Summary of Todo: Token Downsampling For Efficient Generation Of High-resolution Images, by Ethan Smith et al.
ToDo: Token Downsampling for Efficient Generation of High-Resolution Images
by Ethan Smith, Nayan Saxena, Aninda Saha
First submitted to arxiv on: 21 Feb 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper explores the role of attention mechanisms in generative image models, specifically focusing on dense attention in stable diffusion models. Traditional attention-based approaches can be computationally expensive, limiting their ability to process large images efficiently. To address this issue, the authors propose a novel training-free method called ToDo that leverages token downsampling for key and value tokens. This approach accelerates inference by up to 2x for common sizes and up to 4.5x or more for high resolutions like 2048×2048. The paper demonstrates that ToDo outperforms previous methods in balancing throughput and fidelity. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how attention works in computer-generated images. Right now, it takes a long time to make big pictures because the computers need to look at every tiny part of the picture. But what if we could make it faster? The researchers found a way to do this by making some parts of the picture less important. This helps the computers work faster and still makes good pictures. |
Keywords
* Artificial intelligence * Attention * Inference * Token
