Munich🥨NLP’s cover photo
Munich🥨NLP

Munich🥨NLP

IT Services and IT Consulting

Join the Munich NLP discord community to learn and exchange about Natural Language Processing and Machine Learning

About us

MunichNLP is a volunteer-run community for NLP & ML researchers and professionals in Munich and beyond. We bring together experts, researchers, and hobbyists to share knowledge, collaborate, and innovate in the fields of Natural Language Processing and AI. We organize online talks, hands-on demos, and interactive discussions, as well as in-person meet-ups in Munich creating opportunities to learn from industry leaders, researchers, and fellow enthusiasts. Whether you're a seasoned professional or just starting out, our free to join events provide a space to explore the latest trends, exchange ideas, and build meaningful connections. Our mission is to bridge the gap between experts in NLP, fostering a vibrant, inclusive community where everyone can contribute and grow. By hosting regular online and offline events, we aim to make NLP and AI accessible, engaging, and collaborative. We do online talks, discussions, demos as well as live meet-ups in Munich 🥨 - want to learn more Join 🫵👉👇: https://discord.gg/bCuya2mKts

Website
https://munich-nlp.com/
Industry
IT Services and IT Consulting
Company size
2-10 employees
Headquarters
Munich
Type
Nonprofit
Specialties
Natural Language Processing, NLP, Speech Recognition, Munich, Computational Linguistics, Non Profit, Data Science, artificial intelligence, Meetups, Text Mining, and Gen Ai

Locations

Employees at Munich🥨NLP

Updates

  • View organization page for Munich🥨NLP

    2,195 followers

    Munich🥨NLP Meetup Recap: Multilingual LLMs and AI Hardware at LRZ We spent yesterday at the Leibniz Supercomputing Centre (LRZ) diving into how specialized hardware is breaking the speed and language barriers of modern AI. Here are the highlights: 🎤 Michael Hoffmann & Jophin John (LRZ) — Llama-GENBA-10B: English, German, and... Bavarian? *The "Beergarten" Origin: What started as a casual chat in a beer garden turned into a serious research project: the first trilingual model for English, German, and Bavarian. *The Data Challenge: Low-resource languages are notoriously difficult. The team harvested Bavarian sentences from German newspapers and built a custom tokenizer to handle dialect nuances. *The Training Recipe: By using continuous pre-training followed by fine-tuning, they achieved state-of-the-art results for Bavarian, overcoming the total lack of standard benchmarks. *Efficiency at Scale: Trained on the Cerebras CS-2, the model saw 66 days of training with 35 MWh of energy consumption—a fascinating look at efficient foundation models. 🎤 Dmitry Gaynullin (Cerebras Systems) — Building Impossible Things Hardware Architecture: Unlike GPU clusters, the Cerebras system creates a static graph directly into its 850,000 cores. With "weight streaming," it handles massive models with the simplicity of a single node. Speed = Reasoning: Why aren't reasoning models everywhere? The main answer is speed. Dmitry showed how the Wafer-Scale Engine makes the "Thought-Act-Observe" loop significantly faster, making deep research agents practical for real-time use. 💻 Hands-on Workshop: Deep Research at Warp Speed We wrapped up with a hands-on session building our own deep research agents. The inference speed on the Cerebras hardware was truly eye-opening; seeing a reasoning agent move that fast changes how you think about AI workflows! A massive thank you to our hosts and speakers for making this happen! 🙏 From the Leibniz-Rechenzentrum team: Nicolay Hammer Michael Hoffmann Jophin John Leo Mak 🙏 From the Cerebras team: Alexander Mikoyan Dmitry Gaynullin And of course, thank you to the Munich🥨NLP community for the great questions and for braving the trip out to Garching! 🍕 #MunichNLP #LRZ #Cerebras #BavarianAI #LLM #GenerativeAI #WaferScale #AIInfrastructure #DeepResearch

    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
      +1
  • Munich🥨NLP reposted this

