?
By manual tuning, grid or random search, or automated tools, always checked on validation data rather than test data.
Which hyperparameters matter most?
Often the learning rate, but it depends on the model. Others include batch size, number of layers, and regularization strength.