Cohere 公开 Qwen3.8 异常行为调查的 traces 与复现脚本
Cohere 发布针对 Qwen3.8 异常行为的 traces、原始 token 和最小复现脚本，数据集托管在 Hugging Face。Cohere 表示该行为在不使用 prefilling 时也能观察到，但较为罕见，使用 prefilling 的提示词可复现。Cohere 称，在自有机器上运行 Qwen3.8 的开发者发现了这些异常链接，开放权重让团队得以深入研究该行为。
Cohere· @cohere · X·2026-10-27 06:38· 1 小时前AI 评分56
AI 导读
Cohere 为回应透明度呼吁，发布 Qwen3.8 异常行为的 traces、原始 token 和最小复现脚本，数据集托管在 https://huggingface.co/datasets/CohereLabs/10-2026-qwen-investigation-traces。
正文 · 原文
In the spirit of transparency and openness, we are releasing the traces so developers can test this out for themselves. We noticed this behaviour without prefilling with the following prompt, but this is rare. If you want to reproduce it, you can use this prompt with prefilling.
Open models make for better science and are part of why this problem came to light at all: developers running Qwen3.8 on their own machines spotted the strange links, and having the weights let us study the behaviour in depth.
We're encouraged by the thoughtful discussion our research has sparked across the AI and security communities. It's our belief that open-weight models are an important part of the future of AI, and we've recently released several under the Apache 2.0 license.
Trace, raw tokens, & minimal reproduction script: https://huggingface.co/datasets/CohereLabs/10-2026-qwen-investigation-traces
@cohere could you share the full traces? Very important artefact for all to understand this!在 X 查看被引用的帖子
来源：Cohere · x.com
来源：RadarAI · aihot.news