Epoch
Last Updated: July 29, 2026 | By Mihail Sebastian | AI Dictionary
One complete pass through the entire training dataset. Models train for many epochs, each made up of multiple parameter updates on mini-batches.
What is an Epoch?
An epoch is one complete pass through the entire training dataset during model training. Training runs are measured in epochs: a model trained for 50 epochs has seen every example in the training set 50 times.
One pass is rarely enough. Weights move a little with each update, so the model needs repeated exposure to the data before the loss settles.
How Epochs, Batches, and Iterations Relate
Three terms describe the same training loop at different zoom levels. The dataset is split into mini-batches; processing one batch and updating the weights once is an iteration; completing every batch in the dataset is an epoch.
The arithmetic ties them together: iterations per epoch equals dataset size divided by batch size. A dataset of 10,000 samples with a batch size of 100 yields 100 iterations per epoch, so 50 epochs means 5,000 weight updates.
The number of epochs is a hyperparameter with a trade-off on each side. Too few epochs leave the model underfitting; too many let it memorize the training data and start overfitting.
Example of an Epoch
A team trains a fraud classifier on 200,000 transactions with a batch size of 500, so each epoch is 400 iterations. After every epoch they check the loss on a held-out validation set.
For 12 epochs the validation loss falls. At epoch 13 it turns upward while training loss keeps dropping: the model has begun memorizing. Early stopping kicks in, and the weights from epoch 12 become the final model.
Related AI terms: Mini-Batch · Early Stopping · Overfitting · Stochastic Gradient Descent · Training Set
Did you like the Epoch gist?
Learn about 250+ need-to-know artificial intelligence terms in the AI Dictionary.
Mihail Sebastian — Writes about AI governance, regulation, and the technology behind them. Placeholder bio — replace with a real credential line. About