jshan

- Location: Exhibit Hall C, D, E — Poster #304 - More Info: lnkd.in/gCGiMrKw Hope to see you there—looking forward to meeting researchers working on #ReinforcementLearning and #Planning!

If you’re interested in offline RL, goal-conditioned agents, or temporally abstract reasoning, I’d be glad to connect. - Poster Title: Option-aware Temporally Abstracted Value for Offline Goal-Conditioned Reinforcement Learning - Session Time: Fri, Dec 5 — 11 a.m. to 2 p.m. PST

I’ll be presenting at #NeurIPS2025 Spotlight in San Diego next week! Our work explores how value learning can remain stable and performant in long-horizon tasks by introducing temporal abstraction directly into the temporal-difference learning process.

@jiazhi_yang2024 @VilleKuosmanen +1 I’m also curious

@VilleKuosmanen curious how to assign the progress label for failure data.

more examples on the reward model - while the start and end state look the same, the trajectories look very different which should allow us to tell the difference between the two the reward model was trained with both success and failure data

#RSS2025 #rss2025

3/ Real‑world performance Evaluated on 9 real domains with novel objects: 78.9 % success rate despite training entirely in sim 📦🤖 Detailed evaluation in the paper.

🦾 Want home robots that manipulate everyday objects? They should topple, roll, slide—not just grasp. But current non‑prehensile policy fails in diverse envs (stair, cabinet). Meet HAMNet Accepted #RSS2025, @RoboticsSciSys roboticsconference.org/program/papers… 🧵👇

@iScienceLuvr Yes please! I would like to join

Is there an RL group chat where people talk about latest methods, papers, practitioner advice, etc.? If not maybe I should make one?

More results on the website. 
🌐 unicorn-hamnet.github.io Code & benchmark open-sourcing soon! This work wouldn't have been possible without amazing collaborators: Yoonyoung Cho, @Junhyekh, Jisu Han, and Beomjoon Kim. #robotics #manipulation

@XingdongZ @philfung @huggingface @LeRobotHF Sure! 👍

@jisu_han__ @philfung @huggingface @LeRobotHF Thanks so much Jisu!! Definitely would love to collaborate together next time Let's keep in touch

Just joined the LeRobot Hackathon Seoul district— and honestly, there’s something about building things with others who are genuinely curious and kind 🤖✨ Our team made a simple robot that can catch mosquitoes (world-wide enemy🦟) How about your team? @huggingface @LeRobotHF

WraspRobot, a bug swatter that made its way into the top 10 finalists! x.com/jisu_han__/sta…

@XingdongZ @philfung @huggingface @LeRobotHF Thanks, Xingdong! Our team was busy building everything from scratch, so we didn’t get much time to chat. I was really impressed how your team trained on datasets for language practicality. Would love to team up with you sometime, maybe at another hackathon or other chance! 😊

@jisu_han__ @philfung @huggingface @LeRobotHF What a creative idea!! Huge congrats for the amazing work Jisu! Unfortunately couldn't got a chance to chitchat in the hackathon. Up up for the winner ranking!!

@Ash7Win10 @_1nF0rmed_ @LeRobotHF Thanks a lot!

@jisu_han__ @_1nF0rmed_ @LeRobotHF @jisu_han__ here's our PR github.com/huggingface/le… , still a bit rough around the edges , but it has both teleop , recording and inference code for bimanual mode :)

Here is a better demonstration video of the 2 so-101 arm collaboration for the @LeRobotHF hackathon @Ash7Win10

@partabot_ @huggingface @LeRobotHF Thanks for your kind comment! We considered multiple choice (e.g., companion robot) but we narrowed down to a robot that really mattered in our situation. Summer in Seoul is quite tough due to the fight with mosquitoes, so we all agreed to that topic and proceeded.

@jisu_han__ @huggingface @LeRobotHF Looks like I will take my so-arm with me on my trip across south asia this summer!! I’m curious on how you came up with the idea?? 😃