Diffusion/flow matching is a go-to generative AI paradigm, but... it's slow, sample inefficient and complicated. Do we really need all this complexity? We design a small and simple one-step model, ROMS-IMLE, which attains an FID of 2.56 on ImageNet. Joint w/ Chirag Vashist 1/6show more

Generative models (diffusion/flow) are taking over robotics 🤖. But... do we really need to model the full action distribution to control a robot? We suspected the success of Generative Control Policies (GCPs) might be "Much Ado About Noising." We rigorously tested the myths. 🧵👇show more

“I need Chet, Chet needs me, we both need... Hart and Dub, and K-Rich and so on. We all need each other. Jokic is a really talented guy but all we care about is winning the game and we got one tonight and we got to go get one more.” -SGAshow more

We just brought flow maps to language modeling for... one-step sequence generation 💥 Discrete diffusion is not necessary -- continuous flows over one-hot encodings achieve SoTA performance and ≥8.3× faster generation 🔥 We believe this is a major step forward for discrete generative modeling and language modeling alike. 🚀 Full thread from first author Chanhyuk Lee:show more

We are releasing 4M-21 with a permissive license, including... its source code and trained models. It's a pretty effective multimodal model that solves 10s of tasks & modalities. See the demo code, sample results, and the tokenizers of diverse modalities on the website. IMO, the multitask learning aspect of multimodal models has really taken a step forward. We can train a single model on many diverse tasks with ~SOTA accuracy. But a long way to go in terms of transfer/emergence. 🌐 ⌨️ Joint work w/ EPFL Apple.show more

The co-founders of Flapping Airplanes call the current RL... paradigm for model training "environment slop." They explain: "The reinforcement paradigms of today are shockingly inefficient. You don't really get much generalization across tasks, you teach a model through one kind of learning and then you teach it the next one. It's kind of like whack-a-mole. We look at this and think it's kind of crazy. The next paradigm of AI will not be environment slop." "Human level intelligence is not the ceiling, it is merely the floor on what is possible. If you can train models with vastly less data and possibly more compute in very different ways, what is going to happen? We actually don't know. But I do think they'll be different and weird and they'll have interesting capabilities that we'll find really valuable ways to use."show more

GenCast is a diffusion model, similar to the machine... learning models which also power generative AI. 🎨 We trained it on 40 years of historical data from ECMWF, which included variables such as temperature, wind speed, and pressure at various altitudes - enabling it to learn global weather patterns.show more

We release a technical blog on training PocketTTS, our... 100M parameter on-device TTS, through drifting, a recent one-step generative objective from Deng et al. Less than 1% WER, high quality and voice cloning. To the best of our knowledge, it's the first speech model, and the first autoregressive model, trained this way.show more

“Soccer is not a men's game. It's the people's... game. And we all belong. Young girls need to know….they belong on television for the entire world to see." This Sinclair speech is one of a kind. She is a role model, an icon, and a Portland legend.show more

The best founders in the world like Elon Musk... or tobi lutke “go ultra-hardcore on deletion and simplification.” Luca Ferrari says Bending Spoons employs this idea of “Relentless Simplification” across their businesses: “Humans have a tendency to add complexity and do things that destroy value.” “If you leave an organization, almost any environment unattended, and you don't provide guidance in this regard, it will tend to become more complicated. People will add parts. It could be expanding a team or adding a step to a process, or adding an entire new process. If it's a product, adding a feature to the product. New rules. It applies to almost any human endeavor.” “People tend to add pieces, they very rarely remove pieces.” “If you don't make a conscious effort to achieve simplicity, and avoid this increasing value-destroying complexity, that's how we got to our modern society with all the bureaucracy and complicated regulation.” “We have this principle whereby we ask everyone who works here, every time someone is suggesting that we should be adding complexity, the burden of proof is on those making that suggestion. The people who support the thesis that we shouldn't be adding complexity don't need to prove it. They just have to raise a flag and say, ‘I don't think we should.’ The burden of proof is on those who want to add complexity.” “The other part of relentless simplification is that we want people to be on the lookout for existing complexity and suggest that we should remove it.”show more

🔬 New Science pod with CuspAI! We are entering... a new era where materials science and discovery is transitioning from slow, manual experimentation, to a high-speed search problem powered by generative AI and "physics processing units." Max Welling argues that the foundation of all modern technology—from GPUs to climate solutions—is a materials problem, and that unifying the mathematics of stochastic thermodynamics with generative AI will unlock a new paradigm of automated scientific discovery.show more

[ENG TRANS] SUZY’s narration at 59th Baeksang Arts Awards... “As always, all of us live a simple and an ordinary life. We go to school, we go to work, we meet friends and ask each other how they have been doing, but how are we really?” cont.show more

