Beginner · Fundamentals

Training

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. The process of adjusting a model's parameters so it learns patterns from labeled or unlabeled data.

Technical Definition

The process of adjusting a model's parameters so it learns patterns from labeled or unlabeled data.

How it works

Training feeds data through a model, computes a loss measuring how wrong its predictions are, and updates parameters via an optimizer (typically gradient descent). The loop repeats over many epochs until the loss plateaus. Training quality depends on data volume, data quality, model capacity, and hyperparameters like learning rate and batch size.

Related Concepts

• Loss Function — A mathematical function that measures how far the model's predictions are from the actual values, guiding the learning process.
• Inference — Using a trained model to make predictions on new data — the deployment phase of machine learning.
• Epoch — One full pass through the entire training dataset during model training.