NbAiLab/nb-llama-3.2-1B
NbAiLab/nb-llama-3.2-1B is a 1 billion parameter multilingual generative model developed by the National Library of Norway (NB-AiLab). Built on the Llama 3.2 architecture, it is fine-tuned to provide full support for Norwegian Bokmål, Norwegian Nynorsk, and English, with partial support for Swedish and Danish. This model excels at Norwegian-specific tasks like summarization and question-answering, while maintaining strong multilingual generation capabilities.
Loading preview...
NB-Llama-3.2-1B: Multilingual Model for Nordic Languages
NB-Llama-3.2-1B is a 1 billion parameter generative language model developed by the National Library of Norway (NB-AiLab). It is part of the NB-Llama-3.2 series, fine-tuned from the Llama-3.2-1B base model to enhance support for Norwegian languages while preserving English capabilities.
Key Capabilities
- Multilingual Support: Full support for Norwegian Bokmål, Norwegian Nynorsk, and English, with partial support for Swedish and Danish.
- Norwegian-Specific Tasks: Optimized for tasks such as question-answering and summarization in Norwegian contexts.
- Training Data: Trained on publicly available multilingual datasets, including CulturaX, HPLT, NCC, and Wikipedia, with a focus on high-quality Norwegian, English, Swedish, and Danish sources. Data selection utilized custom Corpus Quality Classifiers to prioritize educational value and linguistic quality.
- Architecture: Utilizes the Llama 3.2 architecture, an auto-regressive language model, with fine-tuning via Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO).
Intended Use Cases
- Dialogue systems.
- General multilingual text generation and language modeling.
- Norwegian-specific tasks like text summarization and question-answering tailored to cultural and linguistic contexts.
This model is released under the Llama 3.2 Community License and is considered a research project, with an emphasis on experimental use for adapting models for the Norwegian language.