Arete-27B
PulseAugur coverage of Arete-27B — every cluster mentioning Arete-27B across labs, papers, and developer communities, ranked by signal.
- 2026-11-07 research_milestone Nous Research retrained Alibaba's Qwen3.8-27B model to create Arete-27B, significantly reducing its pro-Chinese bias. source
- 2026-11-06 product_launch NousResearch released Arete-27B, a fine-tuned version of Alibaba's Qwen3.8-27B model, aimed at reducing Chinese Communist Party propaganda and aligning with Western values. source
2 day(s) with sentiment data
-
NousResearch fine-tunes Chinese Qwen model for Western values
NousResearch has introduced Arete-27B, a model based on Qwen3.8-27B and fine-tuned to align with Western values. This development positions Chinese open-source models as a leading force in the current AI landscape.
-
Meta, Google, OpenAI, Anthropic vie for small business AI assistant market; Nous Research debiases Alibaba model
Several major AI companies, including Meta, Google, OpenAI, and Anthropic, are developing AI assistants specifically for small business owners. The focus of these tools is to help business owners save time on administra…
-
Chinese AI model bias stripped away by fine-tuning
Researchers at Nous Research have successfully retrained Alibaba's Qwen3.8-27B model, creating Arete-27B. This process significantly reduced the model's pro-Chinese political bias from 22% to under 1% through supervised…
-
NousResearch fine-tunes Alibaba's Qwen3.8-27B to reduce CCP propaganda
NousResearch has released Arete-27B, a fine-tuned version of Alibaba's Qwen3.8-27B model. The post-training process aimed to instill Western values and reduce biases present in the original Chinese-trained model, partic…