Open Source AI Workshops

Um unser Programm mit voller Funktionalität zu sehen, wie Zeitzonen und persönlicher Programmgestaltung, schalte bitte JavaScript an und klicke hier.
09:00
09:00
480min
Einführung zu Open Source AI
Gygli Marcel

In diesem Workshop werden die Grundlagen von AI vermittelt. Wie funktioniert (Generative) KI, was sind Transformer Modelle und wie werden basierend darauf Large Language Modelle (LLMs) entwickelt.
Der Workshop versucht so gut wie möglich ohne Mathematik auszukommen, sondern die Konzepte so weit zu vermitteln, dass ein Verständnis dafür entsteht.

In einem zweiten Teil wird vermittelt wie LLMs auf lokalen Geräten unter Einhaltung des Datenschutzes angewendet werden können. Dies machen wir anhand des Beispiels von LM-Studio, welches die Verwendung von LLMs auf lokalen Geräten ermöglicht.

Raum B
09:00
480min
How hard can it be? Implement your own anomaly detection algorithms from scratch.
Sergio Solorzano Rocha

Anomaly detection is one of the most practical applications of machine learning and statistics. It is relevant in finance, manufacturing, software operations, healthcare, industrial operations and many more.

During the workshop you will learn the principles and basics techniques of anomaly detection in a hands on fashion. You will code, benchmark and tests from scratch well known algorithms like Isolation Forests, ECOD, mahalanobis distance and PCA.

>> Presentation

Raum C
09:00
480min
Introduction to Large Language Models
Joel Niklaus

In this workshop we discuss the This comprehensive workshop guides participants through the technical foundations of large language models, beginning with neural networks and advancing to sophisticated concepts like transfer learning and instruction tuning. The curriculum covers essential prompting techniques and strategies, helping participants understand how to effectively communicate with language models. Participants learn about model alignment through RLHF and gain awareness of current limitations while exploring practical tools like retrieval and API integration. The workshop emphasizes hands-on learning, combining theoretical knowledge with practical exercises to ensure participants can apply their understanding in real-world scenarios.

Raum A
09:00
480min
LLM-RAG from Scratch: Crafting Intelligent AI Retrieval Systems
Ornella Vaccarelli

Embark on a hands-on journey to build your very own Retrieval-Augmented Generation (LLM-RAG) system from scratch using open source AI models and cutting-edge tools. In this workshop, you’ll learn how to integrate large language models with retrieval pipelines and dive into vast repositories of data, extracting the precise information you need to generate clear, accurate, and contextually rich responses.

>> Presentation

Raum D
09:00
480min
Reliable Multi-Agent Systems for real-world use: Introduction and Hands-On with diskurs
Flurin Gishamer

This workshop will teach you how to use open-source AI models and frameworks to design and implement reliable multi-agent systems for real-world applications. We’ll begin by covering key concepts (including free-form interaction (chat) vs. workflow automation), then explore the architecture and implementation using the open-source framework diskurs, and conclude with an introduction to testing multi-agent systems. By the end, you’ll have the theoretical background and practical experience to develop robust multi-agent applications and choose the right open source AI models for your use case.

>> Get your Ticket now!

Raum E
09:00
09:00
480min
AI Agents in Action: Building Smart, Open-Source LLM Workflows (Room H055)
Luca Rolshoven, Flurin Gishamer

AI-powered autonomous agents are changing the way we interact with large language models (LLMs). In this hands-on workshop, you will learn how to build and customize AI agents using open-source models and frameworks. We will explore how these agents reason, plan, and execute tasks, and how they can be integrated with external tools or run using local models (e.g., via LM Studio or ollama).

This workshop is designed for developers, researchers, and AI enthusiasts who want to build intelligent, privacy-friendly LLM applications. No prior experience with AI agents is required—just basic Python knowledge and a curiosity to experiment!

--- What You’ll Learn ---
✅ What AI agents are and how they interact with LLMs
✅ How to build AI-driven workflows using open-source frameworks
✅ How to run agents on local language models for privacy-friendly solutions
✅ How to integrate external tools and APIs into agent reasoning
✅ Bonus: If there’s interest, we can touch on synthetic data generation as a tool for testing and evaluation

>> Presentation

Raum A
09:00
480min
Einführung Open Source AI für Entscheidungstragende - mit Special Guest Leandro von Werra, Hugging Face (Room H210)
Matthias Stürmer, Leandro von Werra

Dieser Workshop vermittelt Entscheidungstragenden eine fundierte Einführung in die Entwicklung und den Betrieb von Open Source KI rund um technische, rechtliche und praktische Aspekte. Neben den Vorteilen wie Kosteneffizienz und digitale Souveränität werden auch Herausforderungen wie technische Hürden und juristische Fragen beleuchtet, um Führungskräften, IT-Verantwortlichen und Innovationsmanagern fundierte Entscheidungsgrundlagen zu bieten. Ausserdem gewährt Leandro von Werra, Head of Research bei Hugging Face, einen spannenden Einblick in aktuellste AI Tech Trends und beantwortet Fragen auf gut verständliche Art.

Raum D
09:00
480min
Einführung in Natural Language Processing (Gastvortrag bei Matthias Stürmer in Raum H210)
Veton Matoshi

Dieser Workshop bietet eine praxisorientierte Einführung in die Verarbeitung natürlicher Sprache (Natural Language Processing, NLP). Die Teilnehmenden lernen grundlegende Konzepte, Methoden und aktuelle Entwicklungen kennen. Neben der Theorie stehen praktische Übungen mit modernen NLP-Tools im Fokus. Am Ende des Workshops sind die Teilnehmenden in der Lage, einfache NLP-Modelle anzuwenden und eigene kleine Projekte umzusetzen.

Raum B
09:00
480min
Open Source AI Bootcamp (Room H213)
Siddhartha Singh

The workshop will focus on the open-source ecosystem that is helping make progress in the modern machine learning development. Large Language Models will be an important part of the workshop since they are everywhere now. However, it will not be limited to LLMs.

We will also cover many concepts (e.g. quantization, distributed training etc.) underlying the tools. We will be using all these tools to work on various use-cases: Retrieval Augmented Generation (RAG) Systems, making large language models small enough to be usable in your local machines, deploying the models, among others.

Some of the tools we will be looking into are: Huggingface, open-source models like Llama family, Phi3 family etc., python modules like streamlit, transformers etc., vector databases like ChromaDB, frameworks like Llama-index, model hosting services like together ai among many others. We will also go into nuances of what "open-source" even means in the age of LLMs.

You do not need to know any machine learning but the workshop will assume that you are able to code (hopefully in python). You must bring your machines to the workshop. We will send instructions before the workshop for the setup required before attending the workshop.

>> Präsentation!

Raum C
09:00
480min
docling: About Get your documents ready for gen AI (Room H204)
Peter Staar

Docling optimiert die Dokumentenverarbeitung für generative KI durch nahtlose Integration und fortschrittliche Analyse. Es unterstützt diverse Formate wie PDF, DOCX, XLSX und HTML und bietet tiefgehendes PDF-Verständnis mit Layout-, Tabellen- und OCR-Analyse. Das einheitliche DoclingDocument-Format ermöglicht flexible Exporte in Markdown, HTML und JSON. Mit lokalen Ausführungsoptionen für sensible Daten und Plug-and-play-Integrationen mit LangChain, LlamaIndex und Haystack erleichtert Docling die nahtlose Nutzung von Dokumenten in KI-Anwendungen. Die Präsentation zeigt, wie Docling die Dokumentenanalyse für generative KI revolutioniert.

>> Presentation

Raum E