IBM Research

🐝 - Introducing #BeeAI, an experimental platform designed to help developers discover, run, and compose open-source AI agents from any framework and in any coding language: https://ibm.co/4hIGxi0 🗣️ - Calling for contributions from the community, leaders of the project, Kate Blair and Ismael Faro from IBM Research unveiled the newly redesigned platform at the 2025 AI Developer Conference. 📚 - Bookmark the project on GitHub here: https://ibm.co/4hkRKp8 ----- #IBM #Research #AI #Agents #NLP #opensource #AIDEV2025

Can you build an AI agent that performs in-depth research while remaining cost effective? The answer is yes. We'll show you how with IBM Granite ⤵

We are excited to announce a major milestone release (V1.0) of #TerraTorch - our #PyTorch-based fine-tuning and inference toolkit for leveraging and benchmarking foundation models in (geospatial) computer vision. It supports the Prithvi model family, developed by #IBM in collaboration with #NASA, along with many other foundation models. Among numerous improvements, V1.0 introduces support for Hyperparameter Tuning, Neural Architecture Search, and Benchmarking through TerraTorch-Iterate. TerraTorch is the reference implementation for #GeoBench. https://lnkd.in/ekTTCEy9

Qiskit SDK v2.0 is here! https://ibm.co/43Cl2ML The latest release of the Qiskit SDK brings exciting new features and performance improvements that set the foundation for even more powerful tools and capabilities to come. ⚙️ C API: We’ve introduced a C API for the ‘SparseObservable’ class, laying the groundwork for the much more extensive C interface we'll be building over the minor releases to come. We believe Qiskit's C API will facilitate closer collaborations with the HPC community and allow us to extend Qiskit in ways that were previously impossible. 🤸♂️ `stretch` variables: We've added `stretch` variables for `Delay` instructions, a concept originally introduced in OpenQASM 3. This capability will soon allow you to exercise much more nuanced control over instruction timing in dynamic circuits. 📦 Boxes: Another feature, the new `box` instruction, allows you to more easily group instructions in the Qiskit SDK for later processing and custom scheduling. Instructions can be grouped with downstream data attached, passed through transpilation, and transmitted up and down the execution stack. 📈 Performance improvements: Qiskit SDK v2.0 builds on the extensive work done throughout the v1.x series, which helped to make Qiskit the most performant quantum software in the world. Many of the boosts to speed and efficiency we saw in the v1.x era were powered by our work refactoring Qiskit's internals in the high-performance Rust programming language, and we've continued those efforts in v2.0. The latest SDK release demonstrates a 2x speedup in circuit construction and ~20% faster transpilation in our open-source Benchpress benchmarking suite. We’re laser-focused on maintaining Qiskit’s position as the world's most performant quantum SDK, and these new features and improvements help us to do just that. Head over to the IBM Quantum blog linked above for a full deep dive on Qiskit 2.0.

A clue in today's edition of The Crossword by The New York Times "Maker of the world's largest quantum computer".... we wonder who it can 🐝 ?

For over two decades, the collaboration between IBM and TOKYO ELECTRON LIMITED has been a driving force in semiconductor innovation. Today, we're excited to announce the next chapter, 𝙖 𝙛𝙞𝙫𝙚-𝙮𝙚𝙖𝙧 𝙖𝙜𝙧𝙚𝙚𝙢𝙚𝙣𝙩 𝙛𝙤𝙘𝙪𝙨𝙚𝙙 𝙤𝙣 𝙥𝙤𝙬𝙚𝙧𝙞𝙣𝙜 𝙩𝙝𝙚 𝙖𝙜𝙚 𝙤𝙛 𝙜𝙚𝙣𝙚𝙧𝙖𝙩𝙞𝙫𝙚 𝘼𝙄. This renewed partnership will leverage our combined expertise to push the boundaries of semiconductor nodes and chiplet architectures. Together, we're developing hardware for the future of AI, enabling enhanced performance and energy efficiency in tomorrow’s electronic devices. Learn more: https://ibm.co/4clT9Lg --- #IBM #Research #AI #Semiconductors #Chips

