← the wiki

Model (ML)

The "program" that machine learning produces: patterns learned from training data, ready to make predictions.

A model is what you get after training: a bundle of learned numbers (parameters) that turns new inputs into predictions. Feed a fruit's weight and color into a trained fruit model and it answers "probably an orange".

Models range from a handful of numbers (a line through data points) to trillions (the large language models behind modern chatbots) — but the idea is identical at every size: patterns extracted from examples, replayed on new cases.

The best way to make it stick: watch it run. Start an interactive lesson →