AI Glossary Term

What does “Dataset” mean?

Definition

A collection of structured or unstructured examples used for training, validation, or testing.

Browse the full AI Glossary Explore all free AI guides

How Dataset works

Example

A hospital might assemble a dataset of 10,000 anonymized chest X-rays with diagnoses. Most are used to train a model, some to tune it, and a held-out portion to measure it at the end.

Why it matters

A model reflects the data it learned from. Knowing what a dataset contains, and what it leaves out, tells you where the model is likely to work or fail.

Common questions

What is in a dataset?

Examples such as images, text, or table rows, often with labels, plus metadata about where and how they were collected.

Why split a dataset?

To check performance on examples the model has not seen, which gives a more honest estimate than testing on the training data.

What is a datasheet?

A document describing a dataset's contents, collection method, and limitations, intended to help users judge if it fits their purpose.

Evaluation SetA held-out dataset used to measure model quality after training.Validation SetA dataset used during development to tune models and prevent overfitting.Data AugmentationTechniques that create modified training examples to improve model generalization.BenchmarkA standardized test or dataset used to measure and compare model performance.Supervised LearningTraining a model with labeled examples that map inputs to known outputs.Synthetic DataArtificially generated data used to augment, simulate, or protect sensitive training data.

Learn more in our free guides

LAION and Open Datasets Model Cards and Datasheets for Datasets Building an LLM Eval Dataset Opting Out of AI Training Datasets

Test yourself

LAION and Open Datasets Quiz Model Cards and Datasheets for Datasets Quiz Building an LLM Eval Dataset Quiz