@johnkellden
Account based inSweden
About this account
- Account based in
- Sweden
- Connected via
- Web
Account-level information from X, not a live location or the device used for a specific post.
Cards catalyzing stories, Conversations that mind and matter, Digital communities and collaborative narratives
Sweden
Joined September 2007
- Tweets33K
- Following3.9K
- Followers1.7K
- Likes31.7K
johnkellden retweeted
Thinking Machines Lab exists to empower humanity through advancing collaborative general intelligence.
We're building multimodal AI that works with how you naturally interact with the world - through conversation, through sight, through the messy way we collaborate. We're excited that in the next couple months we’ll be able to share our first product, which will include a significant open source component and be useful for researchers and startups developing custom models. Soon, we’ll also share our best science to help the research community better understand frontier AI systems.
To accelerate our progress, we’re happy to confirm that we’ve raised $2B led by a16z with participation from NVIDIA, Accel, ServiceNow, CISCO, AMD, Jane Street and more who share our mission.
We’re always looking for extraordinary talent that learns by doing, turning research into useful things. We believe AI should serve as an extension of individual agency and, in the spirit of freedom, be distributed as widely and equitably as possible. We hope this vision resonates with those who share our commitment to advancing the field. If so, join us. thinkingmachines.paperform.c…
johnkellden retweeted
i am inspired by the men and women pursuing greatness , remaining kind , staying focused , being true to themselves
johnkellden retweeted
Cells aren’t robots executing their DNA as code. They are better understood inferential learning systems, with expression levels of DNA in the gene regulatory network roughly corresponding to the weights and activations of a transformer or neural net.
johnkellden retweeted
It's a beautiful thing to have faith that all parts of yourself, all of your relationships, experiences, and interests, no matter how seemingly disparate, disjointed, and dysfunctional, are forming an organic and essential divine unity that will be revealed in due time
johnkellden retweeted
When did elite institutions shift from being largely embedded local & national entities into international & global ones?
Think of elite law, capital, medicine, business, education?
When local ties that bind (“religio”) were attenuated, it was only a matter of time until the implicit A/B testing would illuminate the real (unnamed) value.
You can bury it for a time, but once you move into the spreadsheet imaginary, you forfeit the value of the personal, the local, the familial.
johnkellden retweeted
The physicist John Wheeler, inventor of the phrase, "it from bit" at a Princeton blackboard discussing what in nature can be quantized but as if he were looking at Walton Ford's, "Falling Bough," a fantastic depiction of collective behavior.
johnkellden retweeted
This is the most important paper in a long time . It shows with strong evidence we are reaching the limits of quantization. The paper says this: the more tokens you train on, the more precision you need. This has broad implications for the entire field and the future of GPUs🧵
[1/7] New paper alert! Heard about the BitNet hype or that Llama-3 is harder to quantize? Our new work studies both! We formulate scaling laws for precision, across both pre and post-training arxiv.org/pdf/2411.04330. TLDR;
- Models become harder to post-train quantize as they are overtrained on lots of data, so that eventually more pretraining data can be actively harmful if quantizing post-training!
- The effects of putting weights, activations, or attention in varying precisions during pretraining are consistent and predictable, and fitting a scaling law suggests that pretraining at high (BF16) and next-generation (FP4) precisions may both be suboptimal design choices!
Joint work with @ZackAnkner @spectorb @blake__bordelon @Muennighoff @mansiege @CPehlevan @HazyResearch @AdtRaghunathan.
johnkellden retweeted
All of this means that the paradigm will soon shift from scaling to "what can we do with what we have". I think the paradigm of "how do we help people be more productive with AI" is the best mindset forward. This mindset is about processes and people rather than technology.
johnkellden retweeted
In 2024, I had the opportunity to attend (and speak at) a handful of industry machine learning conferences. Whenever there was a break, I'd jot down some insights and lessons in apple notes. Here are 39 lessons I took away from conferences this year.
eugeneyan.com/writing/conf-l…
johnkellden retweeted
💡 How to Build an Experimentation Machine and Where Most Go Wrong with @rameshjohari (Stanford, Airbnb, Uber, Bumble, and more) 🎙️
In this episode of High Signal, I talk to Ramesh Johari—Stanford professor and advisor to platforms like Airbnb, Uber, and Bumble—about building systems for experimentation that drive innovation and avoiding the traps most organizations fall into.
We explore practical takeaways like:
- ⚡ How to combat risk aversion by running faster, smaller tests that reduce stakes while increasing learning.
- 🔄 Why a successful experimentation strategy requires learning across tests instead of treating experiments as one-offs.
- 🌟 How to create a self-learning organization, where experimentation fuels continuous improvement and innovation.
🎬 Watch the clip where Ramesh shares how organizations can build a "learning flywheel" to stay adaptive in dynamic environments.
💻 Check out the full episode via the link in the comments!
johnkellden retweeted
Peter Thiel making the case for open source, decentralized AI
"The real AI risk is a totalitarian world government... Global compute governance would be very heavy-handed, creating a dystopian one-world nanny-state government."
youtu.be/42iVcEg5SOM?si=qp07…
johnkellden retweeted
Important finding: atmospheric rivers shifted 6-10 degrees (400-700 miles) toward the 2 poles over the past 4 decades, mostly due to a poorly-understood shift toward a more La Nina-like climate. More drought for California, and flooding for BC and Alaska. theconversation.com/atmosphe…
