Hyperparameter
Hyperparameter
A value that humans set in advance, before training begins, to control how an AI model learns.
In Simple Terms
A hyperparameter is a rule-setting value that humans decide before training an AI. Things like how fast the AI learns, or how many times it goes through the training data, are numbers that people specify in advance. When building AI, developers adjust these values and run tests repeatedly to improve the model's accuracy.
Behind the Name
The name combines "hyper," meaning "above" or "beyond," with "parameter," a setting or value. Unlike the parameters an AI updates on its own during training, a hyperparameter sits a level above that process — it's a setting that controls the rules and structure of the learning itself, from the outside.
Take a Closer Look!
A hyperparameter is a configuration value that humans set in advance, before an AI or machine learning model starts training. As the model trains, it automatically learns patterns from the data — but the framework that shapes how that learning happens is something a person has to specify.
Roughly speaking, think of it like setting a study plan before you start studying. How many pages you cover per day, or how many times you review the material — these settings shape how efficiently the whole process goes. For an AI, the equivalent numbers include the learning rate, which controls how fast it learns, and how many times it passes through the training data.
If these settings aren't right, the AI's training might not go well, or it might end up learning patterns that are skewed toward just a narrow slice of the data. That's why developers run experiment after experiment, adjusting the values little by little, to find the combination that produces the best results.