    🚘 The Munich MLOps community Meetup is back! 🚘 This time with AI Agents in style 🕺🏻💃🏻 Tomorrow, 21.01.2026, we will be hosted at Slalom with two amazing talks from: 1️⃣ Pradeep will talk about #AI Agents at Scale with #AWS Lambda and Bedrock Agents 2️⃣ Meryem from Doubleword will share about using Async #AI Agents in production. Seats are limited. The insights aren't. Register here: https://luma.com/ylrvmefz See you soon! Wilson CELY Jitendra Kasaudhan Marina Bevacqua

    • No alternative text description for this image
  • Munich🥨NLP reposted this

    𝐀𝐟𝐭𝐞𝐫 𝐨𝐮𝐫 𝐬𝐮𝐜𝐜𝐞𝐬𝐬𝐟𝐮𝐥 𝐲𝐞𝐚𝐫 𝐤𝐢𝐜𝐤-𝐨𝐟𝐟 𝐚𝐭 Eraneos, 𝐰𝐞'𝐫𝐞 𝐭𝐡𝐫𝐢𝐥𝐥𝐞𝐝 𝐭𝐨 𝐢𝐧𝐯𝐢𝐭𝐞 𝐲𝐨𝐮 𝐭𝐨 𝐨𝐮𝐫 𝐧𝐞𝐱𝐭 𝐢𝐧-𝐩𝐞𝐫𝐬𝐨𝐧 𝐦𝐞𝐞𝐭𝐢𝐧𝐠 𝐟𝐞𝐚𝐭𝐮𝐫𝐢𝐧𝐠 𝐚𝐧 𝐞𝐱𝐩𝐞𝐫𝐭 𝐭𝐚𝐥𝐤 𝐚𝐧𝐝 𝐚 𝐡𝐚𝐧𝐝𝐬-𝐨𝐧 𝐰𝐨𝐫𝐤𝐬𝐡𝐨𝐩 𝐨𝐧 𝐭𝐡𝐞 𝐥𝐚𝐭𝐞𝐬𝐭 𝐢𝐧 𝐍𝐋𝐏 𝐚𝐧𝐝 𝐀𝐈. 📅 Date: January 20, 2026 ⏰ Time: Starting at 18:00 📍 Location: LRZ, Boltzmannstr. 1, 85748 Garching bei München Agenda: 18:00-18:20: Arrival & Networking 18:20-18:30: Welcome & Intro by LRZ & MunichNLP 𝟭𝟴:𝟯𝟬-𝟭𝟵:𝟬𝟬: "𝗟𝗹𝗮𝗺𝗮-𝗚𝗘𝗡𝗕𝗔-𝟭𝟬𝗕: 𝗔 𝗧𝗿𝗶𝗹𝗶𝗻𝗴𝘂𝗮𝗹 𝗟𝗮𝗿𝗴𝗲 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹 𝗳𝗼𝗿 𝗚𝗲𝗿𝗺𝗮𝗻, 𝗘𝗻𝗴𝗹𝗶𝘀𝗵 𝗮𝗻𝗱 𝗕𝗮𝘃𝗮𝗿𝗶𝗮𝗻"  Speakers: Michael H. and Jophin John (Research Scientist BDAI at LRZ) Description: Michael and Jophin will introduce LLama-GENBA-10B, the first large language model designed for English, German, and Bavarian. The presentation focuses on the challenges of building multilingual systems with low-resource languages and the solutions the team developed to overcome them. 𝟭𝟵:𝟯𝟬-𝟮𝟬:𝟭𝟬: "𝗕𝘂𝗶𝗹𝗱 𝗜𝗺𝗽𝗼𝘀𝘀𝗶𝗯𝗹𝗲 𝗧𝗵𝗶𝗻𝗴𝘀 𝘄𝗶𝘁𝗵 𝗖𝗲𝗿𝗲𝗯𝗿𝗮𝘀 – 𝗪𝗵𝗮𝘁'𝘀 𝗡𝗲𝘄?" Speaker: Gokul Ramakrishnan ( Applied Scientist Lead & Senior Manager @ Cerebras) Description: Gokul will provide a hands-on exploration of building multi-agent workflows using Cerebras inference. The talk will demonstrate practical approaches to creating agent systems and share insights into the latest developments at Cerebras. 🍕 Food and drinks will be provided! 🙏 𝗧𝗵𝗮𝗻𝗸𝘀 𝘁𝗼 𝗼𝘂𝗿 𝘀𝗽𝗼𝗻𝘀𝗼𝗿𝘀 Leibniz-Rechenzentrum 𝗮𝗻𝗱 Cerebras 𝗳𝗼𝗿 𝗵𝗼𝘀𝘁𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝘂𝘀 𝘁𝗵𝗲 𝗲𝘃𝗲𝗻𝘁. 𝗪𝗲 𝗹𝗼𝗼𝗸 𝗳𝗼𝗿𝘄𝗮𝗿𝗱 𝘁𝗼 𝘀𝗲𝗲𝗶𝗻𝗴 𝘆𝗼𝘂 𝘁𝗵𝗲𝗿𝗲! RSVP NOW: https://lnkd.in/dmN7Q-5m #MunichNLP #NLP #AI #Networking #Innovation #AgenticAI #LLM #AIinProduction #MunichTech #ResearchToProduction #Munich

