Epoch is one complete pass of the training algorithm through the entire training dataset. Models are usually trained for many epochs.
What it means in plain English
During training, a model does not learn everything in a single look at the data. Instead, it passes through the whole dataset repeatedly, improving a little each time. One full pass through all the training examples is called an epoch. Training for multiple epochs lets the model refine its parameters gradually — but too many epochs risks overfitting, where it starts memorising the data rather than learning general patterns.
The number of epochs is a setting the practitioner chooses and monitors.
A simple example
If you train a model for 10 epochs on a dataset of 10,000 images, the model looks at all 10,000 images ten times over the course of training, adjusting itself a little on each pass.
Why it matters
The number of epochs is a key training decision: too few and the model underfits, too many and it overfits. Understanding epochs clarifies that training is an iterative process of repeated exposure, not a single event.
Related terms
- Batch Size — how many examples are processed at once within an epoch.
- Overfitting — a risk of training for too many epochs.
- Training Data — the dataset an epoch passes through.
Frequently asked questions
What is an epoch in training?
One epoch is a single complete pass through the entire training dataset. Models are typically trained for many epochs so they can learn gradually.
How many epochs are needed?
It varies by problem: too few and the model underfits, too many and it can overfit. Validation performance is used to decide when to stop.
