Tracking instruction-tuned LLM openness. Paper: Liesenfeld, Andreas, Alianda Lopez, and Mark Dingemanse. 2023. “Opening up ChatGPT: Tracking Openness, Transparency, and Accountability in Instruction-Tuned Text Generators.” In Proceedings of the 5th International Conference on Conversational User Interfaces. doi:10.1145/3571884.3604316.
A mistral-based Orca-tuned open model: https://huggingface.co/Intel/neural-chat-7b-v3-1
GWDG uses it for researchers: https://info.gwdg.de/news/en/gwdg-llm-service-generative-ai-for-science/