  • 𝐀𝐟𝐭𝐞𝐫 𝐨𝐮𝐫 𝐬𝐮𝐜𝐜𝐞𝐬𝐬𝐟𝐮𝐥 𝐲𝐞𝐚𝐫 𝐤𝐢𝐜𝐤-𝐨𝐟𝐟 𝐚𝐭 Eraneos, 𝐰𝐞'𝐫𝐞 𝐭𝐡𝐫𝐢𝐥𝐥𝐞𝐝 𝐭𝐨 𝐢𝐧𝐯𝐢𝐭𝐞 𝐲𝐨𝐮 𝐭𝐨 𝐨𝐮𝐫 𝐧𝐞𝐱𝐭 𝐢𝐧-𝐩𝐞𝐫𝐬𝐨𝐧 𝐦𝐞𝐞𝐭𝐢𝐧𝐠 𝐟𝐞𝐚𝐭𝐮𝐫𝐢𝐧𝐠 𝐚𝐧 𝐞𝐱𝐩𝐞𝐫𝐭 𝐭𝐚𝐥𝐤 𝐚𝐧𝐝 𝐚 𝐡𝐚𝐧𝐝𝐬-𝐨𝐧 𝐰𝐨𝐫𝐤𝐬𝐡𝐨𝐩 𝐨𝐧 𝐭𝐡𝐞 𝐥𝐚𝐭𝐞𝐬𝐭 𝐢𝐧 𝐍𝐋𝐏 𝐚𝐧𝐝 𝐀𝐈. 📅 Date: January 20, 2026 ⏰ Time: Starting at 18:00 📍 Location: LRZ, Boltzmannstr. 1, 85748 Garching bei München Agenda: 18:00-18:20: Arrival & Networking 18:20-18:30: Welcome & Intro by LRZ & MunichNLP 𝟭𝟴:𝟯𝟬-𝟭𝟵:𝟬𝟬: "𝗟𝗹𝗮𝗺𝗮-𝗚𝗘𝗡𝗕𝗔-𝟭𝟬𝗕: 𝗔 𝗧𝗿𝗶𝗹𝗶𝗻𝗴𝘂𝗮𝗹 𝗟𝗮𝗿𝗴𝗲 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹 𝗳𝗼𝗿 𝗚𝗲𝗿𝗺𝗮𝗻, 𝗘𝗻𝗴𝗹𝗶𝘀𝗵 𝗮𝗻𝗱 𝗕𝗮𝘃𝗮𝗿𝗶𝗮𝗻"  Speakers: Michael H. and Jophin John (Research Scientist BDAI at LRZ) Description: Michael and Jophin will introduce LLama-GENBA-10B, the first large language model designed for English, German, and Bavarian. The presentation focuses on the challenges of building multilingual systems with low-resource languages and the solutions the team developed to overcome them. 𝟭𝟵:𝟯𝟬-𝟮𝟬:𝟭𝟬: "𝗕𝘂𝗶𝗹𝗱 𝗜𝗺𝗽𝗼𝘀𝘀𝗶𝗯𝗹𝗲 𝗧𝗵𝗶𝗻𝗴𝘀 𝘄𝗶𝘁𝗵 𝗖𝗲𝗿𝗲𝗯𝗿𝗮𝘀 – 𝗪𝗵𝗮𝘁'𝘀 𝗡𝗲𝘄?" Speaker: Gokul Ramakrishnan ( Applied Scientist Lead & Senior Manager @ Cerebras) Description: Gokul will provide a hands-on exploration of building multi-agent workflows using Cerebras inference. The talk will demonstrate practical approaches to creating agent systems and share insights into the latest developments at Cerebras. 🍕 Food and drinks will be provided! 🙏 𝗧𝗵𝗮𝗻𝗸𝘀 𝘁𝗼 𝗼𝘂𝗿 𝘀𝗽𝗼𝗻𝘀𝗼𝗿𝘀 Leibniz-Rechenzentrum 𝗮𝗻𝗱 Cerebras 𝗳𝗼𝗿 𝗵𝗼𝘀𝘁𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝘂𝘀 𝘁𝗵𝗲 𝗲𝘃𝗲𝗻𝘁. 𝗪𝗲 𝗹𝗼𝗼𝗸 𝗳𝗼𝗿𝘄𝗮𝗿𝗱 𝘁𝗼 𝘀𝗲𝗲𝗶𝗻𝗴 𝘆𝗼𝘂 𝘁𝗵𝗲𝗿𝗲! RSVP NOW: https://lnkd.in/dmN7Q-5m #MunichNLP #NLP #AI #Networking #Innovation #AgenticAI #LLM #AIinProduction #MunichTech #ResearchToProduction #Munich

