TerDiT: Ternary Diffusion Models with Transformers

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Lu, Xudong, Zhou, Aojun, Lin, Ziyi, Liu, Qi, Xu, Yuhui, Zhang, Renrui, Yang, Xue, Yan, Junchi, Gao, Peng, Li, Hongsheng
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909567744475136
author Lu, Xudong
Zhou, Aojun
Lin, Ziyi
Liu, Qi
Xu, Yuhui
Zhang, Renrui
Yang, Xue
Yan, Junchi
Gao, Peng
Li, Hongsheng
author_facet Lu, Xudong
Zhou, Aojun
Lin, Ziyi
Liu, Qi
Xu, Yuhui
Zhang, Renrui
Yang, Xue
Yan, Junchi
Gao, Peng
Li, Hongsheng
contents Recent developments in large-scale pre-trained text-to-image diffusion models have significantly improved the generation of high-fidelity images, particularly with the emergence of diffusion transformer models (DiTs). Among diffusion models, diffusion transformers have demonstrated superior image-generation capabilities, boosting lower FID scores and higher scalability. However, deploying large-scale DiT models can be expensive due to their excessive parameter numbers. Although existing research has explored efficient deployment techniques for diffusion models, such as model quantization, there is still little work concerning DiT-based models. To tackle this research gap, we propose TerDiT, the first quantization-aware training (QAT) and efficient deployment scheme for extremely low-bit diffusion transformer models. We focus on the ternarization of DiT networks, with model sizes ranging from 600M to 4.2B, and image resolution from 256$\times$256 to 512$\times$512. Our work contributes to the exploration of efficient deployment of large-scale DiT models, demonstrating the feasibility of training extremely low-bit DiT models from scratch while maintaining competitive image generation capacities compared to full-precision models. Our code and pre-trained TerDiT checkpoints have been released at https://github.com/Lucky-Lance/TerDiT.
format Preprint
id arxiv_https___arxiv_org_abs_2405_14854
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle TerDiT: Ternary Diffusion Models with Transformers
Lu, Xudong
Zhou, Aojun
Lin, Ziyi
Liu, Qi
Xu, Yuhui
Zhang, Renrui
Yang, Xue
Yan, Junchi
Gao, Peng
Li, Hongsheng
Computer Vision and Pattern Recognition
Machine Learning
Recent developments in large-scale pre-trained text-to-image diffusion models have significantly improved the generation of high-fidelity images, particularly with the emergence of diffusion transformer models (DiTs). Among diffusion models, diffusion transformers have demonstrated superior image-generation capabilities, boosting lower FID scores and higher scalability. However, deploying large-scale DiT models can be expensive due to their excessive parameter numbers. Although existing research has explored efficient deployment techniques for diffusion models, such as model quantization, there is still little work concerning DiT-based models. To tackle this research gap, we propose TerDiT, the first quantization-aware training (QAT) and efficient deployment scheme for extremely low-bit diffusion transformer models. We focus on the ternarization of DiT networks, with model sizes ranging from 600M to 4.2B, and image resolution from 256$\times$256 to 512$\times$512. Our work contributes to the exploration of efficient deployment of large-scale DiT models, demonstrating the feasibility of training extremely low-bit DiT models from scratch while maintaining competitive image generation capacities compared to full-precision models. Our code and pre-trained TerDiT checkpoints have been released at https://github.com/Lucky-Lance/TerDiT.
title TerDiT: Ternary Diffusion Models with Transformers
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2405.14854