Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909220499095552 |
|---|---|
| author | Sun, Peize Jiang, Yi Chen, Shoufa Zhang, Shilong Peng, Bingyue Luo, Ping Yuan, Zehuan |
| author_facet | Sun, Peize Jiang, Yi Chen, Shoufa Zhang, Shilong Peng, Bingyue Luo, Ping Yuan, Zehuan |
| contents | We introduce LlamaGen, a new family of image generation models that apply original ``next-token prediction'' paradigm of large language models to visual generation domain. It is an affirmative answer to whether vanilla autoregressive models, e.g., Llama, without inductive biases on visual signals can achieve state-of-the-art image generation performance if scaling properly. We reexamine design spaces of image tokenizers, scalability properties of image generation models, and their training data quality. The outcome of this exploration consists of: (1) An image tokenizer with downsample ratio of 16, reconstruction quality of 0.94 rFID and codebook usage of 97% on ImageNet benchmark. (2) A series of class-conditional image generation models ranging from 111M to 3.1B parameters, achieving 2.18 FID on ImageNet 256x256 benchmarks, outperforming the popular diffusion models such as LDM, DiT. (3) A text-conditional image generation model with 775M parameters, from two-stage training on LAION-COCO and high aesthetics quality images, demonstrating competitive performance of visual quality and text alignment. (4) We verify the effectiveness of LLM serving frameworks in optimizing the inference speed of image generation models and achieve 326% - 414% speedup. We release all models and codes to facilitate open-source community of visual generation and multimodal foundation models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_06525 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation Sun, Peize Jiang, Yi Chen, Shoufa Zhang, Shilong Peng, Bingyue Luo, Ping Yuan, Zehuan Computer Vision and Pattern Recognition We introduce LlamaGen, a new family of image generation models that apply original ``next-token prediction'' paradigm of large language models to visual generation domain. It is an affirmative answer to whether vanilla autoregressive models, e.g., Llama, without inductive biases on visual signals can achieve state-of-the-art image generation performance if scaling properly. We reexamine design spaces of image tokenizers, scalability properties of image generation models, and their training data quality. The outcome of this exploration consists of: (1) An image tokenizer with downsample ratio of 16, reconstruction quality of 0.94 rFID and codebook usage of 97% on ImageNet benchmark. (2) A series of class-conditional image generation models ranging from 111M to 3.1B parameters, achieving 2.18 FID on ImageNet 256x256 benchmarks, outperforming the popular diffusion models such as LDM, DiT. (3) A text-conditional image generation model with 775M parameters, from two-stage training on LAION-COCO and high aesthetics quality images, demonstrating competitive performance of visual quality and text alignment. (4) We verify the effectiveness of LLM serving frameworks in optimizing the inference speed of image generation models and achieve 326% - 414% speedup. We release all models and codes to facilitate open-source community of visual generation and multimodal foundation models. |
| title | Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2406.06525 |