Nous Research:

Nous Research unveils Arete-27B, a post-trained Qwen3.8-27B model that reinforces Western values, saying the underlying Chinese model's bias layer is shallow  —  We post-trained over Alibaba's Qwen3.8-27B to reinforce Western values on a Chinese-trained model.  We call this new model Arete-27B.

Nous Research:

Nous Research unveils Arete-27B, a post-trained Qwen3.8-27B model that reinforces Western values, saying the underlying Chinese model's bias layer is shallow  —  We post-trained over Alibaba's Qwen3.8-27B to reinforce Western values on a Chinese-trained model.  We call this new model Arete-27B.

Source: TechMeme
Source Link: https://www.techmeme.com/261108/p3#a261108p3