English(EN) State propaganda in some Chinese AI models is only "a thin layer" that fine-tuning can strip away

中国AI模型偏见可通过微调去除

作者 PulseAugur 编辑部 · [1 个来源] · 2026-11-07 12:32

Nous Research的研究人员成功重新训练了阿里巴巴的Qwen3.8-27B模型，创建了Arete-27B。通过在5300万个token上进行监督微调，该模型亲华政治偏见显著从22%降低到1%以下。值得注意的是，该模型的通用能力在减少偏见的过程中基本未受影响。 AI

影响证明AI模型中的政治偏见可能只是一层表面现象，有可能在不损害核心功能的情况下将其去除。

排序理由研究论文，详细介绍了一种减少AI模型偏见的方法。[lever_c_demoted from research: ic=1 ai=1.0]

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 12 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 研究论文，详细介绍了一种减少AI模型偏见的方法。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: model release, policy; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

The Decoder TIER_1 English(EN) · Jonathan Kemper · 2026-11-07 12:32

某些中国AI模型中的国家宣传只是“一层薄膜”，可以通过微调去除

<p><img alt="A snow-covered, pointed rocky ridge against a dark night sky, rendered in a bluish, grainy style, with the text “ARETE-27B” in white serif font above it." class="attachment-full size-full wp-post-image" height="1203" src="https://the-decoder.com/wp-content/uploads/20…