Concept dictionary
Epoch
An epoch is one full pass through a training dataset during model training.
- Review state
- Reviewed in POC
- Freshness
- Current
- Version
- 2
Also called
- Epoch
- Training Epoch
Remember it this wayOne epoch is one full pass through the training set.
In plain language
An epoch is completed when a training process has worked through every example in the training set once. One epoch usually contains many batches and parameter updates.
Why it matters
Epochs provide a common way to describe training progress, compare learning curves, and decide when additional passes stop helping or begin to overfit.
Example
A dataset has 1,000 examples and the batch size is 100. Ten training iterations complete one epoch.
Common misconception
One epoch is not one update. The number of updates in an epoch depends on the dataset and batch size.
Related terms
Sources
- Google for Developers, “Machine Learning Glossary”provides established boundaries for core machine-learning vocabulary.
Learn more
- Compare [[batch|Batch]], [[training|Training]], [[dataset|Dataset]], and [[loss-function|Loss Function]] to place Epoch in the surrounding system.
- Follow the cited source material above for the method, standard, or research context behind this draft.