Summary of Data-free Group-wise Fully Quantized Winograd Convolution Via Learnable Scales, by Shuokai Pan et al.
Data-Free Group-Wise Fully Quantized Winograd Convolution via Learnable Scales
by Shuokai Pan, Gerti Tuzi, Sudarshan Sreeram, Dibakar Gope
First submitted to arxiv on: 27 Dec 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 ways to reduce the computational and storage costs of large-scale text-to-image diffusion models while maintaining their quality. The authors investigate quantization methods to speed up inference time, focusing on convolution layers which account for a significant portion of computations in these models. They propose a finer-grained group-wise quantization approach that can handle fully quantized Winograd convolutions and reduce range differences in the Winograd domain. This method allows for near-lossless quality (FID and CLIP scores) in text-to-image generation tasks, outperforming state-of-the-art methods in image classification tasks on ResNet18 and ResNet-34. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about making big computers work faster without losing their ability to create realistic images. The authors want to make these computers more efficient so they can be used for many different tasks. They are trying to find a way to make the computer’s calculations faster and use less memory, while still keeping the quality of the images good. They have developed a new method that works well for both image generation and classification tasks. |
Keywords
» Artificial intelligence » Classification » Diffusion » Image classification » Image generation » Inference » Quantization » Resnet