  • View organization page for Munich🥨NLP

    2,195 followers

    We kicked off the new year with a packed room at Eraneos, staying true to our philosophy: mixing real-world industry lessons with deep academic research. Here are the highlights: 🎤 David Satomi, Fabian Stermann: From Scratch to Production: LLM Based E2E Service in Swiss Banking - The Swiss Challenge: Handling customer service across 4 official languages requires robust architecture. - The Router Pattern: They showcased moving beyond simple tool calling by using a router to decide if a request needs a direct sub-agent or a RAG lookup. - Iterate, don't just prompt: Complex banking requests need a multi-action system that iterates step-by-step for reliability. 🎤 Ludovic Tessier 🇺🇦 & Claudio Garutti, PhD: Beyond Retrieval: The Technical Core of Reliable Agent Reasoning - RAG Reality Check: Discussed the trade-offs of context overload and what to do when documents conflict  - "Astute RAG": Moving from classic retrieval to a system-in-the-loop: Internal Knowledge ➝ Retrieval ➝ "Thought-Act-Observe." - Agentic Trade-offs: While agentic systems (like a research assistant) are non-deterministic and costlier than classic RAG, they solve the "consistency" weakness of online data. 🎤 Dr. Matthias Assenmacher: Advances in Open-Ended Text Generation and Evaluation - The Hidden Decoder: A deep dive into the decoder part of the Transformer architecture. While often overlooked, the decoding strategy has a massive impact on final results. - The Evaluation Dilemma: Evaluating open-ended text is difficult because there are so many "axes" to measure. While we need robust, unified scores, different use cases often require different "flavors" of output. - Optimizing Generation: Matthias shared their work on a decoding method that balances competing needs while removing the friction of tuning extra hyperparameters - showing impressive results across several evaluation criteria. A huge thank you to Eraneos for hosting us and providing the space to connect (and catering!)- Claudia Schulze, Jakob Klement, Luca Mülln, Fabienne G. And thanks to our speakers and community for the great discussions. 🗓️ Next up: Jan 20 @ LRZ We are keeping the momentum going with a workshop on hardware and multilingual models! Topic: Multilingual LLMs & Hands-on Cerebras Workshop Location: LRZ (Leibniz Supercomputing Centre), Boltzmannstr. 1 https://lnkd.in/dV3hcb6X #MunichNLP #AgenticAI #RAG #LLM #AIinProduction #MunichTech #ResearchToProduction

