NorskGPT-Mistral-7b: Norwegian Language Model
NorskGPT-Mistral-7b is a 7 billion parameter language model developed by bineric, based on the Mistral-7b-v0.1 architecture. It has been specifically fine-tuned using a curated dataset of Norwegian instruction pairs to enhance its performance in the Norwegian language.
Key Capabilities & Performance
- Norwegian Language Proficiency: Designed to understand and generate text effectively in Norwegian.
- Leaderboard Ranking: As of February 2nd, 2024, it holds a strong position on the Mainland Scandinavian NLG leaderboard, tied for 2nd place overall and recognized as the top Norwegian-specific model after GPT-3.5.
- Instruction Following: Tuned for assistant-like chat interactions and instruction-based tasks.
Intended Use & Limitations
This model is primarily intended for personal and research use within the Norwegian language context. It performs well on tasks such as summarization, question answering, and general chat. However, as an LLM, it does not possess specialized knowledge beyond its base model and fine-tuning data, and may require prompt tuning for optimal results. Users should be aware that the base model lacks censorship, and the fine-tune does not directly address this, meaning it can potentially generate harmful content if explicitly prompted. The model is released under the Attribution-NonCommercial-ShareAlike 4.0 International license, restricting commercial use.