阿里披露过滤 Qwen4-Max 训练数据,弱化网络攻击能力并降低生化核风险
terryyuezhuo · x · 2026-11-16
According to a South China Morning Post report relayed via Techmeme, Alibaba said it filtered Qwen4-Max's training data to remove offensive cyber capabilities and mitigate chemical, biological, radiological, and nuclear (CBRN) risks. The poster adds a correction: it's not a full removal — "weaken the cyber capabilities" is more accurate. A rare public detail from a major model vendor about safety-oriented training data curation.
「模型」频道最新
- Qwen4-max发布页被吐槽:只有标题和评论区,无模型卡无评测 — petrusenko_max · 2026-11-16
- SLDBench v2发布:Claude Opus 5.5仅用4%算力成本超越经典拟合基线 — jiqizhixin · 2026-11-16
- Grok Bot 1.0 发布:机器人可入会、主动行动并自我改进 — mattyp · 2026-11-16
- 用户实测惊呼:Qwen4 Max Preview 表现强到「不该出现在人类世界」 — rickasaurus · 2026-11-16
- 曝新模型或为 Claude 新档位:从 $10/$50 降至 $8/$40 — lyraxana · 2026-11-16
- NVIDIA Parakeet每30秒硬补句号:80%转述被无声截断 — shauntrennery · 2026-11-16