Zineng Tang

I am a first year Ph.D student at BAIR (Berkeley Artificial Intelligence) working with Alane Suhr. Previously, I had wonderful experience working with Mohit Bansal at UNC-NLP, MURGe-Lab and Ziyi Yang at Microsoft. I did my undergrad at UNC Chapel Hill!

Email / CV / Google Scholar / Twitter / Github

Research

My primary research interests lie in the area of multi-modal learning, natural language processing, and machine learning.

	CoDi: Any-to-Any Generation via Composable Diffusion Zineng Tang, Ziyi Yang, Yang Liu, Chenguang Zhu, Michael Zeng, Mohit Bansal Arxiv, 2023 Project Page / github / arXiv We built a framework for any to any modality mapping.
	Unifying Vision, Text, and Layout for Universal Document Processing Zineng Tang, Ziyi Yang, Guoxin Wang, Yuwei Fang, Yang Liu, Chenguang Zhu, Michael Zeng, Cha Zhang, Mohit Bansal CVPR, 2023 (Highlight; 2.5% acceptance rate) github / arXiv We built a unified framework for document processing.
	TVLT: Textless Vision-Language Transformer Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal NeurIPS*, 2022 (Oral; 1.76% acceptance rate) github / arXiv We built a textless vision-language transformer with a minimalist design.
	Vidlankd: Improving language understanding via video-distilled knowledge transfer Zineng Tang, Jaemin Cho, Hao Tan, Mohit Bansal NeurIPS, 2021 github / arXiv We built a teacher-student transformer for visually grounded language learning.
	Decembert: Learning from noisy instructional videos via dense captions and entropy minimization Zineng Tang, Jie Lei, Mohit Bansal NAACL, 2021 github / paper We built a video-language transformer that addresses the issues of noisy ASR data by dense captions and entropy minimization.
	Continuous language generative flow Zineng Tang, Shiyue Zhang, Hyounghun Kim, Mohit Bansal ACL, 2021 github / paper We built a language framework basde on continuous generative flow.
	Dense-caption matching and frame-selection gating for temporal localization in VideoQA Hyounghun Kim, Zineng Tang, Mohit Bansal ACL, 2020 github / arxiv We built a video QA framework basde on various techniques like dense captions.