Silo AI

Gathering the country's most notable AI experts, the AI Finland network of the Teknologiateollisuus - Technology Industries of Finland has established an advisory board as part of an ambitious mission to position Finland as a global leader in AI development and application. Board member, and Silo AI co-founder Peter Sarlin played a key role in assembling the group, stating; “Artificial intelligence is not just technology – it has the potential to strengthen Finland's competitiveness and act as a broader societal force for change. Success in this requires not only data and algorithms but also an understanding of service and process development. AI Finland strengthens the broader ecosystem that helps companies commercialize innovations. I am part of the Advisory Board because I believe that through such collaboration, we can accelerate development and achieve significant impact.” All the nationally and internationally recognized members of the AI Finland Advisory Board include: 🔹 Kaj-Mikael Björk – Director of Center for Intelligent Computing, Turun yliopisto - University of Turku 🔹 Bengt Holmstrom – Professor, Board professional 🔹 Timo Harakka – Member of Parliament, author, journalist 🔹 Maija Hovila – Chief Data & AI Strategist, Futurice 🔹 Juha Inberg – CTO, Ponsse 🔹 Pekka Manninen – Director, Science and Tech, CSC - IT Center for Science 🔹 Petri Myllymäki – Professor, Vice-Director, Finnish Center for Artificial Intelligence 🔹 Liisa Peltonen – General Manager, Surface Finland, Microsoft 🔹 Peter Sarlin – CEO and Co-founder, Silo AI 🔹 Risto Siilasmaa – Founder, Chair of the Board, WithSecure 🔹 Henry Tirri – CTO, Nokia 🔹 Tommi Vilkamo – Director, RELEX Solutions For more information, check the links in the comments. #AIforPeople

Modular, open-source based software stacks for generative AI workloads give companies flexibility when choosing component frameworks, and increased control to choose models that best suit specific requirements. ROCm is designed to easily connect to frameworks such as PyTorch and model hubs like Hugging Face. GenAI development platforms bring together all building blocks needed for building and deploying genAI applications, inviting AI communities across the world to contribute to the global AI ecosystem. #TogetherWeAdvance

Europe's tech future hinges on more than just AI. While the tech sector is the primary driver for the widening productivity gap, the region must also address a more fundamental challenge: Europe needs significantly more scalable digital companies to truly capitalize on AI's potential. To seize the opportunity, Europe must: 🔷 Encourage institutional investors 🔷 Simplify the regulatory landscape 🔷 Encourage public and private funding 🔷 Further foster links between academia and industry Read more in our blog, link in comments. #AIforPeople #Europe

Not only did Financial Times list AMD’s CEO, Dr. Lisa Su, among the 25 most influential women of 2024, TIME chose her as the CEO of the year! Glass ceiling shattering leaders are needed as role models in the AI industry to ensure fair and inclusive AI. As the FT article declares: “She is a role model, as well as an example of perseverance and strength who inspires us all.” and “Su has made pivotal contributions to new semiconductor technologies that have introduced new industry standards”. Read more about Dr. Su and the other 24 inspiring leaders in the articles below. #TogetherWeAdvance

Congratulations to CSC - IT Center for Science's LUMI supercomputer and others on being selected as one of Europe's first AI Factories by the EuroHPC Joint Undertaking (EuroHPC JU)! 🚀 LUMI has powered the development of our Viking and Poro model families through a customized open source training framework. The new LUMI AI factory will be one of seven new AI Factories representing a total €1.5B investment. The initiative will more than double EuroHPC's compute capacity, addressing a critical need for European AI development. The LUMI AI Factory will feature an experimental platform for developing and testing innovative AI models, with five countries already joining the consortium. The facility will provide access to massive computing power that startups, industry, and researchers need to develop their AI models - from European large language models to specialized vertical AI models focusing on specific sectors and domains. As one of five brand-new world-class AI-optimized supercomputers in the initiative, LUMI AI Factory builds on the success of the current AMD MI250X-powered LUMI system, which has proven its capabilities in scaling AI training. This new dedicated AI optimization represents exactly the kind of infrastructure Europe needs to drive AI innovation forward. #AIforPeople #TogetherWeAdvance #AIFactories #DigitalEU Barcelona Supercomputing Center CINECA NCSR "DEMOKRITOS" LuxProvide University of Stuttgart HLRS - High-Performance Computing Center Stuttgart GRNET - Greek Research & Technology Network Linköping University Leibniz Supercomputing Centre GWDG - Gesellschaft für wissenschaftliche Datenverarbeitung mbH Göttingen Karlsruhe Institute of Technology (KIT) SICOS BW National Technical University of Athens Growthfund, the National Fund of Greece Luxembourg Institute of Science and Technology (LIST) Luxembourg National Data Service (LNDS) Luxinnovation GIE University of Luxembourg Fundação para a Ciência e a Tecnologia (FCT) TÜBİTAK NAISS – National Academic Infrastructure for Supercomputing in Sweden RISE Research Institutes of Sweden

