generative model
A generative model is a statistical or neural model that attempts to learn or approximate the underlying probability distribution of data. Depending on the model family, a generative model can do one or both of the following:
- Generate new samples consistent with that distribution
- Assign likelihoods or densities for observed data points, though implicit models like GANs can only sample and can’t score them
Modern generative techniques include autoregressive models, variational autoencoders (VAEs), normalizing flows, diffusion and score-based models, and generative adversarial networks (GANs).
These generative models are trained using objectives designed to make their learned distribution approximate the true data distribution, such as maximum likelihood, adversarial losses, or denoising score matching under a variational bound, as diffusion models use.
In practice, that means you can either draw new samples from the model or score how likely a given observation is under it.
Related Resources
Tutorial
Generative Adversarial Networks: Build Your First Models
In this step-by-step tutorial, you'll learn all about one of the most exciting areas of research in the field of machine learning: generative adversarial networks. You'll learn the basics of how GANs are structured and trained before implementing your own generative model using PyTorch.
For additional information on related topics, take a look at the following resources:
By Leodanis Pozo Ramos • Updated Aug. 25, 2026
