Huge congratulations to my phenomenal graduate student, Ziyi Wu, for completing his PhD at the University of Toronto. 🎉 Ziyi has done some of the pioneering work on endowing generative models for image and video generation with fine-grained control and better alignment. 1/n

If LLMs are great at generating code to control robots, their best use is probably helping generate controllers and, subsequently, data for pre-training robot models. Similar to how they currently help interface with game/physics engines used to help train video models.

Today was the last day of our brilliant research coordinator, Jazmin Diaz, at the University of Toronto. She kept @davelindell.bsky.social's lab and mine running smoothly. We can't imagine what things would be like without her. She is leaving giant shoes to fill and will be dearly missed. 1/2

📢 We are looking for Postdocs at @vectorinstitute.ai Join us to work with my group or any of our other brilliant researchers. Apply by Aug 31 to receive full consideration for the fall cohort: workforcenow.adp.com/mascsr/defau...

This is a robot failing to grasp a ball. Almost every robot lab produces clips like this daily… and almost all of them get thrown away. This is the most abundant but underused resource in robot learning. We’re collecting all of it now as “OopsieData”, please join us at oopsie-data.com! (1/13)

A new method from @utoronto.ca computer scientists trains AI faster and more reliably using less data and computing power. This could help bring smarter robots and autonomous systems into everyday life. Learn more: uoft.me/cyu @cvoelcker.bsky.social @igilitschenski.bsky.social

‘Nobel Prize of mathematics’: #UofT mathematician Jacob Tsimerman awarded prestigious Fields Medal 🏅 uoft.me/jacob

I am optimistic on 3d for robotics. It matters for policy learning, as some sensing modalities are inherently 3d; for HRI because words are not always enough; for scenario design in simulation & testing; and for bootstrapping training data.

At #ICLR2026 presenting our first poster in the morning on Relative Entropy Pathwise Policy Optimization. Stop by at #4613. 🧑‍🎓 @cvoelcker.bsky.social, @axelbrunnbauer.bsky.social, Michal Naumann, Pieter Abbeel, @ericeaton.bsky.social, Radu Grosu, @sologen.bsky.social @igilitschenski.bsky.social

Excited to share that the code and weights for our #NeurIPS2025 relighting (UniRelight) and lighting estimation (LuxDiT) works are now public. 📢 Code 🧑‍💻 💡 github.com/nv-tlabs/Uni... 📷 github.com/nv-tlabs/Lux... Weights 🏋️‍♀️ 💡 huggingface.co/nvidia/UniRe... 📷 huggingface.co/nvidia/LuxDiT

Enjoyed my trip to Philadelphia to speak at @upenn.edu GRASP Lab Robotics Seminar this week. Thank you, Kostas Daniilidis, for hosting me. It was wonderful meeting with Antonio Loquercio, Rachel Holladay, Dinesh Jayaraman, Pratik Chaudhari, Lingjie Liu, and many of your brilliant mentees.

Every year around this time, I wish for introducing a matching procedure for CS grad school as in medical residency. This would be a mental health improvement for everyone involved. How are doctors better at this than computer scientists? Where is our occupational pride?

Had a great time speaking at Stanford's Center for Image System Engineering and Vision seminars over the last two weeks. 👨‍🏫 Thank you for hosting me, Gordon Wetzstein and Wenlong Huang.

🚀 Excited to share REPPO, a new on-policy RL agent! TL;DR: Replace PPO with REPPO for fewer hyperparameter headaches and more robust training. REPPO, led by @cvoelcker.bsky.social, will be presented at ICLR 2026. How does it work? 🧵👇

Given the diverse work on world models, there is a debate on where to draw the line between world models and other dynamic systems models. I'd argue that, like LLMs for general-purpose language modelling, world models are not designed for a single specific task or phenomenon.

2025/2026 may be remembered as the years when many grad students "stopped" and many professors "resumed" programming.

I'll be at #NeurIPS2025 in San Diego from Thu to Sat, and I am looking for PostDocs in Embodied AI, particularly in world modeling and simulator learning. Please reach out if you are interested.

I'm looking for graduate students to join my group in fall 2026. We work at the intersection of Computer Vision, Deep Learning, and Robotics. The goal of our work is to create and understand organic 🍀 simulation systems, i.e., controllable data-curation engines built from real-world data. 1/n

Together with our brilliant graduate researcher Ziyi Wu and our fantastic undergrad, Brayden Zhang, we had a wonderful time at #ICCV2025!

Interesting @anthropic.com work on LLM poisoning. I'd argue that the glass is half full, and this can be interpreted as a positive result: Regardless of model size, the same number of samples seems to be sufficient to learn a new "skill". www.anthropic.com/research/sma...

A small number of samples can poison LLMs of any size

Anthropic research on data-poisoning attacks in large language models

anthropic.com

Excited to have been appointed as a faculty member at @vectorinstitute.ai. 🎉 This would not have been possible without the amazing work of my collaborators and the brilliant students in my group within @uoftcompsci.bsky.social. We all look forward to deepening ties with the Vector community. ❤️

Want to help train the next generation of leaders, researchers, and innovators in CS? We are looking for teaching stream faculty to join us at @uoftcompsci.bsky.social. Check out our job posting and join a team of brilliant colleagues and amazing students. 🇨🇦❤️💻 academicjobsonline.org/ajo/jobs/30410

Proud German advisor moment: Discovered a pair of Birkenstocks under a grad student's desk.

Would you be surprised to learn that many empirical implementations of value-aware model learning (VAML) algos, including MuZero, lead to incorrect model & value functions when training stochastic models 🤕? In our new @icmlconf.bsky.social 2025 paper, we show why this happens and how to fix it 🦾!

Thrilled to share the work that our lab will present at @cvprconference.bsky.social. Check out the papers and meet @kai-he.bsky.social, @yashkant.bsky.social, @dazitu616.bsky.social, and Toshiya Yura in Nashville at their poster sessions!

Ai4Science is doing a great job at applying AI to accurately model physics for one (or a few) phenomena at a time. Video models sacrifice such accuracy in favour of huge diversity. The big challenge for robot learning researchers is to strike a balance between these two.

In Germany, there is a tradition of creating funny hats for doctoral graduates. 🎓 @cvoelcker.bsky.social brought this tradition to my group and, together with Umangi Jain, spearheaded the construction of a masterpiece for our first PhD graduate, @ashmrz.bsky.social. 1/2

Compared to huge budget cuts, barring Harvard from enrolling any international students may seem unimportant. The opposite is true. Graduate students are the primary generators of research. Staying world-class requires the world's brightest minds. www.cnn.com/2025/05/22/u...

Today is one of the most exciting days of my academic career: my first PhD student graduated. 🎉 Huge congrats, @ashmrz.bsky.social! 🥂 I am grateful for the opportunity to have worked with you and could not have wished for a better start. 🚀 1/4

Great meeting @sramamoorthy.bsky.social and David Richardson from @edinburgh-uni.bsky.social