https://hellochang.github.io/tags/diffusion/ Recent content in Diffusion on Life of Chang https://hellochang.github.io/%3Clink%20or%20path%20of%20image%20for%20opengraph,%20twitter-cards%3E https://hellochang.github.io/%3Clink%20or%20path%20of%20image%20for%20opengraph,%20twitter-cards%3E Hugo -- 0.147.5 en Copyright © Chang Liu https://hellochang.github.io/projects/ Mon, 01 Jan 0001 00:00:00 +0000 https://hellochang.github.io/projects/ <blockquote> <p>Checkout my <a href="https://github.com/hellochang">GitHub</a> for coding projects</p></blockquote> <h2 id="towards-smaller-diffusion-models---gaussian-mixture-masks-and-unet-scaling">Towards Smaller Diffusion Models - Gaussian Mixture Masks and UNet Scaling</h2> <p><em>#Diffusion Models #Deep Learning #Research</em></p> <p>Denoising diffusion probabilistic models (DDPMs) have demonstrated superior image generation capabilities but suffer from slow inference and high computational costs. As a first step to address these challenges, we propose two novel modifications to enhance small-scale diffusion models- Gaussian mixture masks and scaled skip connections. More details in the <a href="../assets/projects/gmm_report.pdf">report</a> or <a href="../assets/projects/gmm_poster.pdf">poster</a>.</p>