Summary of Hart: Efficient Visual Generation with Hybrid Autoregressive Transformer, by Haotian Tang et al.
HART: Efficient Visual Generation with Hybrid Autoregressive Transformer
by Haotian Tang, Yecheng Wu, Shang Yang, Enze Xie, Junsong Chen, Junyu Chen, Zhuoyang Zhang, Han Cai, Yao Lu, Song Han
First submitted to arxiv on: 14 Oct 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 introduces the Hybrid Autoregressive Transformer (HART), a novel visual generation model that can directly generate high-quality 1024×1024 images, comparable to state-of-the-art diffusion models. The HART model addresses limitations in existing autoregressive (AR) models by introducing a hybrid tokenizer that decomposes continuous latents into discrete and residual components. This approach enables scalable-resolution AR modeling of the discrete component and lightweight residual diffusion for the continuous component. The paper demonstrates significant improvements over previous AR models, achieving 31% better reconstruction FID and 4.5-7.7x higher throughput with lower MACs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary HART is a new way to create high-quality images using a special type of model called an autoregressive transformer. This kind of model is good at creating images that look like they were taken from real life. The authors of the paper made HART better by adding a special part that helps it learn to make more detailed pictures. |
Keywords
» Artificial intelligence » Autoregressive » Diffusion » Tokenizer » Transformer
