I've always appreciated the different experiments on reviewing done by NeurIPS PCs over the years. However the ever increasing verbosity of rebuttals these days does call for ways to reduce length and the 1 page pdf from CVPR does not look that bad or obsolete.

Christian Wolf@chriswolfvision.bsky.social · 2d ago

Rebuttals and discussions need to be reduced an extreme minimal length. CVPR style, 1 single PDF page. No discussion. Else, human ACs and reviewers will have a disadvantage compared to the world of agents out there now.

It’s hot across Europe right now. Millions of us are feeling it - in cities, towns, and villages across the continent. When heatwaves hit, small actions matter ⬇️

Home

Tired: Google Summer of Code Wired: NeurIPS Summer of OpenReview Scrolling

Multiple #ECCV2026 outstanding reviewer awards at @valeoai.bsky.social! Congrats to A. Boulch, @abursuc.bsky.social, @anhquancao.bsky.social, N. Samet, @t-martyniuk.bsky.social, Y. Xu, to our visiting PhD student E. Karypidis, and to our alumnus @gillespuy.bsky.social!

European Conference on Computer Vision #ECCV2026@eccv.bsky.social · last wk.

Thank you to all our #ECCV2026 reviewers. Special thanks to the Outstanding Reviewers for their exceptional contributions! 👏 eccv.ecva.net/Conferences/...

How can AI better understand 3D driving scenes? 🚗 In this ELLIOT Research Spotlight 🔬, we highlight LOSC from @valeoai.bsky.social: a method for open-vocabulary LiDAR segmentation that helps assign meaning to 3D point clouds, beyond fixed object categories. Paper + code in comments 👇

The NAVSIM challenges were key to the progress of E2E driving models. But, we start to reach their limits and we need more complex challenges to advance further. The AlpaSim E2E Closed Loop Challenge is a major step. Happy to share that key dates, registration and leaderboard access are now up!

KE:SAI - Kyutai ELLIS Scalable Autonomous Intelligence@kesai.eu · 4w ago

The AlpaSim E2E Closed Loop Challenge 2026 is open for registration and leaderboard access in a soft-open period. Build AV policies for realistic closed-loop simulation, where each policy's own decisions shape future scenes, observations, and interactions.

I've been offline for a few days at a scout camp in a tent. I didn't have much FOMO (still stayed up to date with many things via my phone), but man, does it feel good to be back at my PC!

valeo.ai is heading to #ICML2026 and #ACL2026 with two papers: - diverse LLM decoding with LoRA-MCL, - counterfactual explanations for driving LLMs with DRIV-EX 🧵👇

35°C is normal, we used to just call that summer! Here's how many summers topped 32°C 70s 4 80s 2 90s 8 2000s 6 2010s all of them 2020s all of them And here's how many summers topped 35°C 70s 1 80s 0 90s 2 2000s 2 2010s 3 2020s 4 (out of 6 so far)

We explored the impact of variability sources in generative modeling. Turns out, we've been neglecting the error bars associated with training variability all along! We should aim to report results that we are sure of their scientific validity, instead of seed engineering!

Kyutai@kyutai-labs.bsky.social · 2mo ago

🎰 Welcome to the FID Lottery. We pulled the lever 25 times on the same machine. Identical diffusion model, identical ImageNet class-cond recipe, only the seed changed. The house paid out anywhere from 33.59 to 35.69 FID. A 2.1-point spread, pure luck. Step onto the floor 👇🧵

The #WorldEngine tech report is now up! It's our post-training framework to deal w/ scarcity of long-tail safety-critical scenarios. Post-training strategies for autonomous driving are rarely tackled and discussed in the open. We hope that this will open more this area. arxiv.org/abs/2606.19836

Andrei Bursuc@abursuc.bsky.social · 4mo ago

🌍WorldEngine is one of the most exciting projects in AD in the past years! It's a post-training framework tackling the scarcity of long-tail safety-critical scenarios by: mining -> 3DGS reconstruction and dynamic agents control w/ behavior world models -> RL post-training. Blog, code and data are up

Super impressed by how Chinese teams continue releasing super-strong open-weights models! By this time you would assume that private ones are reaching escape velocity thanks to the huge amount of user data and feedback they can harvest, and yet open-models still pop up.

