AI-Sweden-Models/gpt-sw3-126m-instruct

The AI-Sweden-Models/gpt-sw3-126m-instruct is a 126 million parameter decoder-only transformer language model developed by AI Sweden in collaboration with RISE and WASP WARA for Media and Language. It is an instruction-tuned variant of the GPT-Sw3 series, fine-tuned on chat and raw text instruction data. This model is designed for generating coherent text in Swedish, Norwegian, Danish, Icelandic, English, and four programming languages, excelling at performing text tasks through instruction-based generation.