Colloquium: Generating Images and Audio With the Hugging Face Diffusers Library

10 November 2022

Colloquium by Patrick von Platen, PhD, Univ. of Cambridge and Hugging Face

With the release of Dall-E 2, Google’s Imagen and Stable Diffusion, diffusion models have taken the world by storm. Most of you will have have already tried generating entertaining images with prompts like "a koala bear juggling eggs in the sunset in the style of van Gogh" or at least seen examples of what modern diffusion models are capable of.

Patrick von Platen holds a PhD from the University of Cambridge and works as Research Engineer at Hugging Face. He is the core developer of the recently released Hugging Face diffusers library [0], a state-of-the-art toolbox for building and running diffusion models covering unconditional as well as text-to-image, image-to-image, or text-to-audio generation tasks.

Patrick will talk about how you can run published models with just a couple of lines of Python code, or build one from scratch by combining and modifying noise schedulers, vision or audio models and inference pipelines. You will also have the chance to ask him anything about diffusion models, or what it means to build and maintain a library with >7.3k GitHub stars at Hugging Face.

[0]: https://github.com/huggingface/diffusers

Join Zoom Meeting:
Thursday, November 10th, 3pm - 4pm CEST (Berlin/Amsterdam/Paris)
https://uni-hamburg.zoom.us/j/61840700939?pwd=NHJMYWxINDdXSnc4azR6MVFhT1lxZz09
Meeting ID: 618 4070 0939
Passcode: 86724144