Celo2: Towards Learned Optimization Free Lunch Learned optimizers are powerful alternatives to hand-designed update rules like Adam, yet they have seen limited practical adoption since they often fail to meta-generalize beyond their training distr...

All paper links

Celo2: arxiv.org/abs/2602.19142
Celo (its predecessor): arxiv.org/abs/2501.12670

I'll also be presenting these two papers at #ICLR26 main track in 3 weeks at Rio de Janeiro, Brazil 🇧🇷 feel free to reach out if you'd like to chat more in-person!

Our work on building world models from episodic memories is accepted to #ICLR26!
Led by (then undergraduate, now masters student) @herbiehe.bsky.social
Check out the tweeprint!

New paper 🚨 #ICLR26

Most world models predict the future from a past trajectory. But neuroscience suggests that such inference can instead be made from temporally independent experiences.

We built the Episodic Spatial World Model (ESWM), a model that does exactly this:

Video abstract [1/2]

✨Two single author papers accepted to ICLR 2026!✨

Truly excited to present these results at #ICLR2026 !
@iclr-conf.bsky.social #ICLR26 #ReinforcementLearning

✨Two single author papers accepted to ICLR 2026!✨

Truly excited to present these results at #ICLR2026 !

@iclr-conf.bsky.social #ICLR26 #DeepRL #ICLR #ReinforcementLearning

Workshop Making Spatial Reasoning Systems More Adaptive, Robust, and Efficient for Real-World Deployment.

🚨Announcing our #ICLR26 Workshop, The First Workshop on Efficient Spatial Reasoning (ES-Reasoning) @iclr-conf.bsky.social

🚀 Call for Papers!!!
📅 Deadline: 2026/02/09, 11:59PM AOE
📣 AC notification: 2026/03/01

🌐Learn more at our Website: sites.google.com/ucsd.edu/eff...

📢The second edition of ✨GRaM workshop✨ is here this time at #ICLR26.

🌟Submit your exciting works in Geometry-grounded representations.

We welcome submissions in multiple tracks i.e.
📄 Proceedings
📝extended abstract
👩‍🏫Tutorial/blogpost
as well as an exciting challenge!