Aleph Alpha has released Kolibri, an open-weight Large Language Model designed for German and English. This Mixture of Experts model boasts 78.1 billion parameters, with only a fraction active per token. Kolibri was trained on approximately 24 trillion tokens, with a significant portion in German, utilizing 768 NVIDIA B200 GPUs across Germany and Finland. The model is available under the Apache 2.0 license and its weights can be found on Hugging Face. AI
IMPACT Provides a new open-weight LLM option for German and English language tasks, potentially accelerating research and development in these areas.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →