Yuchi Wang

Yuchi Wang

Master Student of Artificial Intelligence

Lanco Lab

Peking University

Biography

Hi! I’m a second-year master’s student at the AAIS (Academy for Advanced Interdisciplinary Studies), Peking University. Currently, I am a member of Lanco Lab, lead by Prof. Xu Sun. Previously, I earned my Bachelor’s degree from the School of Data Science, Fudan University.

My research interests encompass (1) Multimodal learning, including visual understanding (e.g., image captioning), text-guided visual generation (e.g., image/video generation or editing, talking face generation), etc (2) Generative Models such as diffusion models and VAEs (3) LLMs (especially multimodal large language models) and their applications like embodied AI.

Interests
  • Multimodal learning
  • Generativa AI (AIGC)
  • Large language models
  • Diffusion models
Education

  • Master in Data Science, Peking University, 2025 (Expected)

    Lanco Lab; Advisor: Xu Sun; AAIS

  • BSc in Data Science, Fudan University, 2022

    School of Data Science; GPA Rank: 3/85

News

  • [2024.03] One paper accepted by NAACL 2024
  • [2024.01] One paper accepted by ICLR 2024
  • [2023.10] We release GAIA demo!
  • [2023.10] One paper accepted by FMDM@NeurIPS 2023
  • [2023.05] Starting internship at MSRA ML Group
  • [2022.09] Joining in Lanco Lab, Peking University

All Publications

[NAACL 2024] LaDiC: Are Diffusion Models Really Inferior to Autoregressive Counterparts for Image-to-Text Generation?
We explore the previously untapped advantages of diffusion models over autoregressive (AR) methods in image-to-text generation. Through meticulous design of a latent-based diffusion model tailored for captioning, we achieve comparable performance with some strong AR baselines.
Yuchi Wang, Shuhuai Ren, Rundong Gao, Linli Yao, Qingyan Guo, Kaikai An, Jianhong Bai, Xu Sun
PDF Cite Code
[NAACL 2024] LaDiC: Are Diffusion Models Really Inferior to Autoregressive Counterparts for Image-to-Text Generation?
[ArXiv] UniEdit: A Unified Tuning-Free Framework for Video Motion and Appearance Editing
We present UniEdit, a tuning-free framework that supports both video motion and appearance editing by harnessing the power of a pre-trained text-to-video generator within an inversion-then-generation framework.
Jianhong Bai, Tianyu He, Yuchi Wang, Junliang Guo, Haoji Hu, Zuozhu Liu, Jiang Bian
PDF Cite Code Demo Page
[ArXiv] PCA-Bench: Evaluating Multimodal Large Language Models in Perception-Cognition-Action Chain
We present PCA-Bench, a multimodal decision-making benchmark for evaluating the integrated capabilities of Multimodal Large Language Models (MLLMs).
Liang Chen, Yichi Zhang, Shuhuai Ren, Haozhe Zhao, Zefan Cai, Yuchi Wang, Peiyi Wang, Xiangdi Meng, Tianyu Liu, Baobao Chang
PDF Cite Code Dataset
[ICLR 2024] GAIA: Zero-shot Talking Avatar Generation
An inside project in Microsoft. We design a codec to disentangle each frame of talking face videos into motion and appearance representations and then curated a large-scale, high-quality dataset to train our diffusion-based GAIA model. The results demonstrate remarkable naturalness and scalability.
Tianyu He, Junliang Guo, Runyi Yu, Yuchi Wang, Jialiang Zhu, Kaikai An, Leyi Li, Xu Tan, Chunyu Wang, Han Hu, HsiangTao Wu, Sheng Zhao, Jiang Bian
PDF Cite Demo Page (Anonymous version)
[ICLR 2024] GAIA: Zero-shot Talking Avatar Generation
[FMDM@NeurIPS 2023] Towards end-to-end embodied decision making via multi-modal large language model: Explorations with gpt4-vision and beyond
We found that powerful multimodal LLM like GPT4-Vision makes End-to-End embodied decision making more possible than ever. Moreover, we propose a new benchmark called PCA-EVAL and a multi-agent cooperation framework HOLMES for evaluation.
Liang Chen, Yichi Zhang, Shuhuai Ren, Haozhe Zhao, Zefan Cai, Yuchi Wang, Peiyi Wang, Tianyu Liu, Baobao Chang
PDF Cite Code Dataset

Work Experience

 
 
 
 
 
Microsoft Research Asia (MSRA)
Research Intern
Microsoft Research Asia (MSRA)
May 2023 – Present Beijing, China
  • Excited to join the ML group, advised by Junliang Guo and Xu Tan.
  • During my internship up to now, I have primarily focused on generative learning, including talking avatar generation, video generation/editing.

Academic Service

Teaching Assistant

I am currently serving as a teaching assistant for the course Large Language Model in Decision Intelligence (PKU Class 2024 Spring). This course is tailored for undergraduates, aiming to provide them with a foundational understanding of large language models and effective strategies for their utilization.