Mario Meir-Huber

The Rails 8 feature set perfectly complements SQLite's power in creating resilient, high-performance production apps, but still the question lingers: Can I really go all-in on #SQLite? Join Stephen Margheim at #RailsWorld as he illustrates how to leverage Rails and SQLite's full potential in your next venture, and when SQLite does and doesn't make sense for your application. https://lnkd.in/eaBstScy

52

1 Kommentar

A great read on the performance and financials of LLM fine-tuning. Really hoping the outcome is a tool that helps overcome the biggest elephant in the room, Mr. Cost! #llm #finetuning #finance #computaioncosts

2

1 Kommentar

Public Seminar: We recently had Juliane Mueller gave a talk on Surrogate model-based algorithms for expensive black-box optimization. Missed it? Don't worry you can catch all OPTIMA seminars on the OPTIMA YouTube Channel. Juli's talk https://lnkd.in/gEMTG3kS OPTIMA seminars https://lnkd.in/gi6RtPt4 #optimization #seminar #blackbox

11

1 Kommentar

Normal Computing has been collaborating with DSPy, a cutting-edge framework developed by the Stanford NLP group aimed at algorithmically optimizing language model (LM) prompts. Thomas Ahle, Head of Machine Learning at Normal Computing, was excited to highlight our team's recent contributions enabling the DSPy compiler to take advantage of typing constraints. This provides a powerful way to create robust and type-safe predictors while seamlessly integrating with the existing DSPy ecosystem. This combination of type safety, flexibility, and optimization capabilities makes Typed Predictors a valuable addition to the DSPy toolset, empowering developers to build more reliable and efficient AI pipelines. We continue on the mission to build reliable and secure AI at Normal Computing! https://lnkd.in/dScSBqgu

22

Want to get started with AI vectors in the 23ai Oracle database ? Have a look at the post we wrote, my colleague Ulrike Schwinn and I ! In this blog post we cover a few key tasks using SQL, PL/SQL, and ORDS: - Loading an ONNX embedding model into the database - Generating vector embeddings using the loaded vector embedding model - Storing vector embeddings in a table with a data type VECTOR column - Using similarity search in SQL and ORDS Feel free to give it a try with the 23ai Free version and have fun! #oracledatabase #23ai #23aifree #oraclecloudinfrastracture #vector #similaritysearch #AI #ORDS #SQL #tutorial #embeddings Blog post: https://lnkd.in/eZBDRcfu

27

💡What are the limitations of purpose-built database systems? 💡What defines a universal database? 💡How does TileDB fit into the bigger picture? Our very own Stavros Papadopoulos explores these questions and introduces the concept of the universal database as the future-proof alternative to purpose-built databases. Read more 👉 https://hubs.la/Q02trv4n0

94

The tendency of large language models to “hallucinate” continues to trouble #CIOs eyeing production use-cases – even as efforts around fine-tuning and RAG-based optimisations continue. Among those concerned about this “inconsistency” is the US Department of Defense. The Stack took a look at a new programme as well as some fresh research on #LLM validation. 👉 https://lnkd.in/eEuntxqB #artificialintelligence #nationalsecurity #llms #hallucination

2

Is tool-calling all you need? Interaction patterns in multi-agent systems Part I: Introduction For all the hype surrounding large language models (LLMs) like GPT-4, their limitations become apparent in real-world use. LLMs, by themselves, operate in isolation, processing input and generating output based solely on their training data. To overcome this, we use AI agents, essentially "wrappers" around LLMs, enabling interaction with external data and services, like web search. Think of ChatGPT: it can retrieve information from the web and incorporate it into its responses. However, even agents are bound by their predefined instructions, known as prompts. While prompts can be increasingly complex, larger prompts increase the risk of misinterpretation or sections being entirely ignored by the LLM. Multi-agent systems offer a solution by distributing tasks among multiple agents, each with specialized instructions, unlocking greater complexity and efficiency. But a crucial question arises: how do these agents effectively collaborate? This is where orchestration comes into play. Tool calling The most basic orchestration pattern is tool-calling. Imagine an agent needing data from the web; it can utilize a search engine as a "tool," providing a query and receiving relevant results. Interestingly, agents themselves can function as tools! One agent can leverage the capabilities of another, forming a chain of AI interactions. Motleycrew, an open-source framework, is notable for explicitly supporting this agents-as-tools concept. While undeniably powerful for many use cases, is tool-calling the definitive answer for multi-agent orchestration? Tool-calling, though powerful, might not be a universal solution for all multi-agent scenarios. The field is young, and other frameworks employ different approaches. This series aims to review these methods, discern their uniqueness and equivalences, and ultimately, in its final part, delve into the orchestration options we've chosen for Motleycrew, explaining why we believe they strike the right balance between power and simplicity. Intrigued by the potential of multi-agent systems and eager to delve deeper into the nuances of orchestration? Explore the full article https://lnkd.in/eQ2-jjAB for a comprehensive analysis and further insights. #motley_ai

3

🎤 Keynote announced for InfoQ Dev Summit Munich (Sep 26-27): "" AI in the Age of Climate Change"" by @Nischal HP (@nischalhp), Director of Data Science and Data Engineering @Agreena 🌍🤖 Over the last decade, AI has made incredible strides. Simultaneously, climate change has accelerated at an alarming pace. The big question remains: Can AI help combat climate change? 💡 This keynote will explore: 🌱 The contributors to climate change, with a focus on agriculture 💡 How AI can revolutionize our approach to tackling these issues 🌾 Agreena's efforts in enabling regenerative farming, carbon sequestration, and voluntary carbon markets 🤝 Supporting farmers in their mission to revive our planet AI isn't just about technology—it's about creating a sustainable future for humankind. Let's harness its power to make a real difference! 🔗 Find out more: https://bit.ly/3M9EhU1 #ClimateChange #AI #Sustainability #RegenerativeFarming #IDSMunich #SoftwareDevelopment #DevCommunity #DeveloperConference #InfoQEvents #DevTrends #InfoQDevSummit

