WestCode1357/gpt-sw3-6.7b
WestCode1357/gpt-sw3-6.7b is a 7.1 billion parameter base model, a community mirror of AI Sweden's GPT-SW3. It is designed for text completion in Swedish, Norwegian, Danish, Icelandic, and English. This model is trained on 320 billion tokens, including code, making it suitable for research and educational purposes in Nordic languages and English.
Loading preview...
GPT-SW3 6.7B: Multilingual Base Model for Nordic Languages and English
This model, gpt-sw3-6.7b, is a 7.1 billion parameter base model developed by AI Sweden in collaboration with RISE and WASP WARA for Media and Language. It is a community mirror of the original AI Sweden model.
Key Capabilities
- Multilingual Text Completion: Excels at generating text in Swedish, Norwegian, Danish, Icelandic, and English.
- Extensive Training: Trained on a substantial dataset of 320 billion tokens, encompassing web data and code.
- Research Focus: Primarily intended for scientific and research use, allowing exploration of large language models in a multilingual context.
Intended Use and Limitations
GPT-SW3 6.7B is provided as-is for research and educational purposes only. It is explicitly not intended for commercial use due to potential biases inherited from its training data and a lack of alignment or safety-tuning. Users are responsible for content generated and must review the AI Sweden RAIL license before any deployment. It is recommended for use strictly in controlled research settings to mitigate risks of generating inaccurate, offensive, or inappropriate content.