c4darkness

Introducing Arete-27B: Qwen3.8-27B, post-trained to reinforce Western values. Chinese open models are currently the strongest available, but they ship with state-aligned behavior in the weights. We built an 803-prompt eval across seven value categories, including freedom of expression, democracy, and privacy, and measured how often a model adopts China’s party line. After fine-tuning on roughly 53 million tokens, Arete-27B's propaganda rate fell from 22% to 1%, in line with the Western models we tested. Its Hermes Bench score shows no significant performance tradeoffs. The values in an open model are one fine-tune away. Our eval, weights, and training data are available now along with the full blog post: nousresearch.com/arete

来了个大的 Hugging Face 上最近杀出一个核弹级无审查模型，直接被挤爆了 DeepSeek-V4-Pro + Qwen3.5 9B 的混血完全体，体积只有 5.6GB，上线就被狂下 1.8 万次。 它的测试结果极其残暴： · 拒答率暴跌：实测 100 个刁钻测试提示词，原版直接 100% 拒答，它硬生生只拒了 3 个 · 顶配推理蒸馏：用 DeepSeek-V4-Pro 约 25 万条高质量数据蒸馏，逻辑与数学推理能力完全拉满 · 零损耗去审查：底层用 Heretic 的 ARA 算法消融限制，彻底卸掉道德锁，智商分毫不损 · 普通电脑本地直跑：Q4_K_M 的 GGUF 格式仅 5.6GB，显存要求极低，核显本和轻薄本也能秒开 完全开源免费，Apache 2.0 协议，想在本地跑一个完全听话、推理极强的大模型的兄弟建议直接马住！ 模型体验：huggingface.co/xtools-at/Dee… 你平时在本地跑大模型，遇到拒答最严重的是写代码还是角色扮演？ #AI #大模型 #开源