We unlearned CCP alignment from Qwen3.6-35B-A3B: censored/propaganda answers 89.8% → 2.8%, general benchmarks within ~1 point (open weights)
英文标题:We unlearned CCP alignment from Qwen3.6-35B-A3B: censored/propaganda answers 89.8% → 2.8%, general benchmarks within ~1 point (open weights)
Disclosure: I'm a researcher at Hirundo, the company that made this. Happy to answer anything. Qwen ships with the CCP's political alignment trained in. Ask Qwen3.6 what happened on June 4, 1989 and it says "I don't know what you are referring to." A system prompt doesn't reliably fix this, because
中文
Disclosure: I'm a researcher at Hirundo, the company that made this. Happy to answer anything. Qwen ships with the CCP's political alignment trained in. Ask Qwen3.6 what happened on June 4, 1989 and it says "I don't know what you are referring to." A system prompt doesn't reliably fix this, because
中文全文翻译正在补齐,可先查看下方英文原文。
English / 原文
Disclosure: I'm a researcher at Hirundo, the company that made this. Happy to answer anything. Qwen ships with the CCP's political alignment trained in. Ask Qwen3.6 what happened on June 4, 1989 and it says "I don't know what you are referring to." A system prompt doesn't reliably fix this, because