What if you could turn any number of photos (3, 8, 15, or even 60) into one clean 3D surface (pts & mesh) with Flow Matching? Check out our new work, Surflo: Consistent 3D Surface Flow Model with Global State. 🧵 1/N 🔗https://anttwo.github.io/surflo/

I made and saw multiple tutorials on self-supervised learning in the past, but this video by @jbhuang0604.bsky.social on this is quite something! It goes from InfoMax to contrastive learning to self-distilation to masked autoencoders to JEPAs w/ planning, including LeWorldModel in only 37 min!

Jia-Bin Huang@jbhuang0604.bsky.social · 2mo ago

I didn't know what JEPA is, and at this point I am too afraid to ask ...😬 so I made a video tracing the last 30+ years of self-supervised learning, covering ideas from contrastive learning, distillation, masked modeling, JEPA, and world models. youtu.be/gVEr2cnDE_8

We're hiring a postdoc to push a moonshot that mixes world-modeling, unsupervised RL, and autonomy. More details here, and please help us get the word out! docs.google.com/document/d/1...

EMERGE Lab World Models / Unsupervised RL Postdoc

POSTDOCTORAL RESEARCHER WORLD MODELING / UNSUPERVISED RL EMERGE Lab The EMERGE lab at NYU is looking to hire a postdoc for an exciting moonshot project aimed at exploring the limits of learning in ...

docs.google.com

Just stumbled upon this short & intense post by Gabriel Synnaeve on Research, FOMO, strong team, focus: -whatever you do, research ideas fail 4/5 to 3/4 times -you can't hold too many active problems in your head -no remedy for FOMO, but focus & strong teams help syhw.github.io/index.html#f...

⏱️ 1 day left to apply to the ELLIS Summer School on Autonomous Driving 🚗 Let's meet our amazing speakers! Today we introduce Fatma Güney (Professor at Koç University in Istanbul, Türkiye) 📝 Deadline: Friday, June 12, 11:59 PM AoE pages.cvc.uab.es/ssadriving/

Interested in the ELLIS Summer School on Autonomous Driving? 🚗 Let's meet our speakers! Today we introduce @kashyap7x.bsky.social (Co-Founder & Chief Technology Officer at @kesai.eu) 📝 (New) Application deadline: June 12, 2026 pages.cvc.uab.es/ssadriving/

⏰🚨 𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗱𝗲𝗮𝗱𝗹𝗶𝗻𝗲 𝗲𝘅𝘁𝗲𝗻𝗱𝗲𝗱!! Still time to apply for the ELLIS Summer School on Autonomous Driving 🚗 📅 New deadline: Friday, June 12, 11:59 PM AoE More info and registrations 👉 pages.cvc.uab.es/ssadriving/

Andrei Bursuc@abursuc.bsky.social · 3mo ago

📣 ELLIS | ELLIOT Summer School on Autonomous Driving 🚗 Super happy to announce the hottest event this summer. Apply now! 📅 September 15-18, 2026 📍Barcelona, Spain 🌐 pages.cvc.uab.es/ssadriving/ @ellis.eu, ELLIS-Barcelona, CVC-Barcelona, @valeoai.bsky.social ai.bsky.social, @elliot-eu.bsky.social

⏱️ 4 days left to apply to the ELLIS Summer School on Autonomous Driving 🚗 Let's meet our speakers! Today we introduce Holger Caesar (Assistant Professor at TU Delft in the Netherlands) 📝 Application deadline: June 8, 2026 pages.cvc.uab.es/ssadriving/

⏱️ 6 days left to apply to the ELLIS Summer School on Autonomous Driving 🚗 Let's meet our speakers! Today we introduce @maxigl.bsky.social (Staff Research Scientist at Nvidia) 📝 Application deadline: June 8, 2026 pages.cvc.uab.es/ssadriving/

7 days left to apply to ELLIS Summer School on Autonomous Driving (SSAD 2026) in Barcelona!

