moab.arar
@ArarMoab
Followers
358
Following
606
Media
11
Statuses
302
FAIR (@AIatMeta) | Ph.D. | Generative World Models | ex Google
Joined March 2021
Checkout our work "GameNGen". A Gaming engine powered by a diffusion-model that simulates DOOM in Real-Time! Find out more: https://t.co/1RKiwMCZZJ Amazing effort and fun collaboration with the incredible @daniva, @yanivle, and @shlomifruchter!
Google presents Diffusion Models Are Real-Time Game Engines discuss: https://t.co/RQTrNSqGHE We present GameNGen, the first game engine powered entirely by a neural model that enables real-time interaction with a complex environment over long trajectories at high quality.
2
17
51
Check out our recent paper on motion editing: MotionsV2V. Let’s say you have a video you like, but want some of the objects to behave a bit differently while preserving the rest: for instance, you don't want the cat to look at the fish.
1
3
15
Frontend dev is changed forever!
Gemini 3 is out in the wild and it’s a pleasure to share that generative UI is coming to Google Search in AI Mode and Gemini. Gen UI enables an AI model to generate not just content but also the interface. Gemini 3 analyzes the prompt and designs a custom response just for you!🧵
0
0
1
If you missing publishing your research and contributing to the open community - my research group is hiring and provide competitive offering See our latest paper release in the comments Additional models and papers are already in the oven - great time to join us 🚀
1
2
12
Wow 🤯
Generating an image from 1,000 words. Very excited to release Fibo 😃, the first ever open-source model trained exclusively on long, structured captions. Fibo sets a new standard for controllability and disentanglement in image generation [1/6] 🧵
0
0
1
We present DyPE, a framework for ultra high resolution image generation. DyPE adjusts positional embeddings to evolve dynamically with the spectral progression of diffusion. This lets pre-trained DiTs create images with 16M+ pixels without retraining or extra inference cost. 🧵👇
9
32
102
Our insertion model is now available to try! It's very fun to 'touch' a video by adding new things to it.
Veo is getting new precision editing capabilities that let you easily add or remove elements from a scene - all while preserving the integrity of your original video. 🎥
2
1
9
@hila_chefer is presenting “From Still to Moving: Temporal Priors as Creative Tools for Personalization” at our @P13N_Workshop at @ICCVConference
0
1
8
@YVinker is giving her talk “Personalization Methods for Design and Artistic Creation” in our @P13N_Workshop at @ICCVConference
1
5
22
Time update: I'll be presenting at 11:00 AM (instead of 11:45) at the AI4VA Workshop. See you there!
AI for Visual Arts Workshop Sun 19.10 • 11:45 • Room 313A I'll talk about generative tools for designers beyond pixel generation, showing examples from our recent works NeuralSVG and SketchAgent 🔗
0
1
13
It was really difficult to use Google to find solutions to undergraduate calculus problems. Maybe AGI is not around the corner… but search is improving 😅
@kevinweil Hi, as the owner/maintainer of https://t.co/69gOJM7Ci7, this is a dramatic misrepresentation. GPT-5 found references, which solved these problems, that I personally was unaware of. The 'open' status only means I personally am unaware of a paper which solves it.
0
0
0
Decreasing validation loss gives me the adrenaline - can’t sleep now!
1
0
1
Emergent behaviors are showing up in video models like Sora 2 and Veo 3. People are asking them to solve real-world tasks. They’re not fully reliable yet, but with more scale and efficiency, they will become general-purpose AI, going beyond creativity and entertainment.
1
0
1
Thrilled to share that two papers got into #NeurIPS2025 🎉 ✨ FlowMo (my first last-author paper 🤩) ✨ Revisiting LRP I’m immensely proud of the students, who not only led great papers but also grew and developed so much throughout the process 👇
Beyond excited to share FlowMo! We found that the latent representations by video models implicitly encode motion information, and can guide the model toward coherent motion at inference time Very proud of @ariel__shaulov @itayhzn for this work! Plus, it’s open source! 🥳
4
13
158
We release Code World Model (CWM)! 👩💻🌎📊 A coding LLM designed to advance code generation research through agentic reasoning and world-model-based planning. Super excited about this release and proud of the team’s work! 😃 See Gab's post for more info 👇
(🧵) Today, we release Meta Code World Model (CWM), a 32-billion-parameter dense LLM that enables novel research on improving code generation through agentic reasoning and planning with world models. https://t.co/BJSUCh2vtg
0
12
50
OpenAI was the first to scale video generation with Sora. I wonder if Sora 2 is next? Perhaps reaching on the order of 100B parameters?
Over the next few weeks, we are launching some new compute-intensive offerings. Because of the associated costs, some features will initially only be available to Pro subscribers, and some new products will have additional fees. Our intention remains to drive the cost of
0
0
1
I asked an Architect, Gemini (🍌), and ChatGPT to redesign my daughters’ room to fit a new desk. While generative models can generate beautiful designs, they are still missing the ability to understand the practical constraints a human designer can see. Future generative models
0
0
6
🧵 PractiLight: Practical Light Control Using Foundational Diffusion Models 🔗 https://t.co/YweM7ir948 Diffusion models already “know” a lot about light transport. No need for massive finetuning that hurts generalization. {1/6} 👇
6
42
266
Whoever came up with the name Nano Banana either typo’d Nana Banana or is messing with us… my tongue is struggling… 🍌
0
0
3