Zhuofan Xia

Zhuofan Xia

Auto-generated profile, based on publicly available information

Zhuofan Xia

Tsinghua University

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 November 2024
Verified
Book Chapter
Closed access
PDF file
DHDongchen Han
Dongchen Han
TYTianzhu Ye
Tianzhu Ye
YHYizeng Han
Yizeng Han
ZXZhuofan Xia
Zhuofan Xia
SPSiyuan Pan
Siyuan Pan
GHGao Huang
Gao Huang

22 November 2024
Verified
Book Chapter
Closed access
PDF file
YPYifan Pu
Yifan Pu
ZXZhuofan Xia
Zhuofan Xia
JGJiayi Guo
Jiayi Guo
DHDongchen Han
Dongchen Han
QLQixiu Li
Qixiu Li
XLXiu Li
Xiu Li

3 June 2025 in Neural Networks
Verified
Journal Article
Closed access
JLJiayun Li
Jiayun Li
YCYuxiao Cheng
Yuxiao Cheng
YLYiwen Lu
Yiwen Lu
ZXZhuofan Xia
Zhuofan Xia
YMYilin Mo
Yilin Mo
GHGao Huang
Gao Huang

11 August 2024 in arXiv
Verified
Preprint
Open access
PDF file
YPYifan Pu
Yifan Pu
ZXZhuofan Xia
Zhuofan Xia
JGJiayi Guo
Jiayi Guo
DHDongchen Han
Dongchen Han
QLQixiu Li
Qixiu Li
XLXiu Li
Xiu Li
This paper identifies significant redundancy in the query-key interactions within self-attention mechanisms of diffusion transformer models, particularly during the early stages of denoising diffusion steps. In response to this observation, we present a novel diffusion transformer framework incorporating an additional set of mediator tokens to engage with queries and keys separately. By modulating the number of mediator tokens during the denoising generation phases, our model initiates the denoising process with a precise, non-ambiguous stage and gradually transitions to a phase enriched with detail. Concurrently, integrating mediator tokens simplifies the attention module's complexity to a linear scale, enhancing the efficiency of global attention processes. Additionally, we propose a time-step dynamic mediator token adjustment mechanism that further decreases the required computational FLOPs for generation, simultaneously facilitating the generation of high-quality images within the constraints of varied inference budgets. Extensive experiments demonstrate that the proposed method can improve the generated image quality while also reducing the inference cost of diffusion transformers. When integrated with the recent work SiT, our method achieves a state-of-the-art FID score of 2.01. The source code is available at https://github.com/LeapLabTHU/Attention-Mediators.