ELLIOT Project@elliot-eu.bsky.social · 2mo ago

The schedule for the ELLIS Summer School on Autonomous Driving (SSAD 2026) is LIVE! 🏎️✨ 4 days of top-tier AI & robotics research in Barcelona, proudly supported by ELLIOT. ⏳ 7 days left to apply! Secure your spot: 🔗 pages.cvc.uab.es/ssadriving/ @ellis.eu @valeoai.bsky.social @uab.cat

Over the years models got bigger, tasks more diverse and harder, datasets larger, etc. However, the message between the lines in AZ’s talks remained the same: everyone, look really carefully at your data, so many questions and solutions come from there. #CVPRinParis

Full house at #CVPRinParis today. Ever suprised by how many familiar faces but also so many new ones gathered here. Kudos to @davidpicard.eurosky.social @vickykalogeiton.bsky.social @matthieucord.bsky.social & team for making this possible

Our recent research will be presented at #CVPR2026 @cvprconference.bsky.social! We’ll present nine papers in the main conference and we’ll be speaking in two workshops & one tutorial Here is a quick overview on the works we’ll be presenting in Denver: valeoai.github.io/posts/cvpr-2...

See you at CVPR@Paris 2026 next week!

ELLIS@ellis.eu · 2mo ago

CVPR@Paris 2026 🇫🇷 — June 1st, co-organised by ELLIS Unit Paris. A one-day local event ahead of CVPR, open to all. Oral & poster sessions for CVPR 2026, CVPR workshops & ICLR 2026 papers. 🔗 https://cvprinparis.github.io/CVPR2026InParis/

Thrilled to share that MIRO is accepted to ICML 2026 @icmlconf.bsky.social ! 🎉 By training on the reward scores, we can simply condition the model on high rewards at inference time to guarantee top-tier, aligned outputs. We’ve updated our paper with some additional results!

MIRO: MultI-Reward cOnditioned pretraining improves T2I quality and efficiency

The default paradigm of post-training text-to-image generators includes post-hoc selection of generated images, and subsequent training with one reward model to align the generator to the reward, typi...

arxiv.org

Nicolas Dufour@nicolasdufour.bsky.social · 9mo ago

We introduce MIRO: a new paradigm for T2I model alignment integrating reward conditioning into pretraining, eliminating the need for separate fine-tuning/RL stages. This single-stage approach offers unprecedented efficiency and control. - 19x faster convergence ⚡ - 370x less FLOPS than FLUX-dev 📉

Hello world from @kesai.eu! Here is a new non-profit open-science AI lab based in Tübingen and Paris pushing the frontiers of Physical AI and self-driving in particular. They have a top-notch founding team, an ambitious agenda and incredible supporters (the team behind @kyutai-labs.bsky.social).

Kashyap Chitta@kashyap7x.bsky.social · 3mo ago

Open-source fueled the LLM revolution, but Physical AI hasn't fully benefited from this flywheel yet. Today, we're launching kesai.eu, our mission to democratize robotics research! First milestone: training a frontier-level self-driving policy using significantly less data than typically required.

py123d: @danieldauner.bsky.social et al. did all the crazy and dirty work to unify the highly heterogeneous autonomous driving datasets into a single efficient data format. nuscenes + nuplan + Waymo + Physical AI AV, etc., are all there. This is how you accelerate open-source AD research!

Kashyap Chitta@kashyap7x.bsky.social · 3mo ago

pip install py123d 3,300 hours of multimodal driving data at your fingertips (Physical AI AV + nuPlan + … + can add custom datasets)! 📄 Paper: arxiv.org/abs/2605.08084 ⭐️ GitHub: github.com/kesai-labs/py123d Monumental effort led by @danieldauner.bsky.social to accelerate open-source AV research.