Hugging Face Introduces HF Diffusers, Tools For Diffusion Models
Hugging Face has release a new library supporting the development and inference of Diffusion models.
Created on July 21|Last edited on July 25
Comment
Hugging Face has revealed a new library they've developed called Hugging Face Diffusers. This library comes with a host of tools for the development and inference of diffusion models, the type of model that DALL·E 2 is. Diffusion models work by taking a noise vector input and passing it iteratively through a model pipeline, and while many of the most popular diffusion models generate images, diffusion models can turn the noise vector into any other mode of data as well.
Hugging Face Diffusers comes with a few key things: Pre-made pipelines which you can use for a quick inference setup, pre-made models you can use as building blocks for your own models, and noise schedulers to manage noise and diffusion steps.
Various tutorials and examples are provided in the GitHub repository, including a couple Colab notebooks to follow, so head over there if you want to try your hand at the new library.
Find out more
Add a comment
Tags: ML News
Iterate on AI agents and models faster. Try Weights & Biases today.