Hello, I'm Jung Yi

I am an MS Student at KAIST CVLAB, advised by Seungryong Kim. I am interested in Efficient World Models, especially KV cache management in video world models. I received my BS from Yonsei University in 2024.

CV Scholar Github Twitter

News

  • May 2026: WorldKV, our paper on Efficient World Memory, has been released on arxiv.
  • May 2026: DeepForcing, our paper on Autoregressive Video World Model, has been accepted to ICML 2026.
  • Mar 2026: MV-TAP, our paper on Point Tracking, has been accepted to CVPR 2026.
  • Mar 2025: Chrono, our paper on Point Tracking, has been accepted to CVPR 2025.

Work Experience

Reactor
Research Scientist Intern Sep 2026 - Now
Project: Efficient Inference & Infrastructure for Video World Models. San Francisco, California
Kakao
Research Intern Mar 2026 - Jun 2026
Project: Real-time Talking Avatar with Autoregressive Video Diffusion. Pangyo, South Korea

Publications

WorldKV: Efficient World Memory with World Retrieval and Compression.

WorldKV: Efficient World Memory with World Retrieval and Compression.

Jung Yi, Minjae Kim, Paul Hyunbin Cho, Wooseok Jang, Sangdoo Yun, Seungryong Kim
arxiv, 2026

Efficient World Memory by combining World Retrieval and World Compression.

Project Page PDF arXiv
Deep Forcing: Training-Free Long Video Generation with Deep Sink and Participative Compression.

Deep Forcing: Training-Free Long Video Generation with Deep Sink and Participative Compression.

Jung Yi, Wooseok Jang, Paul Hyunbin Cho, Jisu Nam, Heeji Yoon, Seungryong Kim
ICML, 2026

Long-video generation by combining Deep Sink and Participative Compression.

Project Page PDF arXiv
Lip Forcing: Few-Step Autoregressive Diffusion for Real-time Lip Synchronization.

Lip Forcing: Few-Step Autoregressive Diffusion for Real-time Lip Synchronization.

Paul Hyunbin Cho*, Jinhyuk Jang*, SeokYoung Lee, Joungbin Lee, Siyoon Jin, Heeseong Shin, Jung Yi, Yunjin Park, Chulmin Park, Seungryong Kim
arxiv, 2026

Real-time lip synchronization using few-step autoregressive diffusion.

Project Page PDF arXiv
DiTracker: Repurposing Video Diffusion Transformers for Robust Point Tracking.

DiTracker: Repurposing Video Diffusion Transformers for Robust Point Tracking.

Soowon Son, Honggyu An, Chaehyun Kim, Hyunah Ko, Jisu Nam, Dahyun Chung, Siyoon Jin, Jung Yi, Jaewon Min, Junhwa Hur, Seungryong Kim
arxiv, 2025

Video Diffusion Model for Point Tracking.

Project Page PDF arXiv
MV-TAP: Tracking Any Point in Multi-View Videos

MV-TAP: Tracking Any Point in Multi-View Videos

Jahyeok Koo*, Ines Hyeonsu Kim*, Mungyeom Kim, Junghyun Park, Seohyeon Park, Jaeyeong Kim, Jung Yi, Seokju Cho, Seungryong Kim
CVPR, 2026

Multi-view point tracker that uses camera and cross-view attention.

Project Page PDF arXiv
Exploring Temporally-Aware Features for Point Tracking

Exploring Temporally-Aware Features for Point Tracking

Ines Hyeonsu Kim*, Seokju Cho*, Jiahui Huang, Jung Yi, Joon-Young Lee, Seungryong Kim
CVPR, 2025

Feature backbone specifically designed for point tracking with built-in temporal awareness.

Project Page PDF arXiv
Fast sun-aligned outdoor scene relighting based on tensorf

Fast sun-aligned outdoor scene relighting based on tensorf

Yeonjin Chang, Yearim Kim, Seunghyeon Seo, Jung Yi, Nojun Kwak
WACV, 2024

Outdoor scene relighting for Neural Radiance Fields (NeRF).

Project Page PDF arXiv