A machine-learning epoch is one pass through the training data: in the standard fixed-dataset case, every training example is processed once. An epoch may contain many batches and training updates; it is not a single update.
Epoch, batch and iteration: what each term means
These terms describe different levels of a training process:
- Epoch: A full pass over the training set, so each example is processed once under the standard definition. Google’s Machine Learning Glossary defines it as “A full training pass over the entire training set such that each example has been processed once.”
- Batch: A group of examples processed together during training.
- Iteration (or step): One training update. In mini-batch training, the model processes a batch and typically updates its parameters once. A neural-network iteration involves a forward pass followed by a backward pass.
So an epoch is a pass through the data, while an iteration is an update made using a batch—or, in other training approaches, a different amount of data.
How many iterations are in one epoch?
For a fixed dataset of N examples and batch size B, the number of iterations is commonly about N/B. The precise count depends on what the training implementation does with a final incomplete batch.
#1 Best Overall
- Use scikit-learn to track an example ML project end to end
- Explore several models, including support vector machines, decision trees, random forests, and ensemble methods
- Exploit unsupervised learning techniques such as dimensionality reduction, clustering, and anomaly detection
- Dive into neural net architectures, including convolutional nets, recurrent nets, generative adversarial networks, autoencoders, diffusion models, and transformers
- Use TensorFlow and Keras to build and train neural nets for computer vision, natural language processing, generative models, and deep reinforcement learning
For example, Google’s documentation works through a dataset of 1,000 examples:
| Training setup | Examples per batch | Iterations in one epoch |
|---|---|---|
| Mini-batch example | 50 | 20 |
| Mini-batch example | 100 | 10 |
These are illustrative calculations, not performance measurements. With the same dataset, a smaller batch means more iterations per epoch; a larger batch means fewer. It does not follow that the two setups learn equally well.
Rank #2
Why update counts differ by training method
Google’s worked example shows how the amount of data used for each update changes the number of updates in an epoch:
- Full-batch training: Uses the full dataset for one update, so it updates once per epoch.
- Stochastic gradient descent: Uses one example per update, so it updates once per example.
- Mini-batch SGD: Uses a batch per update, so it updates once per batch.
That is why “one epoch” and “one update” are not interchangeable: the update count depends on the training method and batch size.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Does every epoch visit every example exactly once?
That is the standard definition for a fixed training set, but the word can describe a practical training phase rather than a literal, guaranteed traversal in every setup. Keras describes an epoch as an “arbitrary cutoff,” generally one pass through the dataset, used to divide training into phases for logging and periodic evaluation.
For streaming or dynamically sampled data, repeated examples, or an input pipeline limited to a custom number of steps, an epoch may not mean that every distinct example was visited exactly once. Interpret it according to the framework’s convention and how that training loop supplies data. The term concerns training data; validation or test evaluation is separate.
Rank #4
What does the number of epochs tell you?
Training commonly reuses the training set over multiple epochs. More epochs take more training time and can improve a model, but there is no universally correct count: the appropriate amount depends on the task and should be determined through experimentation. Increasing epochs does not guarantee better results.
When comparing training runs, epoch counts alone can be misleading if the batch sizes or data-sampling rules differ. Compare the batch size, updates per epoch, total examples processed, training time and validation results to understand what each run did and how it performed.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