    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
      +2
  • Munich🥨NLP reposted this

    **New post: PyTorch and CPU–GPU Synchronizations (Writing Fast PyTorch Code) 🚀** PyTorch GPU execution is mostly asynchronous. Performance issues often come from implicit CPU-GPU synchronizations: In this case, the CPU becomes blocked and stops scheduling new PyTorch operations for the GPU. In this post I cover: * What syncs look like in Nsight Systems: long CPU-side CUDA API calls like `cudaStreamSynchronize`, or blocking device-to-host transfers. * Subtle sync triggers in real code: Dynamic-shape patterns where output sizes depend on GPU data (e.g. boolean indexing / data-dependent slicing). * How to unit test for syncs: `torch.cuda.set_sync_debug_mode()` to surface sync points as warnings/errors in tests. Post: https://lnkd.in/epsGFUUc #PyTorch #CUDA #performance #nsight #GPUs

  • View organization page for Munich🥨NLP

    2,195 followers

    Start your new year with a deep dive into the latest advancements in NLP and LLMs! 🚀 🎇 New Year Kick Off - Munich NLP Meetup! 🎇 📅 Date: January 8, 2026 ⏰ Time: 6:30 PM - 9:30 PM MEZ 📍 Location: Eraneos Office, Klenzestraße 41, München 🌟 Agenda: 🌟 𝐅𝐫𝐨𝐦 𝐒𝐜𝐫𝐚𝐭𝐜𝐡 𝐭𝐨 𝐏𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧: 𝐋𝐋𝐌 𝐁𝐚𝐬𝐞𝐝 𝐄2𝐄 𝐒𝐞𝐫𝐯𝐢𝐜𝐞 𝐢𝐧 𝐒𝐰𝐢𝐬𝐬 𝐁𝐚𝐧𝐤𝐢𝐧𝐠 Speakers: Jakob Klement, David Satomi, Fabian Stermann (Eraneos) Discover how an end-to-end LLM-based voice and chatbot system was developed and implemented at a Swiss bank, overcoming technical and regulatory challenges. 𝐁𝐞𝐲𝐨𝐧𝐝 𝐑𝐞𝐭𝐫𝐢𝐞𝐯𝐚𝐥: 𝐓𝐡𝐞 𝐓𝐞𝐜𝐡𝐧𝐢𝐜𝐚𝐥 𝐂𝐨𝐫𝐞 𝐨𝐟 𝐑𝐞𝐥𝐢𝐚𝐛𝐥𝐞 𝐀𝐠𝐞𝐧𝐭 𝐑𝐞𝐚𝐬𝐨𝐧𝐢𝐧𝐠 Speaker: Ludovic Tessier 🇺🇦, Claudio Garutti, PhD (Google) Explore the next step in AI development: reliable agent reasoning. Learn about the technical core components that enable AI agents to analyze, synthesize, evaluate, and act upon retrieved data. 𝐀𝐝𝐯𝐚𝐧𝐜𝐞𝐬 𝐢𝐧 𝐎𝐩𝐞𝐧-𝐄𝐧𝐝𝐞𝐝 𝐓𝐞𝐱𝐭 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐨𝐧 𝐚𝐧𝐝 𝐄𝐯𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧 𝐟𝐨𝐫 𝐋𝐋𝐌𝐒 Speaker: Dr. Matthias Assenmacher (Ludwig-Maximilians-Universität München) Delve into the evolving landscape of open-ended text generation and decoding strategies for LLMs. 🍕 Catering and drinks will be provided 📢 RSVP: https://lnkd.in/d9bxGivq Sponsored by Eraneos. Organized by Munich🥨NLP-Community! #NLP #MachineLearning #AI #Meetup #MunichNLP #Eraneos #Innovation #Technology #Networking #Munich

