Kunchang Li

@likunchang1998

Intern @BytedanceTalk. Ph.D. student in video understanding and multi-modality learning.

Shanghai, China

andy1621.github.io

Joined January 2022

101 Following

500 Followers

Replies Reposts Media

Kunchang Li
@likunchang1998
May 21, 2025
We're happy to release BAGEL, a powerful unified model: 🔓 Open-source: weights & code 🧠 Unified MoT arch for text, image, & video 🏆 Top performance on 10+ benchmarks ✨ Emerging properties: From basic T2I/I2T to intelligent manipulation ⚡ Try it now: bagel-ai.org
6
Kunchang Li
@likunchang1998
Dec 13, 2024
Check our lastest VLN paper: ✨ Fully-automatic VLN Data Self-Refining Flywheel via multi-agent (navigator+generator) collaboration and mutual improvement 🚀 𝐒𝐎𝐓𝐀 results on 8 VLN task, even surpassing/approaching 𝐇𝐔𝐌𝐀𝐍 performance Check details ⬇️
Zun Wang
@ZunWang919
Dec 12, 2024
🚨Excited to introduce the Self-Refining Data Flywheel (SRDF) for generating high-quality and large-scale navigational instruction-trajectory pairs, achieving SoTA on 8 VLN tasks: ➡️ Automatic, Iterative Data Loop for Self-Improvement: Navigator refines instruction generator via
4
Kunchang Li
@likunchang1998
Mar 12, 2024
Thank @_akhaliq for sharing our work! All the code and models can be found at github.com/OpenGVLab/Vide…
AK
@_akhaliq
Mar 12, 2024
VideoMamba State Space Model for Efficient Video Understanding Addressing the dual challenges of local redundancy and global dependencies in video understanding, this work innovatively adapts the Mamba to the video domain. The proposed VideoMamba overcomes the limitations of
1
Kunchang Li
@likunchang1998
Feb 27, 2024
#CVPR2024 Excited to share that Vlogger is accepted to @CVPR
Kunchang Li
@likunchang1998
Jan 22, 2024
🌟 Introducing Vlogger: AI storytelling evolved! Using ShowMaker, our robust short-video model, Vlogger crafts engaging, minute-long narratives. Step into the future of AI vlogging! 🎥🚀 📄 Paper: arxiv.org/abs/2401.09414 💻 Code: github.com/zhuangshaobin/…
Kunchang Li
@likunchang1998
Feb 27, 2024
#CVPR2024 Excited to share that MVBench is accepted to @CVPR
Kunchang Li
@likunchang1998
Nov 29, 2023
Delighted to unveil MVBench📽️: 20 challenging tasks for MLLM's temporal comprehension.🤔 We also offer VideoChat2, a solid baseline model excelling in 15 tasks!🏆 Paper📄: arxiv.org/abs/2311.17005 Code & Model & Data & Benchmark🥳: github.com/OpenGVLab/Ask-… 🧵

Kunchang Li

Kunchang Li’s posts