An impressive paper that explores the balance between generalization and memorization in diffusion models. The authors argue that diffusion models do not globally underfit their training objectives but instead exhibit selective underfitting: they fit the empirical score function…
I gave a talk today at The Curve on the state of open models.
Here are the slides, recording soon.
Topics include: Chinese ecosystem, reflections on DeepSeek, the demise of Llama, who will fill the U.S. market, what local models do, ATOM project & ai2, and more topics
Thoughts on Sora 2
1) The main difference between Veo3 and Sora 2 was marketing. The model itself is better but is not that much of a step forward over Veo 3.
2) OpenAI engineered a very clever marketing strategy using influencers and invite codes. The invite code made Sora…
Token compression causes irreversible information loss in video understanding. 🤔 What can we do with sparse attention? We introduce VideoNSA, a hardware-aware and learnable hybrid sparse attention mechanism that scales to 128K context length.
New Image model "Lavender" is currently being tested on the Artificial Analysis Arena
The model is likely related to the GPT-image-1 model, possibly an update
Prompt: A cartoonish 1940s detective’s office rendered in a loose watercolor style: a gumshoe in a pinstriped suit,…
🔥 Ming-UniAudio: The 「Nano Banana」moment for speech is here! A single model for universal understanding, generation & free-form editing.
First Unified Continuous Tokenizer 「MingTok-Audio」and Unified Und & Gen Speech LLM built on it.
First Universal Free-form Speech Editing…
Indeed! Glad to see lot of interesting explorations of new video paradigms going on in the community. I am also glad to see Diffusion Forcing making an impact beyond video gen, in RL and robotics too!
Indeed! Glad to see lot of interesting explorations of new video paradigms going on in the community. I am also glad to see Diffusion Forcing making an impact beyond video gen, in RL and robotics too!
FREE SORA WATERMARK REMOVER IS LIVE!!!
Absolutely free tool to use. Print out UGC videos with SORA2 and use topyappers.com to remove the watermark, change metadata, brightness, hue, and a bunch of other stuff.
Make your videos undetectable when uploading to social…
🚀 Introducing Ming‑UniVision & MingTok — the first autoregressive model to natively unify vision understanding & generation in a continuous unified representation space.
First unified continuous visual tokenizer .
>3.5x faster convergence
Seamless multi‑round interaction…
Sora 2 Pro, the higher-fidelity video generation model, is now rolling out with options for both 10s and 15s duration videos, plus high and standard resolution
People often ask about the relationships between the different Qwen models and the logic behind their naming. Currently, we have several distinct model families: LLM, Coder, VL, Omni, and a newer addition, Image. While our long-term goal is to unify them into a single, truly…
🚨New Anonymous Image Editing Model: Riverflow 🚨
The performance is definitely SOTA! They are testing it in Artificial Analysis right now. The naming seems very Google-like, however, I am not sure about this yet.
Prompt: Make the woman appear to be suppressing a laugh.
492 Followers 6K FollowingIlluminati protect me through my war thanks for Illuminati, if you want riches protection you can get it here, dm me privately so I can link you up to riches 💫
120 Followers 5K FollowingGuiding @Elonmusk’s vision for a better future through SpaceX, Tesla, Neuralink and more 🚀 I teach enthusiasts, dream chaser and innovation advocate 🌟
2K Followers 2K FollowingOwner and head designer of own fashion studio in Berlin. specializes in designs that combine Nordic minimalism with German elegance
140K Followers 45K FollowingEx @Streamlit @Snowflake Maestro 🪄 • X about AI agents, LLMs, web apps, Python & SEO • My ❤️ is open source • DM for collabs 📩
90 Followers 1K FollowingNo porn, no sales, a Thai woman living in Toronto who likes shopping, traveling, fitness, swimming, investing, and amateur golf turf killers
14K Followers 3K Followingresearch @MIT_CSAIL @thinkymachines. work on scalable and principled algorithms in #LLM and #MLSys. in open-sourcing I trust 🐳. she/her/hers
19K Followers 1K Followingapplied AI @openai. I work with the world's leading startups and developers to bring the benefits of safe AI to every human. views my own 🇮🇳 @dukeu
145K Followers 1 FollowingClaude is an AI assistant built by @anthropicai to be safe, accurate, and secure. Talk to Claude on https://t.co/ZhTwG8dz3D or download the app.
2K Followers 479 FollowingPh.D. student @HKUSTGuangZhou, Researcher @MetaGPT_, Cofounder of OpenManus, previously at RUC, Lenovo Research AI Lab, Zhipu AI.
212K Followers 3K FollowingSave your best highlights from Kindle, Twitter, Pocket, Instapaper, iBooks, and 30+ others.
Then revisit, search, organize, and export them seamlessly.
227K Followers 38 FollowingThe Gemini app turns research into reality, bringing frontier AI experiences like Veo 3, Deep Think, Nano Banana and more to hundreds of millions of people.
2.5M Followers 23K FollowingReposting Trump’s Truth Social posts (with date/time) on X + news/commentary. Unofficial. Profile Artist: @ElenaRuseva1 Not affiliated with @realdonaldtrump.
14K Followers 616 FollowingFounder and CPO at @reve. Former VP of Product at Stability AI. Ex-Adobe. Creator of the Stable Diffusion Photoshop plugin. Writer repped by Gersh.
24K Followers 251 Following@Cohere's research lab and open science initiative that seeks to solve complex machine learning problems. Join us in exploring the unknown, together.
18K Followers 80 Following财经媒体人,写作中国商业深度报道,包括AI/科技巨头/风险投资/人物,也是播客《张小珺商业访谈录》主持人、制作人。Financial writer covering China business world, also the producer and host of "Zhang Xiaojun Podcast."
19 Followers 8 FollowingOnoma AI strives to advance the frontiers of AI technologies to revolutionize and deliver high quality creative contents such as webtoon!
229K Followers 1 FollowingUpdates for developers building with the OpenAI Platform and API • Service status: https://t.co/kZwnwdYqOS • Support: https://t.co/qCi6M5ESZU