Sean Sung-Wook Lee

PhD Student @ UVA Computer Science
Let's summon the robot ghosts!

[Sean Sung-Wook Lee @seanswlee]

@YouJiacheng Actually looks extremely scaling-pilled if you do -log(100-val)

[Sean Sung-Wook Lee @seanswlee]

@seohong_park Great read! How are you addressing the geometric symmetry of the cube when you are feeding the state inputs? Does anything change if you represent orientation as (roll-axis re-ordered from largest to smallest, roll % 90 deg) instead of quaternion (presumably what you are using)?

[Sean Sung-Wook Lee @seanswlee]

Finally, I would like to say thanks to all my amazing collaborators for making this work possible! @JoshuaK78925 @yenling_kuo

[Sean Sung-Wook Lee @seanswlee]

A bit of throwback from the last year, but we presented CLASS in CoRL 2025, an action-centric representation learning framework that induces policy invariance to useful properties like camera viewpoints, object visuals, etc.

[Sean Sung-Wook Lee @seanswlee]

If you find our work interesting, please try it out! Also, feel free to reach out if you talk more in this direction. We hope that this work opens up more opportunities toward learning strong action-centric vision representations.
Project webpage: https://t.co/7V5R1hb0cY