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.
The Yi series models are the next generation of open-source large language models trained from scratch by 01.AI.
Targeted as a bilingual language model and trained on 3T multilingual corpus,
https://huggingface.co/01-ai/Yi-34B-Chat