Byungwoo Jeon (@rootyjeon) profile picture

Ph.D. student @kaist_ai

Seoul, Korea
Joined February 2026
Photo by Byungwoo Jeon (@rootyjeon)Photo by Byungwoo Jeon (@rootyjeon)Photo by Byungwoo Jeon (@rootyjeon)Photo by Byungwoo Jeon (@rootyjeon)Photo by Byungwoo Jeon (@rootyjeon)Photo by Byungwoo Jeon (@rootyjeon)Photo by Byungwoo Jeon (@rootyjeon)Photo by Byungwoo Jeon (@rootyjeon)
Byungwoo Jeon retweeted
LARC: Latent Reasoning in Continuous Space Instead of reasoning in discrete token space, LARC reasons over continuous hidden states, boosting image generation quality and enabling test-time scaling. SOTA on GenEval (+7% over BAGEL).
4
29
2
146
7,013
Robots still need teleoperated demos for each new task. Human videos of the task are cheap, but visual and action gaps make them hard for robots to learn from. 👉 RoboTryOn uses virtual try-on to turn human videos into robot demonstrations. 🌐 ryan-seungyong-lee.github.io… 🧵 [1/8]
7
10
26
1,127
LLMs get better the longer they think. Unified multimodal models don't. Why? 🌐 rootyjeon.github.io/larc-umm… 📍 NeurIPS 2026 Introducing LARC: Latent Reasoning in Continuous Space. Reasoning moves into the model's hidden states, so image generation finally scales at test time. 🧵
3
27
142
23,043
LARC improves text-to-image generation, editing and visual reasoning over its base model BAGEL and text-reasoning UMMs. It is the first reasoning-augmented UMM with clear test-time scaling: more latent steps, monotonically better images, ahead of best-of-N at equal compute.
1
185
📄 arxiv.org/abs/2610.42149 💻 github.com/rootyJeon/LARC 🤗 huggingface.co/papers/2610.4… w/ @yoonwoojeong , @hyunseok_i and co-advisors Minsu Cho & @JinwooShin0417 @KAIST_AI · @postech1986 · @RLWRLD_ai 🙏 🟢 This work was supported by @nvidia Academic Grant Program.
4
180
Byungwoo Jeon retweeted
VLAs move like the teleoperators who trained them: slowly, even when they don't need to. Introducing PaceVLA 🤖⚡, a VLA that picks, per action chunk, how coarsely and how far ahead to act. Up to 1.8× faster, same success. 📄 arxiv.org/abs/2610.25961 🌐 pacevla.github.io/ 🧵
4
23
5
110
17,956
Byungwoo Jeon retweeted
A recent line of WAMs generates actions from VDM hidden states of a single denoising step, yet their quality gets little attention. 👉 We introduce FRED to improve them via self-distillation, with large gains on real robot experiments. 🌐 huiwon-jang.github.io/FRED/ 🧵 [1/7]
3
19
1
40
1,594
Byungwoo Jeon retweeted
複数カメラの映像で物体の動きを一致させる「Video-Point Model(VPM)」🤖 🔹NVIDIA Cosmos 3を基盤に開発 🔹3D空間での動きを学習し、シミュレーション映像の品質指標FVDを14〜19%改善 🔹ロボットのゼロショット操作も実証 開発者:Sunghwan Hong氏ら 論文・データ・コード公開中。詳細はリプ欄。
2
3
1
54
3,678
Can explicitly learning future point motion improve the model that generates those future images? We introduce VPM (Video-Point Model), a 4D world model that jointly forecasts future multi-view videos + persistent 3D point trajectories. Short answer: yes. Teaching a video world model to forecast 3D point tracks makes its videos more faithful & 3D-consistent, and makes robots better. 🌐 Project: rootyjeon.github.io/video-po… 🧵 [1/6]
3
23
3
90
15,845
Robots too. Add actions as a token modality and the point-supervised backbone gives the best success rates on LIBERO, RoboCasa and a real Franka arm: the tracks describe the robot's own future motion, densely and in 3D. Point track is highly correlated with robot actions. 🧵 [5/6]
1
1
176
📄 arXiv: arxiv.org/abs/2610.28594 💻 Code: github.com/rootyJeon/VPM 🤗 HF: huggingface.co/collections/r… Co-led w/ Sunghwan Hong. Big-shout out to my advisor, Jinwoo Shin @JinwooShin0417. This project would not have been possible without my amazing collaborators Dohyeon Kim, Jonghoon Lee, Jungwoo Park, Inhee Lee, Doosan Baek, Jinil Kim, Taehyun Lim, Hanbyul Joo @jhugestar, Marc Pollefeys @mapo1, and Saining Xie @sainingxie. KAIST · ETH Zürich · SNU · NYU · RLWRLD · Physics Sim Lab 🧵 [6/6]
2
233
Byungwoo Jeon retweeted
🏆1st place in the AI Scientist track at Ralphthon @ ICML2026 — $10,000 in OpenAI credits. @happyhappy_jun & @WoominSong from @KAIST_AI We've open-sourced all of it: harness, code, the full log of research traces. See our post: byungjunyoon.ai/writing-driv… How we did it 🧵
6
18
41
5,006
Byungwoo Jeon retweeted
SpatialBoost enhances vision encoders with 3D spatial awareness A framework that injects 3D spatial knowledge into pre-trained vision encoders via language-guided reasoning, boosting DINOv3 performance by 3.8% on ADE20K and advancing 3D perception tasks.
1
11
58
5,248