World models are the future and the future is now! 🌎🚀 Proud to share DreamerV2, the first agent that achieves human-level Atari performance by learning behaviors purely within a separately trained world model. Paper: arxiv.org/pdf/2010.02193… Thread 👇

Deep RL is not really using deep networks! We found that dense connections and deeper networks help improve learning performance. Results hold across various manipulation and locomotion tasks, for both proprioceptive and image observations! sites.google.com/view/d2rl/home

Really exciting multi-agent collaboration environment! Xavier Puig, Tianmin Shu, Shuang Li, Zilin Wang, Josh Tenenbaum, Antonio Torralba #machinelearning

How do we bridge Sim 2 Real gap? What is necessary and what is sufficient! 🤖🤯 Checkout: Dynamics Randomization Revisited A Case Study for Quadrupedal Locomotion Paper: arxiv.org/abs/2011.02404 Project: pair.toronto.edu/understanding-… Dennis Da, Xie Zhaoming @Mvandepanne Buck Babich

矢印の方向に動いて見えますね～？ 人間の脳は単純なので矢印で惑わされてしまうのでーす

Action Concept Grounding Network for Semantically-Consistent Video Generation deepai.org/publication/ac… by Wei Yu et al. #ComputerScience #Learning

Yoshua Bengio will give the next talk in causalcourse.com! Topic: Causal Representation Learning Date/time: Tuesday, December 1, at 1 pm EST Watch the course Slack for a join link. #CausalTwitter

Long term reasoning needs an understanding ofcontinuous changes in the world. "What happens if I open the door" Action Concept Grounding Networks learn these semantics arxiv.org/abs/2011.11201 iclr-acgn.github.io/ACGN/ Wei Yu W.Chen No longer on Xitter University of Toronto Robotics Institute Vector Institute

Fascinating new research suggests an approach to reversing age-related decline by reprogramming some cells to a ‘younger’ state. fiercebiotech.com/research/rever…

Learn more about this work: dpmd.ai/nature_nowcast… Access the code behind the model: dpmd.ai/github_nowcast… 4/4

Kubric: A scalable dataset generator abs: arxiv.org/abs/2203.03570 github: github.com/google-researc…

From logical reasoning to mental simulation, recurrent neural networks (RNNs) possess an incredible capacity for computation. To fully harness this capacity, we developed a programming language for RNNs arxiv.org/abs/2203.05032. Here is a game of pong programmed inside of an RNN.

Introducing Sora, our text-to-video model. Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions. openai.com/sora Prompt: “Beautiful, snowy

Meta announces Aria Everyday Activities Dataset present Aria Everyday Activities (AEA) Dataset, an egocentric multimodal open dataset recorded using Project Aria glasses. AEA contains 143 daily activity sequences recorded by multiple wearers in five geographically diverse indoor

How to become a toxic Reviewer💀 for #NeurIPS ? 🤔 1. List "Lack of technical novelty" as the only weakness. 2. Give the paper a rating 4 and confidence 4. 3. During rebuttal, acknowledge the authors' response but never take them into account. Tell the author that after

MinT beats Sora in multi-event generation! One week after the release of MinT, Sora also released a *storyboard* tool that targets the same task (sequential events + time control). Below are a few comparisons, where MinT shows better event transition and timing: (1/N)

🚀Excited to introduce GEN3C #CVPR2025, a generative video model with an explicit 3D cache for precise camera control. 🎥It applies to multiple use cases, including single-view and sparse-view NVS🖼️ and challenging settings like monocular dynamic NVS and driving simulation🚗.

Come find me at Hall 2B, booth #182 between 3:00–5:30 PM! I’ll be presenting EgoSim, our multimodal-conditioned video generation model. Stop by—I'd love to chat!