The Thirty-Eighth Annual Conference on Neural Information Processing Systems #NeurIPS2024 is starting today in Vancouver! Come meet AMD and ELLIS - European Laboratory for Learning and Intelligent Systems at: 🔷 A workshop on accelerating AI training using AMD Instinct GPUs by Emad Barsoum & Zicheng Liu today, Tuesday 10th. 🔷 The ELLIS Open Source AI Meetup on Thursday 12th. Hope to see you there! NeurIPS #TogetherWeAdvance #AIforPeople

As reliance on AI increases, compute efficiency becomes an ever more important aspect of responsible AI. Several measures can be taken in order to optimize energy efficiency. Here is a list of measures to consider: 🔹Optimize GPU allocation for on-demand deployment. Different types of clustering enables scaling compute based on need, whether clustering nodes for better work sharing or clustering for faster job completion. 🔹Choose the simplest model possible that can accomplish the task at hand. In addition to being more energy efficient, smaller models also increase explainability. 🔹Make use of fine-tuning, distillation and quantization to minimize model size, while still considering the task at hand as to not compromise performance. Smaller models require less energy. 🔹Be mindful of programming language choices. There are significant differences between the languages in terms of energy consumption. No single solution can fully address AI's energy and sustainability challenges. However, taking a holistic approach to AI deployment enables better decisions that compound into meaningful impact. Flexible platforms for training, deployment and maintenance improve sustainability and are crucial for sustainability and responsible AI.

Open-source LLMs that not only excel in linguistic performance but are also attuned to local values and cultures enable companies to build better AI-driven products. For Doctrin, this means enabling interactions with more empathy between healthcare professionals and patients. A fine-tuned version of the Viking 7B model in Doctrin’s care navigation platform supports the use of more empathic language and saves time in documentation and digital communication. This frees up time for human to human interactions, leaving patients feeling seen and heard. Studies show positive experiences of interactions with healthcare systems can result in better health outcomes and increased patient safety. Viking 7B is part of the Viking model family, focusing on Nordic languages. All Viking models are open-source under the Apache 2.0 license, built with optimized training approaches and architecture. The Viking models are trained on LUMI, the most powerful supercomputer in Europe, using a new customized open-source training framework, with simultaneous utilization of up to 1024 AMD Instinct MI250X GPUs, proving the ability to train LLMs at scale. Learn more about our Viking model family through the link in the comment section below.

Hear hear. Reka AI CEO Dani on a mission to build multimodal AI that can be deployed anywhere: “Our partnership with AMD has unlocked a lot of possibilities for us and our customers as well.” 🚀 Hear from Reka AI, a team with a Google DeepMind, Google, Baidu, Inc. and Meta pedigree and investments from DST Global, Radical Ventures, Snowflake and Nat Friedman, on "deploying across AMD platforms from high-performance cloud GPUs, enabling adoption for our customers all the way to even AI workstations and PCs". #AIforPeople #TogetherWeAdvance

While AI dominated discussions at Slush last week, the real question is how companies are turning potential into business value. We want to understand how Nordic companies are creating impact with AI. What AI technologies are companies using, and for what? The Nordic State of AI report sheds light on these, and many other interesting questions. Help shape the Nordic AI ecosystem by sharing your insights. The survey is aimed at leaders and senior tech staff of Nordic companies of all sizes and industries. Your input will help create a comprehensive view of how AI is transforming Nordic businesses The report is produced by Silo AI in collaboration with AI Finland, aiming to strengthen the Nordic tech ecosystem. Complete the survey: https://lnkd.in/dmwbY-Tv #AIforPeople #NSOAI4