153 likesNew · 1d old
Western-values fine-tune slashes propaganda in Chinese LLM
What is Arete-27B?
Arete-27B is a 27-billion-parameter model rebuilt from Alibaba's Qwen3.8-27B and re-trained to favor western democratic values. The fine-tuning cuts politically biased answers from 22% to under 1% while modestly improving factual accuracy.
Who it is for
Researchers and developers who need an open-source LLM with reduced state-aligned bias can use it as a more neutral base for downstream apps.