I’ve really enjoyed working in Ukraine this year, in... a small way helping design their veterans services. The scale and complexity is enormous; they will need lots of support in the year ahead. They are a great people and we must not waver in our commitment to them. 🇺🇦🤝🇬🇧show more

All 11 sectors are firing on all cylinders, and... the reason is people talk a lot about AI in a negative way, but the truth is it's found its way into all 11 sectors, added margin, and added productivity. Numbers are showing up in cash flow and earnings. It's that simple. No one saw this coming. It's just one of those serendipitous moments where technology adds a lot of value to an economy. It's doing it here. It's doing it all around the world.show more

Trump: "We'll be putting an initially small tariffs on... pharmaceuticals. But in one year, one in a half years maximum, it's gonna go to 150% and then it's gonna go to 250%, because we want pharmaceuticals made in our country."show more

Trump: Concerns of AI Taking Over is a Hoax.... Bigelow: "It is no goddamn hoax." (I'll have more on their AI discussion in a bit. It was good.) Bigelow: "This is the Las Vegas Review-Journal, this September 15. So it's this big hand, and it says, 'Who Controls the World?' Up above in small print, it says 'Artificial Intelligence.' And 'Who Controls the World?' It's a really serious thing. "Now, we have here...researcher says human race is in serious danger. President [Trump] says concerns of AI taking over is a hoax. George, it is no goddamn hoax. We don't know enough yet to where we can say, we don't need to be really concerned. We know enough to say we should be really concerned. Okay? "So....and this is actually a really good article. If you haven't read it, you being the journalist and all [Knapp laughs]. Here, I'll give it to you." George Knapp: "All right." Bigelow:"It's a good article."show more

The DEPRESSING reality of AI adoption curves Advanced AI... just broke into it's third major paradigm since launching. Paradigm One was the simple autocomplete engine. GPT-2 and original GPT-3 were glorified autocomplete tools, just "next token predictors." Paradigm 1.5 was when we added "instruct-aligned" GPT-3, which set the stage for Paradigm Two. Paradigm Two was the chatbot era, with ChatGPT being the front-runner. Paradigm 2.5 was when we started adding reasoning, tool use, and RAG, which set the stage for agentic abilities (Paradigm 3). OpenClaw just blew the lid off Paradigm 3, and we're just at the beginning of this new ramp-up in capabilities. The thing is... each of these paradigm shifts creates fundamentally different UX and technical affordances. While most Fortune 500 companies are still struggling to figure out the cyber security, legal, and financial risks of chatbots, the industry is going whole hog into autonomous agents. And many people will try to graft their understanding of chatbots onto agents. But that's like trying to compare the electric lightbulb to the electric motor.... ...yes they both ran on electricity, but their uses, affordances, limitations, and risks were fundamentally different. Yes, autocomplete, chatbots, and agents all run on "next token prediction" but that's like saying "it all runs on electrons." My goal today is to help give you a better intuition as to why adoption is so slow and to give you a new reference frame to understand that agents are a phase change from chatbots. But also... the state and big companies are going to move depressingly slow on all of this...show more

🪐 WORLD CREATORS Generative AI opened the door for... everyone to tell their stories without the need for a crew or a big budget. We started slowly with images and saw an enormous interest in tools like Midjourney, DALL-E, or Leonardo. Now, a bit more than a year later, we find ourselves in front of new technologies that are transforming our way of life, how we create, and how we consume content. ChatGPT is rapidly replacing Googling, and all the generative AI tools are replacing all the stock photos and videos, with tools like Runway enabling storytelling for everyone. And we won't take over Hollywood. We will surpass it in ways we can't even imagine right now. We'll be able to create worlds and stories beyond our wildest imagination. Generative AI is enabling creative people to go 1000x on what was possible before. At the core of everything we'll still need human creativity, and we have to understand that AI tools without it will be nothing. Until then, enjoy a short video I've put together to illustrate a small fraction of the worlds we will create soon. What worlds and stories are you excited to create?show more

🎙️ Jeff Peterson on the trade deadline's success and... asset management: "Over the next 7 drafts, we have 11 first-round picks, 15 second-round picks." "Being able to continue to stockpile those draft picks, it's always something that myself and our front office—it's been a goal of ours. We'll continue to add 'em, and when we have moments like this—when we see a player that we think is worth it or for the price—we'll continue to try to use those picks to go get them." 🎙️ On knowing when to make a big move: "We'll know when the time is right." "A lot of nuances go into it—which players are available, contract lengths, different dynamics that go into all of that. I promise you we will know when the time is right. This recent stretch has been beautiful to watch. All the credit goes to the players, Charles Lee, and his coaching staff... it's been really fun to see. All of this has just continued to add to the sample size that we need. In the summer, we'll figure out exactly what this team is and what else we need to do to continue to take the next step."show more