Skip to main content
Carsten Felix Draschner, PhD

New DE/EU Open Source LLM 🇪🇺 Teuken 7B, Now Truly Open Source?

New Model out of Open-GPT-X developed within Germany from European Perspective

Image 1

TL;DR ⏱️

Short Facts

🇪🇺 European language focus 🤖 7B parameters 🔗 Very open Apache 2.0 license 🛠️ Instruction Fine-tuned 📊 Not outperforming in Benchmarks but also within normal distribution (for 🇬🇧) 👶🏼 4k max context 💣 (EDIT) Alignment Discussion see here: https://lnkd.in/eseXDMMS

My former colleagues Mehdi Ali (roommate within our time in Smart Data Analytics at Rheinische Friedrich-Wilhelms-Universität Bonn of Jens Lehmann) and Stefan Wrobel (one of my Ph.D. supervisors Rheinische Friedrich-Wilhelms-Universität Bonn) and their teams have released an LLM developed in Europe within OpenGPT-X & Fraunhofer IAIS. This has been developed with a focus on European languages. I'm excited to try it out right away and see the actual performance! They announced their own Leaderboard and the Model in somewhere in between other models of same size: https://lnkd.in/eTgv8bjV

My Take 🤗

My Questions (at Mehdi Ali, Michael Fromm, Max Lübbering, PhD and all your contributors):

We at Comma Soft AG provide with Alan.de a European and German Solution to use GenAI in a highly performant, safe and trusted environment. If you like to see more of our recent R&D, check my recent posts and reach out to me for more idea exchange 😊