Hands-On Generative AI with Transformers and Diffusion Models by Pedro Cuenca Apolinário Passos Omar Sanseviero and Jonathan Whitaker

Author:Pedro Cuenca, Apolinário Passos, Omar Sanseviero, and Jonathan Whitaker
Language: eng
Format: epub, mobi
Publisher: O'Reilly Media, Inc.
Published: 2023-03-15T00:00:00+00:00

For comparison, here are the results on MNIST when using the UNet implementation in the diffusers library, which features all of the above improvements:

Warning

This section will likely be expanded with results and more details in the future. We just haven’t gotten around to training variants with the different improvements yet!

Alternative Architectures

More recently, a number of alternative architectures have been proposed for diffusion models. These include:

Transformers. The DiT paper (“Scalable Diffusion Models with Transformers”) by Peebles and Xie showed that a transformer-based architecture can be used to train a diffusion model, with great results. However, the compute and memory requirements of the transformer architecture remain a challenge for very high resolutions.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.