Very interesting, standard attention causes vanishing gradient due to most prob being very small after some training. LASER tackles this by pushing the attention operation on exponential space. i.e., exp_output = sm(QK^T) exp(V) They dont seem to exaggerate on the performance

📢 3DV 2026 – Call for Papers is Out! 📝 Paper Deadline: Aug 18 🎥 Supplementary: Aug 21 🔗 3dvconf.github.io/2026/call-for-… 📅 Conference Date: Mar 20–23, 2026 🌆 Location: Vancouver 🇨🇦 🚀 Showcase your latest research to the world! #3DV2026 #CallForPapers #Vancouver #Canada

I will be at #CVPR25! 🤗 Please come chat with me and Ethan Weber - during our poster session for Pippo, on Sat 5-7pm (Hall D)! 👋

Thrilled to share the papers that our lab will present at #CVPR2025. Learn more in this thread 🧵 and meet Kai He, Yash Kant @CVPR25, Ziyi Wu @ CVPR'25, and our previous visitor Toshiya Yura in Nashville! 1/n

Happening today in room 104 at #CVPR2025 joins in discussing what comes next in data driven simulation for AV.

I'm giving a keynote talk in a very interesting workshop on virtual humans and humanoid robots today at 2:25-2:50pm at 101B #CVPR2025 2025!

We will present the paper this afternoon. Come to chat with us!!

Grateful to everyone who stopped by our oral presentation and posters during the whirlwind of #CVPR2025 — we know you had plenty of options! I'm at the Eyeline Studios Booth (1209) from now–3pm today. Come say hi — I’d love to chat about our research philosophy and how it ties

🚨🎬 MiniMax Hailuo AI (MiniMax) 's latest video model, Hailuo 02, is now available exclusively on fal's API platform. For just 28 cents per video, you can generate these beautiful shots with our API. fal.ai/models/fal-ai/… Links and examples below 👇👇

✨ CVPR 2025 highlight #3 --- MatAnyone: Stable Video Matting with Consistent Memory Propagation MatAnyone is a robust framework for video matting that pairs image-level matting quality with video-level semantic consistency, and it does so while asking the user for just a single

I'm attending Robotics: Science and Systems for the first time #RSS2025 in LA Excited to be giving two invited talks, at the Continual Learning and EgoAct workshops on Sat, June 21 I'll share the latest on 2D/3D motion prediction from human videos for manipulation! Do drop by and say hi :)

One paper accepted to #ICCV2025! 💡 “LightSwitch: Multi-view Relighting with Material-guided Diffusion” 💡 Will release the arxiv, website, and code very soon 🌺

It's finally done cooking, excited to see what everyone does with it!

Presenting DemoDiffusion: An extremely simple approach enabling a pre-trained 'generalist' diffusion policy to follow a human-demonstration for a novel task during inference One-shot human imitation *without* requiring any paired human-robot data or online RL 🙂 1/n

The latest research paper from Eyeline Studios, FlashDepth, has been accepted to the International Conference on Computer Vision (#ICCV2025). Our model produces accurate and high-resolution depth maps from streaming videos in real time and is completely built on open-source

🔊 #ICCV2025 acceptance: ⚡FlashDepth⚡ estimates accurate and consistent depth for 2K-resolution videos in a real-time (24 FPS) streaming fashion on a single A100 GPU. ✊ Kudos to the teamwork led by our intern Gene Chou at Eyeline Studios . 👉 Join us to be the next one

My research group in the Codec Avatars lab at Meta contributed by developing the technology required to display the outputs of FAIR SeamlessNext's dyadic motion models as 3D Full-body Codec Avatars.

Adding horizontal lines to images improves VLM (vision language model) performance of tasks like counting, visual search, spatial understating, scene understanding, and more

Dear NeurIPS Conference -- it seems OpenReview is down entirely, and we cannot submit reviews for the upcoming review deadline tonight. Please share if you are having a similar issue. #neurips2025

Some samples of (positive) feedback from Scouts users this past week! We're continuing to let people in every day. Join the waitlist at yutori .com.