  • 🎉 Welcome aboard, Galit! Thrilled to have you join the Munich🥨NLP team as our newest volunteer organizer. Your energy and passion are already making "waves" can’t wait to see the amazing events you’ll help us bring to life in 2026! For those who don’t know us yet: We’re a volunteer-run community connecting AI researchers, engineers, and enthusiasts in Munich. Expect more insightful talks, networking, and collaboration next year—stay tuned! 👉 Follow us here and on Meetup to catch all the updates https://lnkd.in/eVX86Yws

    I am happy to announce that I have joined the organization team for Munich🥨NLP. I have already attended two events that featured insightful talks and excellent networking opportunities. A special thanks to Thomas Berger for the warm welcome and to Ekaterina Artemova for the original connection. I am excited about what lies ahead, with already two events scheduled for January. Stay tuned for upcoming announcements. *follow us here on linkedIn and in our meetup page! link in the comments :) Photo taken in KI performance GmbH office where the last NLP meetup was hosted

    • No alternative text description for this image
  • View organization page for Munich🥨NLP

    2,195 followers

    Full House at Munich🥨NLP We wrapped up the year with a packed meetup hosted by KI performance GmbH, featuring two talks: 🎤 Dr. Daryna Dementieva (Postdoc, Technical University of Munich) - Responsible & Multilingual NLP for Real-World Impact - Multilingual reality check: you often can’t “just collect thousands of texts” for every language. - Practical recipe: cross-lingual transfer + small in-language data (even ~5k samples can significantly improve results). - Responsible NLP in practice: dataset coverage, content warnings, and evaluation that reflects real users. 🎤 Christoph Bensch (Senior AI Engineer KI performance GmbH) — From LLMs to Multi-Agents (in production) - Why conversational AI is evolving into tool-using / autonomous agent systems. - The hard part is not only building agents, it is evaluating them in production. Next events by Munich🥨NLP (links in comment): - Jan 8 - Munich NLP New Year Meetup @ eraneos (Elsenheimerstr. 41) - Jan 20 - Multilingual LLMs & Hands-on Cerebras Workshop @ LRZ (Boltzmannstr. 1) Big thank you to Irina Kuznetsova (planned and set it up from the KI Performance side, but sadly couldn’t join) and Roland Graser / The Residency (pizza sponsor 🍕). And thanks to the wonderful people that made this event a success. Maher Deebleon j. iwanowitschMykhailo Kozyrev, and Alinur Kozhanov #MunichNLP #AIinProduction #MultiAgentSystems #MultilingualNLP #ResponsibleAI

    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
  • It will be a great event 🔥🔥🔥 Thank you KI performance for hosting us and supporting Munich Tech Communities 🥨

    View organization page for KI performance GmbH

    1,901 followers

    Full House! 🚀 (But there is still a chance...) Wow, Munich! The response has been incredible. While all regular spots for the AI Meetup are taken, the discussion is just heating up.   We are proud to see our Senior AI Engineer, Christoph Bensch, take the stage this Thursday. He is moving the conversation beyond the hype: 🤖 From LLMs to Multi-Agents: Why conversational AI is evolving into autonomous systems. 📉 The Hard Part: Building agents is one thing. Evaluating their impact in production is another. Swipe to see the agenda. 👉 Wanna try your luck? Spots open up occasionally. Join the Waiting List here: 🔗 https://lnkd.in/dVe9cZnx We are excited to contribute to the Munich tech community alongside our friends at Munich🥨NLP. Let’s discuss below: As Christoph asks on the last slide: What is your biggest challenge in evaluating AI agents right now? Reliability? Latency? Or business value? 👇 #MultiAgentSystems #AIinProduction #MunichAI #KIperformance #TeamSpotlight #NLP

Similar pages