Peitian Zhang (张配天)
I'm a 3rd-year master student at Renmin University of China, advised by Zhicheng Dou.
I focus on search models and large language model applications.
News
-
Our paper "Video-XL: Extra-Long Vision Language Model for Hour-Scale Video Understanding" has been accepted to CVPR2025.
-
Our paper "Long Context Compression with Activation Beacon" has been accepted to ICLR2025.
-
Our paper "MemoRAG: Moving towards Next-Gen RAG via Memory-Inspired Knowledge Discovery" has been accepted to WWW2025 as a poster presentation.
Education
-
M.E. Artificial Intelligence, Renmin University of China (RUC)
-
B.E. Computer Science, Renmin University of China (RUC)
Awards
-
JingDong Special Scholarship
-
Renmin University Excellent Graduate
Projects
- BAAI General Embedding (BGE) Core Contributor.
- Foreign Law Large Language Model Leader of Retrieval Module.
Selected Research
Video-XL: Extra-Long Vision Language Model for Hour-Scale Video Understanding
[Paper] [Code]
Yan Shu, Zheng Liu, Peitian Zhang, Minghao Qin, Junjie Zhou, Zhengyang Liang, Tiejun Huang, Bo Zhao
CVPR, 2025
[Paper] [Code]
Yan Shu, Zheng Liu, Peitian Zhang, Minghao Qin, Junjie Zhou, Zhengyang Liang, Tiejun Huang, Bo Zhao
CVPR, 2025
Enabling high-quality and efficient video understanding over thousands of frames on a single A100 GPU.
MemoRAG: Moving towards Next-Gen RAG via Memory-Inspired Knowledge Discovery
[Paper] [Code]
Hongjin Qian, Peitian Zhang, Zheng Liu, Kelong Mao, Zhicheng Dou
WWW, 2025
[Paper] [Code]
Hongjin Qian, Peitian Zhang, Zheng Liu, Kelong Mao, Zhicheng Dou
WWW, 2025
Enhancing retrieval-augmented generation (RAG) by acquiring global understanding of the entire database from a memory module.
Long Context Compression with Activation Beacon
[Paper] [Code]
Peitian Zhang, Zheng Liu, Shitao Xiao, Ninglu Shao, Qiwei Ye, Zhicheng Dou
ICLR, 2025
[Paper] [Code]
Peitian Zhang, Zheng Liu, Shitao Xiao, Ninglu Shao, Qiwei Ye, Zhicheng Dou
ICLR, 2025
A plug-in module for transformer-based LLMs that enables effective, efficient, and flexible compression of long contexts.