NXAI

Thanks AWS for having us at the GenAI Loft in Berlin this year. Great experience, interesting discussions with the audience. Our CEO Albert Ortig presented our approach, when it comes to xLSTM and Industrial Use Cases in robotics, on the shopfloor or on edge devices.

Our paper recommendation for the weekend 📝 X-IL: Exploring the Design Space of Imitation Learning Policies by Xiaogang Jia et al. 2025 Karlsruher Institut für Technologie (KIT) Whole paper on Arxiv: https://lnkd.in/e58NqnJf "Designing modern imitation learning (IL) policies requires making numerous decisions, including the selection of feature encoding, architecture, policy representation, and more. As the field rapidly advances, the range of available options continues to grow, creating a vast and largely unexplored design space for IL policies. In this work, we present X-IL, an accessible open-source framework designed to systematically explore this design space. The framework's modular design enables seamless swapping of policy components, such as backbones (e.g., Transformer, Mamba, #xLSTM) and policy optimization techniques (e.g., Score-matching, Flow-matching). This flexibility facilitates comprehensive experimentation and has led to the discovery of novel policy configurations that outperform existing methods on recent #robot learning benchmarks. Our experiments demonstrate not only significant performance gains but also provide valuable insights into the strengths and weaknesses of various design choices. This study serves as both a practical reference for practitioners and a foundation for guiding future research in imitation learning." What about xLSTM? - LIBERO: Specifically, xLSTM shows great potential in both 20% and 100% data settings - RoboCasa: Specifically, using xLSTM-based models, we achieved a higher average success rate of 53.6%, compared to 40.0% of BC-Transformer - Point Clouds: Experimental results demonstrate that incorporating both modalities significantly enhances performance, particularly for the xLSTM-based model, which achieves a 60.9% success rate Thanks to the research team around Gerhard Neumann

We are very proud of our first Spin-Off Emmi AI 🚀 The team will focus on Industrial Simulation and will achieve tremendous results in the field - for sure. Keep up the great work. We are very proud of you. More to come soon.

We are sure: When it comes to Industrial AI, xLSTM will play a crucial role. But when you want to enter the industrial sector, you have to present your solutions at HANNOVER MESSE. Our Chief Scientist Sepp Hochreiter gave a talk at the traditional Press Preview with journalist from around the world and shared some insights, when it comes to Timeseries, Vision #xLSTM and Large Action Models or Bio xLSTM for Life Science. Besides that, Hochreiter explained: The bitter lesson is over - we need to find new approaches, when it comes to #AI in the industrial sector. We enter now a new phase of AI and need new solutions, we need to rethink architectures. Thanks for having us Dr. Jochen Köckler - it is always a great show. We are looking forward to see and hear more at the trade fair show in March/April.

xLSTM live: Watch Sepp Hochreiter´s presentation at the Hannover Messe Press Preview. He is going to discuss Industrial AI Use Cases and how the industrial sector will benefit from xLSTM as a new architecture. After his presentation there will be a discussion with Fatma Kocer-Poyraz about the potential of AI. 🍿 https://lnkd.in/eWmCxKev

We are so back - Sepp Hochreiter about xLSTM at the Machine Learning Street Talk (MLST) - thanks guys it is a great discussion about the future of #AI Check out the whole episode https://lnkd.in/ehHaTfxT

A whole evening just xLSTM - we are looking forward to the Meetup “Autonomous Driving, Robots & AI” in Munich. Thanks to Felix Friedmann and his team. And they do this on a voluntary basis. Chapeau! Thanks to Siemens Industry and especially Boris Scharinger, who will be our host. We are looking forward to the talks, but above all we want to know what the community is doing with xLSTM. And we also bring some ideas with us.

Sepp Hochreiter at NVIDIA GTC 2025 We’re thrilled to announce that Sepp Hochreiter will be a speaker at NVIDIA GTC, presenting: “Unlocking the Future of LLMs: High-Efficiency xLSTM Architectures for Scalable Performance” What to Expect: 💡 How xLSTM challenges the dominance of transformer architectures 💡 Why efficiency matters: lower energy consumption, faster inference, and reduced compute costs 💡 The real-world impact of xLSTM for LLMs, robotics, industrial AI, and embedded systems 💡 Scaling beyond traditional limitations to make AI more accessible and sustainable xLSTM is proving that AI doesn’t have to be bigger to be better — it has to be smarter. Don’t miss Sepp’s session at NVIDIA GTC 2025! #GTC2025 #xLSTM #AI #SeppHochreiter #LLMs #ScalableAI #IndustrialAI #NXAI #NVIDIA NVIDIA

In this special episode of the Eye on AI podcast, Sepp Hochreiter, the inventor of Long Short-Term Memory (LSTM) networks, joins Craig Smith to discuss the profound impact of LSTMs and of course #xLSTM on artificial intelligence, from language models to real-time robotics. Sepp reflects on the early days of LSTM development, sharing insights into his collaboration with Jürgen Schmidhuber and the challenges they faced in gaining recognition for their groundbreaking work. He explains how LSTMs became the foundation for technologies used by giants like Amazon, Apple, and Google, and how they paved the way for modern advancements like transformers. Topics include: - The origin story of LSTMs and their unique architecture. - Why LSTMs were crucial for sequence data like speech and text. - The rise of transformers and how they compare to LSTMs. - Real-time robotics: using xLSTMs to build energy-efficient, autonomous systems. The next big challenges for AI and robotics in the era of generative AI. Sepp also shares his optimistic vision for the future of AI, emphasizing the importance of efficient, scalable models and their potential to revolutionize industries from healthcare to autonomous vehicles. https://lnkd.in/gMkQfq4a

Advancing AI Research: 3 xLSTM Papers Accepted at ICLR 2025 We are excited to share that three papers on xLSTM have been accepted at ICLR 2025, one of the world’s top machine learning conferences. 🔍 Vision xLSTM: Unlocking new possibilities for computer vision tasks by capturing long-range dependencies ⚡ Flash-RNN: A novel RNN design combining the speed of recurrent networks with transformer efficiency 🧬 Bio-xLSTM: Enabling breakthroughs in biology and chemistry with generative modeling for DNA, proteins, and molecules We’re looking forward to presenting these contributions in Singapore and connecting with the global AI community. What are you most excited about at ICLR 2025? Let us know! #xLSTM #ICLR2025 #MachineLearning #AIResearch #ArtificialIntelligence Singapore EXPO, Sepp Hochreiter, Günter Klambauer, Korbinian Pöppel, Maximilian Beck, Benedikt Alkin