9

🎤 Keynote announced for InfoQ Dev Summit Munich (Sep 26-27): "" AI in the Age of Climate Change"" by @Nischal HP (@nischalhp), Director of Data Science and Data Engineering @Agreena 🌍🤖 Over the last decade, AI has made incredible strides. Simultaneously, climate change has accelerated at an alarming pace. The big question remains: Can AI help combat climate change? 💡 This keynote will explore: 🌱 The contributors to climate change, with a focus on agriculture 💡 How AI can revolutionize our approach to tackling these issues 🌾 Agreena's efforts in enabling regenerative farming, carbon sequestration, and voluntary carbon markets 🤝 Supporting farmers in their mission to revive our planet AI isn't just about technology—it's about creating a sustainable future for humankind. Let's harness its power to make a real difference! 🔗 Find out more: https://bit.ly/3M9EhU1 #ClimateChange #AI #Sustainability #RegenerativeFarming #IDSMunich #SoftwareDevelopment #DevCommunity #DeveloperConference #InfoQEvents #DevTrends #InfoQDevSummit

2

AI Engineer – Rethink the system with AI and stay one step ahead now! In a new episode of #InnovationSnippets, Lara Fehr, Juliane Turzynski and Nianheng Wu talk about T-Systems' AI solution #AIEngineer, which programs and translates old programming languages into new ones - For example, 40-year-old FAME programs from the financial industry in current #Python code. It also automatically creates documentation and tests parts of the program. The AI Engineer has been used in daily work at T-Systems since 2023. If you want to unlock the full potential of your business with the simplicity and efficiency of AI, please contact us and book a workshop at [email protected]. #InnovationSnippets #innotastic #tsystems #innovation #AI #Metaverse

59

1 Kommentar

What do you need to successfully deploy LLMs in your enterprise? Andrei Dmitrenko, Carlos Parés Pulido and Meret Ringwald are running the workshop “Building Agile Data Foundations for Data Mesh & Generative AI” at the Swiss Conference on Data Science | SDS Find more information here: https://lnkd.in/d_3fXhqc #LLMs #DataVault #SDS2024 #IEEEConference #doneai

28

🎉 aGrUM/pyAgrum 1.16.0 is now available! 🎉 The latest version of aGrUM/pyAgrum is packed with new features and updates: 🔹 Enhanced Structure Learning: Now you can use NoParentNode and NoChildNode constraints to fine-tune 👩‍🔬 Bayesian network learning. 🔹 New in Causal Inference: We’ve added (conditional) Average Causal Effect/Average Treatment Effect (ACE/ATE) estimation with meta learners. Curious 🧐 to see it in action? Check out our example notebooks: Causal Effect Estimation (https://lnkd.in/eA44aHW4) Examples of Causal Effect Estimation from Data (https://lnkd.in/ezmkPwif) 👋😢 Farewell, Python 3.9: With this release, we also say goodbye to Python 3.9 support. 💻 Download the latest version to explore all the new features! For the complete list of changes, visit the changelog here: https://lnkd.in/etNV7wR4 #pyAgrum #Python #BayesianNetworks #AI #DataScience

23

🧠 The myth: More data makes an #AI model smarter and therefore more accurate 💪 The truth: It’s not just about data—it’s about #parameters With the around OpenAI’s #GPT4o it might seem that feeding more data into these models makes them smarter, the real game-changer is the number of parameters they have like Meta #Llama 3.1 with 450b parameters 💡 More parameters mean these models can process and understand data in more sophisticated ways, leading to increased accuracy and the appearance of being “smarter.” More parameters = higher trustworthiness, even with more focused data like the proprietary #data in your organization In the world of conversational user interfaces becoming the norm, more parameters will be crucial for a seamless experience. While yet too slow and too expensive for real fast interactions between humans and machines, with #quantumcomputing we will move in a new area of responsible #UI #design #AI #MachineLearning #LLM #conversationaldesign #hmi #Innovation

5

1 Kommentar

When data is systematically collected and stored, this creates opportunities for misuse, whether by companies that gather it, the state, or individuals who hack into systems. 👁️‍🗨️ 🚨👇 Therefore, it is crucial for people to understand, at least conceptually, how data can be used and what tools and techniques in #AI and #machinelearning are available. Without this basic knowledge and understanding, all we can do is speculate about how these technologies might benefit or harm us. 🔦 🤷‍♀️ 🤷 Just like we already start preparing children for reading and writing in kindergarten by training their motor skills, drawing, and painting, the same could be done with computer and #dataliteracy. 💯 🍊 But simply incorporating digital competencies into subjects, as envisioned by the current Slovenian curriculum reform, is not enough. Digital competencies primarily focus on the use of digital tools, and cannot be equated with the knowledge and skills in computerscience and informatics, as there is still a lack of algorithmic thinking and foundational understanding of 🎯 programming, 🎯 algorithms, 🎯 computer systems, 🎯 machinelearning, 🎯 #datascience, and more. Without this basic knowledge, individuals, societies, and economies will struggle to keep pace, make sense of, and remain competitive in today’s world. 🌍 🧭 To bring this knowledge into Slovene primary and secondary schools, our team is involved in several large experimental projects, among them #Katarina, which involves 10 Slovene secondary schools. More about the project here 👉 https://lnkd.in/dXiNSFTB Blaz Zupan, Izjava dneva 👉 https://lnkd.in/dUNuXPWq More on this topic 👉 https://lnkd.in/dH2Y2yts #education

19