In this week’s edition, we are learning how IBM is expanding quantum education by the numbers. We’ll take a closer look at the progress of an international quantum education initiative, announce the first two IBM Quantum Research Chair positions, and hear more from IBM’s CEO, Arvind Krishna, on the future of AI and quantum computing. For more on the latest news, be sure to read here and subscribe:

𝗪𝗵𝗮𝘁'𝘀 𝗕𝗲𝗲𝗔𝗜? This experimental platform from IBM Research lets developers run open-source AI agents from any framework and in any coding language. 𝗪𝗵𝘆 𝗱𝗼𝗲𝘀 𝘁𝗵𝗶𝘀 𝗺𝗮𝘁𝘁𝗲𝗿? This makes it easy for developers to run popular open-source AI agents from different frameworks and build specialized agents of their own. Regardless of where agents originate, or what code they were written in, they can be easily configured to work alone, or with AI teammates, through the IBM Research-designed agent communication protocol (ACP). 𝗟𝗲𝗮𝗿𝗻 𝗺𝗼𝗿𝗲 𝗵𝗲𝗿𝗲 ⤵ BeeAI overview → https://ibm.co/3QWQKwv Check out the project on Github → https://ibm.co/3FZJuxy Read the draft ACP proposal → https://ibm.co/3FKvOq8

Last week, the first of two 𝗦𝗮𝗳𝗲𝗿 𝗠𝗮𝘁𝗲𝗿𝗶𝗮𝗹𝘀 𝗦𝘂𝗺𝗺𝗶𝘁𝘀 took place at IBM 's Almaden innovation lab, welcoming executives, researchers, and industry experts to an interactive discussion examining the challenges of #PFAS (“Forever Chemicals”) proliferation and the opportunity to address them with new AI technology across three fronts: – 𝗜𝗱𝗲𝗻𝘁𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻: What PFAS exposure do I have in my products, processes and supply chain? – 𝗦𝘂𝗯𝘀𝘁𝗶𝘁𝘂𝘁𝗶𝗼𝗻: How can I find existing, suitable “safer” alternatives? – 𝗗𝗶𝘀𝗰𝗼𝘃𝗲𝗿𝘆: If no substitutes exist, how can I develop new replacements? The strategic importance of PFAS identification in manufacturing is now apparent. It’s a massive, complicated challenge where AI can play a crucial role in 𝗿𝗶𝘀𝗸 𝗮𝘀𝘀𝗲𝘀𝘀𝗺𝗲𝗻𝘁 and 𝗺𝗶𝘁𝗶𝗴𝗮𝘁𝗶𝗼𝗻. The next Safer Materials Workshop will be this Thursday, April 3, 2025, from 9 am to 4 pm (followed by an Apéro riche), at IBM Research — Zurich in Rüschlikon, Switzerland. Details here: https://ibm.co/4iPGthS 📸 - Photos: Pedro Pavanato/IBM --- #IBM #Research #AI #AcceleratedDiscovery

🚀 I’m excited to share our newly published paper on AI-driven prediction of influenza antibody activity! This work is the result of a successful collaboration between our IBM Research team and the Cleveland Clinic Florida Research and Innovation Center team led by Giuseppe Andrea Sautto, PhD, MBA. 🎉 Huge kudos to the incredible researchers behind this study: Ella Barkan, Ibrahim Siddiqui, Kevin Jose Cheng, Alex Golts, Yoel Shoshan, Jeffrey Weber, and Yailin Campos Mota. 🙏 A special thanks to our IBM managers Efrat Hexter, Wendy Cornell, Michal Rosen-Zvi, and Jianying Hu for their support in making this study happen! #IBM #IBMResearch #AI #ClevelandClinic