Nous Research unveils Arete-27B, a post-trained model built to reduce Chinese AI propaganda

Nous Research fine-tuned a Qwen base model to cut its propaganda rate from 22% to 1%, naming the result after a classical Greek concept of virtue. Nous Research has released Arete-27B, a fine-tuned language model built on a Qwen base that the company describes as an effort to reinforce Western values on a Chinese-trained foundation. The headline result: a propaganda rate reduced from 22% to 1%, with no significant loss in general performance. The name is deliberately chosen. Arete, written in Greek as ἀρετή, is the classical concept of excellence and virtue, specifically the full realization of something’s highest potential. For Nous, it doubles as a mission statement. What Nous actually did Qwen is Alibaba’s family of open-weight language models, widely used as a base for fine-tuning across the open-source AI community. The models are capable and freely available, which makes them attractive starting points. They also carry measurable ideological tendencies baked in during pre-training on Chinese data, which is where the 22% propaganda rate figure comes from. Nous applied post-training, the technical term for supervised fine-tuning and reinforcement learning steps that happen afte...