Computer Vision Meetup: Deci Diffusion - Triple the Speed of Stable Diffusion

Introducing DeciDiffusion 1.0. A new text-to-image 820 million parameter text-to-image latent diffusion. The model trained is on the LAION-v2 dataset and fine-tuned on the LAION-ART dataset. While it maintains foundational architecture elements from Stable Diffusion, such as the Variational Autoencoder (VAE) and CLIP’s pre-trained Text Encoder, DeciDiffusion’s primary innovation is the substitution of U-Net with the more efficient U-Net-NAS. In this talk, I’ll dive into the architectural innovations and showcase DeciDiffusion in action!

Speaker: Harpreet Sahota is the Deep Learning Developer Relations Manager at Deci AI.

Not a Meetup member? Scroll down on this page and join the Computer Vision Meetup friendliest to your timezone:

Recorded on Oct 12, 2023 at the Computer Vision Meetup.

#computervision #machinelearning #datascience #ai #artificialintelligence