Plum ’s Post

🔶 BIG ANNOUNCEMENT 🔶 Google's Chief Decision Scientist, Cassie Kozyrkov, is joining Yann LeCun as a keynote speaker at AI Native on September 7th! Don't miss Cassie's opening address on The Secret to AI Innovation! REGISTER: https://lnkd.in/gxtWva9y She'll answer key AI questions circling the ecosystem:  🔸 Does AI free us to create groundbreaking solutions or are such notions mere hype?   🔸 Which jobs is AI designed to automate and will those careers disappear?   🔸 Is Generative AI truly something new, or is it just buzz?   🔸 What's one piece of advice that every investor should follow when diving into this space?

Join 5k+ engineers who are already members of the Computer Vision Meetup.  Scroll down to the bottom of this page to join the Meetup nearest you. https://lnkd.in/gN4akVfb We host two virtual Meetups every month in APAC, EMEA and North American-friendly time zones. #computervision #ai #artificialintelligence #machinevision #machinelearning #datascience

Join 5k+ engineers who are already members of the Computer Vision Meetup.  Scroll down to the bottom of this page to join the Meetup nearest you. https://lnkd.in/gN4akVfb We host two virtual Meetups every month in APAC, EMEA and North American-friendly time zones. #computervision #ai #artificialintelligence #machinevision #machinelearning #datascience

Recently at Gemini meet-up: Bridging Perspectives with Searce Inc & Google, I was asked a very interesting question… If Chargeup had the power to invite a superhero, what would it be? The only one I could think of was an Indian superhero - the Hanuman. Lord Hanuman possessed extraordinary abilities—he could grow as large as a mountain or shrink as small as a bee, he could walk, swim & fly and adapt to any situation, just like a modern-age super AI and above all Lord Hanuman, with all his powers always followed his master Lord Ram. I wish, Artificial intelligence will continue to empower humans and never overpower… The discussions at the event were enriched by perspectives from Ketan Dewan and Vishal Sharma, moderated by Pulkita Marwah emphasizing the profound role of AI across industries. At @Chargeup, we're leveraging AI/ML to shift driver behavior from "Reactive" to "Predictive", unlocking immense opportunities for the future #AI #chargeup Varun Goenka Anmol Chouksey Chetan Guglani Riya J. Shivansh Sharma

Llama 2 from Meta AI is taking the open-source LLM by storm. Along with the release, Meta also published a research paper with tons of technical details and insights. Sharing some of my notes: - Llama 2 outperforms other open-sourced models (MPT, Falcon, Vicuna, etc.) by large margins and comes close to GPT-3.5 in performance. - A round of full training of the 70B variant consumed over 1.7 million GPU-hours, which would cost ~$10M, assuming AWS rate for A100 machines. - Tens of thousands high-quality examples are good enough for SFT - quality matters much more than quantity. - RLHF again proved highly effective, despite the belief that SFT data has more information density. There are great discussions on this topic in the research paper. You can find my full notes with charts here - https://lnkd.in/gxjz9yH7 And the research paper - https://lnkd.in/gSa3Zhin

Interested in #ai alignment - join in the learnings and showcases of new methods and techniques to test your models. Sponsored by Apollo Research Date : Friday November 24th 19:00 UTC to Sunday November 26th 2023 You can apply as a team of 5 people but if you don’t have a team one can be assigned for you. https://lnkd.in/e9y2ButD #data #it #business

Google DeepMind hopes to change that with a new tool that uses deep learning to dramatically speed up the process of discovering new materials. Called graphical networks for material exploration (GNoME), the technology has already been used to predict structures for 2.2 million new materials, of which more than 700 have gone on to be created in the lab and are now being tested. It is described in a paper published in Nature today. #Google #deepmind #material #ai #ia #tecnologia #gnome

Was catching up with this event while waiting for my flight and Just found myself recommending this over lunch at the OECD.AI's 9th AIGO meeting.. and thought why not share with everyone: "This video covers three fireside chat conversations with HAI leadership and faculty at the HAI at Five Conference. In the fireside discussion "Removing Impediments to a Robust AI Innovative Ecosystem" HAI Co-Director Fei-Fei Li and venture capitalist Marc Andreessen debate open- versus closed-source AI, describe what a healthy AI ecosystem looks like, consider industry’s role in these decisions, and define a realistic pathway to applied AI. In the session "Guardrails and Regulation for a Vibrant AI Ecosystem" HAI Deputy Director Russell Wald and Mariano-Florentino (Tino) Cuéllar, president of the Carnegie Endowment for International Peace and former California Supreme Court Justice, explore the potential impact of AGI on society, delving into the intricate intersection of technology, philosophy, and humanity's quest for intelligence. In the chat "Accelerating Science with AI" HAI Co-Director Fei-Fei Li talks with computer scientist and industry leader David Siegel about how this technology is advancing science as we know it — in robotic learning, neuroscience, and in our work in higher education — and requiring a more interdisciplinary approach. This video also features incoming Stanford University President, Jonathan Levin, sharing a brief keynote on "AI at Stanford." Speakers: Marc Andreessen , Co-Founder and General Partner, Andreessen Horowitz Fei-Fei Li, Sequoia Professor, Computer Science Department; Denning Co-Director of Stanford HAI Mariano-Florentino (Tino) Cuéllar, President, Carnegie Endowment for International Peace; Visiting Professor, Stanford Law School; Advisory Council Member, Stanford HAI Russell Wald, Deputy Director, Stanford HAI Jonathan Levin, Philip H. Knight Professor and Dean of Stanford Graduate School of Business David Siegel, Co-Founder and Co-Chairman, Two Sigma" https://lnkd.in/dmjdqMCd

Join 5k+ engineers who are already members of the Computer Vision Meetup.  Scroll down to the bottom of this page to join the Meetup nearest you. https://lnkd.in/gN4akVfb We host two virtual Meetups every month in APAC, EMEA and North American-friendly time zones. #computervision #ai #artificialintelligence #machinevision #machinelearning #datascience

If you are deep into AI and are not subscribed to AlphaSignal, you really should get signed up. It’s an email newsletter with summaries of the top recent research papers. Most are highly technical, beyond my meager mental capacity, but getting a sense of where research is today and where capabilities are headed is priceless. The most recent paper that was highlighted is “Textbooks are all you need” https://lnkd.in/gAncZvDJ The raw paper synopsis is a little hard to parse but AlphaSignal’s newsletter provides a more accessible version and additional context. AlphaSignal