Sharath Girish amrita Abhinav Shrivastava David Luebke Shalini De Mello NVIDIA NVIDIA AI Univ. of Maryland amrita will present our poster on Friday 4:30-7:30pm. Please say hi 😀 neurips.cc/virtual/2024/p… #NeurIPS2024 🌐 website: research.nvidia.com/labs/amri/proj… 📄 paper: arxiv.org/abs/2412.04469

1/ Founders live for proof of product market fit. For undeniable evidence that what you're building matters. I've seen it before at Facebook and Dropbox. In 2024 I saw it at South Park Commons. This is what PMF looks like for -1 to 0 🧵

📢 Assistant Professor Ruohan Gao (Ruohan Gao) is joining UMD Department of Computer Science! His research centers on computer vision and machine learning, focusing on multisensory learning—integrating sight, sound and touch. 👁️🎧🦾 Learn more: go.umd.edu/Gao1-2025

This past week has been an amazing blur. Last Wednesday at South Park Commons' demo faire Rohit and I had the opportunity to present what we've been grinding on for the past few months - and I'm blown away by the reception we've been getting! youtube.com/watch?v=NezuDq…

📢 Excited to announce LARP has been accepted to #ICLR2025 ! 🇸🇬 Code and models are publicly available. Project page: hywang66.github.io/larp/index.html

Over the last 11 years at @microsoft Satya Nadella has had one of the greatest runs of any CEO, ever. I'm thrilled to announce he will be joining us March 4th at South Park Commons to tell us how – and share what he sees on the horizon.

wait… what…

New open source reasoning model! Huginn-3.5B reasons implicitly in latent space 🧠 Unlike O1 and R1, latent reasoning doesn’t need special chain-of-thought training data, and doesn't produce extra CoT tokens at test time. We trained on 800B tokens 👇

Success isn't automatic. You always have to be "refounding". Few know this better than Satya Nadella. I'm excited to share our full conversation with Satya at South Park Commons: quantum, AI, taking over Microsoft. Everything you'd want to learn from one of the GOATs.

I feel a little bit for the Google DeepMind team.. You build a world changing model and everyone is posting Ghibli-fied pictures instead. But this is the core problem with Google - they can build the best models in the world but if they don’t focus on the consumer experience

wait for it..... wait for it......... now, there it is. Something fun is cooking!

Exploration is key for robots to generalize, especially in open-ended environments with vague goals and sparse rewards. BUT, how do we go beyond random poking? Wouldn't it be great to have a robot that explores an environment just like a kid? Introducing Imagine, Verify,

Brought to you by the amazing UMD Department of Computer Science students Seungjae Lee Seungjae (Jay) LEE, Daniel Ekpo (Daniel Ekpo), Haowen Liu, and my colleagues Furong Huang and Abhinav Shrivastava Check out the project page for more visual results! ive-robot.github.io

My dream for Physical Artificial Intelligence? An embodied agent that ventures into the wild 🌿, builds its own mental world-model 🧠, and learns like a curious child—by seeing, imagining, verifying, and doing. We built an initial version of this vision in a project led by our

If you’re at #ICRA2025 come see us present P3-PO in the Representation Learning I session. Room 311 at 4:40

prompt engineering leveling up my parenting skills… me: why don’t you want to eat? kiddo (3 y/o): because i’m not hungry me: why aren’t you hungry? kiddo: because i don’t want to eat me (after a few more cycles of this): why aren’t you eating or hungry? kiddo: because mama

Speed up your diffusion models—with evolutionary caching! Proud to share Ani Aggarwal’s first paper (mentored by Matt Gwilliam)! ECAD uses evolutionary search to find Pareto-optimal schedules that speed up diffusion & boost quality—no retraining or tuning. Read the full 🧵below

Your company doesn't start when you launch. It doesn't start with traction. It starts when your ambition sets you on the messy path to attempt greatness. It starts at -1. So should your support.

🎉 Excited to share our paper "Trokens: Semantic-Aware Relational Trajectory Tokens for Few-Shot Action Recognition" has been accepted to #ICCV2025! Equally co-led with Shuaiyi Huang — we advance few-shot action recognition via smart point tracking. 🔗 trokens-iccv25.github.io 🧵👇