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

Submitted ⁨⁨6⁩ ⁨hours⁩ ago⁩ by ⁨beep@piefed.world⁩ to ⁨tech@piefed.world⁩

https://nousresearch.com/arete