おすすめツイート
Excited to introduce Dreamer 4, an agent that learns to solve complex control tasks entirely inside of its scalable world model! 🌎🤖 Dreamer 4 pushes the frontier of world model accuracy, speed, and learning complex tasks from offline datasets. co-led with @wilson1yan
🚀Try out rCM—the most advanced diffusion distillation! ✅First to scale up sCM/MeanFlow to 10B+ video models ✅Open-sourced FlashAttention-2 JVP kernel & FSDP/CP support ✅High quality & diversity videos in 2~4 steps Paper: arxiv.org/abs/2510.08431 Code: github.com/NVlabs/rcm
We’re proud to announce that Genie 3 has been named one of @TIME’s Best Inventions of 2025. Genie 3 is our groundbreaking world model capable of generating interactive, playable environments from text or image prompts. Find out more → goo.gle/3KGqiYa
Today we're sharing the next phase of Reflection. We're building frontier open intelligence accessible to all. We've assembled an extraordinary AI team, built a frontier LLM training stack, and raised $2 billion. Why Open Intelligence Matters Technological and scientific…
Transfusion combines autoregressive with diffusion to train a single transformer, but what if we combine Flow with Flow? 🤔 🌊OneFlow🌊 the first non-autoregressive model to generate text and images concurrently using a single transformer—unifying Edit Flow (text) with Flow…
SOTA humanoid locomotion: 20M params Fruit fly: 50M params
Scientists have mapped the entire neural network of the fruit fly. 50 million synapses between 139,000 neurons.
Humanoid motion tracking performance is greatly determined by retargeting quality! Introducing 𝗢𝗺𝗻𝗶𝗥𝗲𝘁𝗮𝗿𝗴𝗲𝘁🎯, generating high-quality interaction-preserving data from human motions for learning complex humanoid skills with 𝗺𝗶𝗻𝗶𝗺𝗮𝗹 RL: - 5 rewards, - 4 DR…
🔥Veo 3 has emergent zero-shot learning and reasoning capabilities! This multitalented model can do a huge range of interesting tasks. It understands physical properties, can manipulate objects, and can even reason. Check out more examples in this thread!
Veo is a more general reasoner than you might think. Check out this super cool paper on "Video models are zero-shot learners and reasoners" from my colleagues at @GoogleDeepMind.
Dreamer 4 takes world models to a new level - training multi-task agents fully in imagination while never touching the real environment. Powered by a novel shortcut forcing method, it delivers lightning-fast, accurate predictions and crushes benchmarks, beating OpenAI’s VPT…
Excited to introduce Dreamer 4, an agent that learns to solve complex control tasks entirely inside of its scalable world model! 🌎🤖 Dreamer 4 pushes the frontier of world model accuracy, speed, and learning complex tasks from offline datasets. co-led with @wilson1yan
Hard to overstate how big of a deal this is: 1. Train a video model on Minecraft 2. Train agent on video model WITHOUT ACTUALLY PLAYING THE GAME 3. The agent gets better at Minecraft Learn in imagination, act in reality 🤯 ∴ define a great env -> learn anything More…
Excited to introduce Dreamer 4, an agent that learns to solve complex control tasks entirely inside of its scalable world model! 🌎🤖 Dreamer 4 pushes the frontier of world model accuracy, speed, and learning complex tasks from offline datasets. co-led with @wilson1yan
LimX Dynamics' Oli humanoid robot autonomously collects tennis balls from the floor. ⦿ Oli is 5'5" tall, weighs 55 kg (121 lb), has 31 DoF ⦿ Ships with a modular SDK for Python-based development
complex robotics self-learning on video generation will go *very* far. It seems to have very nice scaling properties. Expect the most general robots in 1-2 years to be pretrained on a lot of video and text, and then train robotics skills within generated worlds
Excited to introduce Dreamer 4, an agent that learns to solve complex control tasks entirely inside of its scalable world model! 🌎🤖 Dreamer 4 pushes the frontier of world model accuracy, speed, and learning complex tasks from offline datasets. co-led with @wilson1yan
Clean and well-executed new work from @danijarh @wilson1yan , and it's cool to see shortcut models working at scale! The exciting finding is that you can train the world model largely on *unlabelled* videos, and only need a small action-anchoring dataset.
Excited to introduce Dreamer 4, an agent that learns to solve complex control tasks entirely inside of its scalable world model! 🌎🤖 Dreamer 4 pushes the frontier of world model accuracy, speed, and learning complex tasks from offline datasets. co-led with @wilson1yan
Today, @ekindogus and I are excited to introduce @periodiclabs. Our goal is to create an AI scientist. Science works by conjecturing how the world might be, running experiments, and learning from the results. Intelligence is necessary, but not sufficient. New knowledge is…
Sora 2 the same day as Dreamer 4. Coincidence? I don't think so 🤪
Excited to introduce Dreamer 4, an agent that learns to solve complex control tasks entirely inside of its scalable world model! 🌎🤖 Dreamer 4 pushes the frontier of world model accuracy, speed, and learning complex tasks from offline datasets. co-led with @wilson1yan
United States トレンド
- 1. Auburn 26.6K posts
- 2. Kyle Tucker 1,836 posts
- 3. #UFCRio 52.1K posts
- 4. Penn State 26.6K posts
- 5. Michigan 56.6K posts
- 6. Nuss 5,123 posts
- 7. Indiana 48.2K posts
- 8. Billy Napier 1,897 posts
- 9. James Franklin 14K posts
- 10. Sherrone Moore 1,104 posts
- 11. Hugh Freeze 1,348 posts
- 12. Chad Patrick N/A
- 13. Diane Keaton 218K posts
- 14. Charles 108K posts
- 15. King Miller N/A
- 16. Oregon 69.5K posts
- 17. Fickell 1,895 posts
- 18. #GoBlue 2,159 posts
- 19. Andrew Vaughn 1,380 posts
- 20. Gilligan's Island 2,941 posts
おすすめツイート
-
Noam Brown
@polynoamial -
Sergey Levine
@svlevine -
Karol Hausman
@hausman_k -
Shimon Whiteson
@shimon8282 -
Yi Tay
@YiTayML -
Chelsea Finn
@chelseabfinn -
Brandon Amos
@brandondamos -
Jakob Foerster
@j_foerst -
Tim Rocktäschel
@_rockt -
Durk Kingma
@dpkingma -
Abhishek Gupta
@abhishekunique7 -
Jeff Clune
@jeffclune -
Deepak Pathak
@pathak2206 -
Ofir Nachum
@ofirnachum -
Behnam Neyshabur
@bneyshabur
Something went wrong.
Something went wrong.