Anyscale

🚨 Anyscale Meet-Up Alert 🚨 Learn from the experts shaping scalable AI systems! Join Robert Nishihara, co-founder of Anyscale, and Jason Lopatecki, co-founder of Arize AI, on Wednesday, February 5th, as they discuss how to scale and evaluate production AI effectively. Space is limited—reserve your spot today: https://lnkd.in/gK9NCDcv

Just sat down to read the DeepSeek-R1 paper. We're entering an era where compute isn't primarily for training. It's for creating better data. I expect to see the money & compute spent on data processing (generation / annotation / curation) grow to match and exceed the money & compute spent on pre-training. People have talked about pre-training plateauing because we're "running out of data" on the internet to scrape. While that may be the case, capability improvements are going to continue full steam ahead. The improvements in intelligence are going to come not from putting in more data (scraped from the internet) but rather from putting in more compute (to generate higher-quality data). Intuitively, this feels similar to me to how people learn. You don't learn just by ingesting lots of tokens. In many cases, you learn by thinking more (I am referring to training time, but thinking more also applies at inference time). There are many creative ways to put in more compute to get better data, and this problem will be an important research area for a number of years. - In this paper, they train two models. Why two models? The first one (a reasoning model trained via RL) is used to generate data to train the second. This works by using the first model to generate reasoning traces and then selectively keeping only the high quality outputs (quality is judged by simply checking the results). This approach of "checking the results" works well for domains like math and coding where you can easily check the results. - In drug development, it is super common to put compute into generating better data in two phases. In the first phase, a generative model (e.g., for protein sequences) generates a massive number of candidate drugs. In the second phase, scoring or filtering is done with a slew of predictive models which may predict structure, toxicity, solubility, binding affinity, etc. After all this work is done, you may end up with 100 data points. - In physical domains (e.g., climate applications), expensive but accurate physics simulators exist (these simulations are run on super computers for long periods of time to simulate the physics of the atmosphere or some other system). All of that data can be used to train models, which is showing a ton of promise. The question of "how to put more compute into generating better data" is central to progress in AI right now. Over the last 6 months at Anyscale, we've see a ton of folks in the Ray community working on this.

Only a few days left to shape the future of Ray! Fill out the survey and make your voice heard: https://lnkd.in/g7wea5rB The survey closes on Monday, January 27th.

Last night, Anyscale and Alluxio co-hosted an AI/ML Infrastructure Meetup, and the energy was incredible 🙌 Over 100 attendees joined in person, with 400+ registrants eager to explore the latest in AI infrastructure trends, GPU optimization, and cutting-edge LLM inference techniques. Key highlights: ⚡Robert Nishihara decoded critical AI infrastructure trends ⚡ Bin Fan revealed GPU optimization breakthroughs ⚡ Junchen Jiang shared LLM inference magic - 10x performance improvements! 2025 is off to an amazing start! 🚀 Major thank you Hope Wang and Alluxio co-hosting the meetup.

🚨 Anyscale Meet-Up Alert 🚨 Join Anyscale and Arize AI on Wednesday, February 5th, for an evening of technical insights and engaging discussions. Hear from Robert Nishihara (Co-Founder of @anyscalecompute) on distributed data processing with Ray, and Jason Lopatecki (Co-Founder of Arize AI) on evaluating audio in multimodal AI applications. Space is limited—reserve your spot now: https://lnkd.in/gK9NCDcv

🚨 Join us this Thursday for an AI/ML Infra Meetup you won’t want to miss!🚨 Hosted by Anyscale and Alluxio 🙌 We’ll dive into: -Distributed computing techniques for the real world -Supercharging data access with Alluxio + Ray integration Don’t miss this chance to learn from the experts shaping the future of AI infrastructure. Spots are filling fast—sign up now: https://lu.ma/swc2imah

Join us on February 13th for an exciting meetup, co-hosted by Anyscale and ByteDance! Learn how to build scalable LLM applications with Ray, and hear from ByteDance as they showcase how they scale complex video processing workflows using Ray. Seats are limited -- don’t miss this chance to dive into real-world AI infrastructure innovations! 🙌 👉 https://lu.ma/ji7atxux

Really exciting to see Uber pushing boundaries with Ray! By integrating it into their mobility marketplace allocation system, they achieved 40x performance gain--boosting efficiency, scalability, and real-time optimization while powering mission-critical systems 🙌 🚀 Read more about how Uber leverages Ray to power their advancements: https://lnkd.in/gQ3k9YYm

Help Shape the Future of Ray! 👉 https://lnkd.in/ggA5XBWd The Ray community is the heart of everything we do. Now, as we look toward the next exciting chapter of Ray, we need your voice to help guide us. We’re launching an online survey to capture the pulse of our incredible community. Your input will play a crucial role in shaping the future of Ray. The holidays might be over, but it is still the giving season so to thank you for your time, we’re giving away some exclusive swag, fresh off the press with our previously unannounced Ray rebrand, featuring new styling and a new logo! 🙊🎁🎊 - 🧥 First 10 participants: Ray quarter zip - ☕ Next 20 participants: Ray tumbler - 👕 Next 100 participants: Ray t-shirt - 💻 All participants: Ray laptop stickers To be eligible for swag, don’t forget to provide your email address, company info, and a US shipping address at the end of the survey. This is your opportunity to share your insights, feedback, and ideas. Let’s build the future of Ray together! 🚀

🚨Anyscale Meet-up Alert🚨 Discover how industry leaders scale and evaluate production AI. Join Robert Nishihara co-founder of Anyscale, and Jason Lopatecki, co-founder of Arize AI, at our next technical meet-up on Wednesday Feb. 5th, for insights into building robust, scalable AI systems. Learn what it takes to bring AI applications from vision to reality. 🚀 Space is limited, sign up today: https://lnkd.in/gK9NCDcv