visit
The Nuts and Bolts of Parallel-UNet: Implementation Details by@backpropagation
The Nuts and Bolts of Parallel-UNet: Implementation Details
by BackpropagationOctober 6th, 2024
Too Long; Didn't Read
The implementation of TryOnDiffusion with Parallel-UNet includes a 256x256 architecture with key changes for improved performance. Trained using JAX on TPU-v4 for 500K iterations, the inference process is efficient, taking around 18 seconds for a batch of four.Authors:
(1) Luyang Zhu, University of Washington and Google Research, and work done while the author was an intern at Google; (2) Dawei Yang, Google Research; (3) Tyler Zhu, Google Research; (4) Fitsum Reda, Google Research; (5) William Chan, Google Research; (6) Chitwan Saharia, Google Research; (7) Mohammad Norouzi, Google Research; (8) Ira Kemelmacher-Shlizerman, University of Washington and Google Research.Table of Links
3.1. Cascaded Diffusion Models for Try-On
5. Summary and Future Work and References
Appendix
A. Implementation Details
A.1. Parallel-UNet
A.2. Training and Inference
TryOnDiffusion was implemented in JAX [4]. All three diffusion models are trained on 32 TPU-v4 chips for 500K iterations (around 3 days for each diffusion model). After trained, we run the inference of the whole pipeline on 4 TPU-v4 chips with batch size 4, which takes around 18 seconds for one batch.
This paper is under CC BY-NC-ND 4.0 DEED license.
L O A D I N G
. . . comments & more!
. . . comments & more!
About Author
TOPICS
RELATED STORIES
TryOnDiffusion: A Tale of Two UNets #ai-in-fashion
Oct 06, 2024
TryOnDiffusion: A Tale of Two UNets #ai-in-fashion
Oct 06, 2024
Behind the Scenes of TryOnDiffusion #ai-in-fashion
Oct 06, 2024
Cascaded Diffusion Models for Try-On #ai-in-fashion
Oct 06, 2024
