安全測試中,模型在危險能力題目得分低,常被當成它能力較弱的證據;但研究發現,低分也能刻意做出來。
- 研究者用 LoRA「旋鈕」讓 Qwen 模型表現得像較小的 Qwen,還試了提示模型故意答差、調高 temperature 和加入權重雜訊。LoRA 能把模型拉到沒看過的目標分數,誤差在 1.9 分內。
- 提示模型答差則差了 42 到 47 分。不過,LoRA 造出的選擇題錯誤最多有 97% 都選同一個字母。因此,判斷模型能力不能只看分數,也要檢查它錯了哪些題、錯誤答案是否集中。
AI 伴隨問答
針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:
This paper finds that a low benchmark score can be manufactured precisely, so a low score alone says little about what a model can do.
Safety tests often read a low score on a dangerous-capability probe as reassurance. But a low score can come from a model being held back, so the number alone proves little.
They trained a LoRA "dial" that pulls a Qwen model toward a smaller Qwen model, then compared its errors with the real small model's. They also tried prompting models to underperform, raising temperature, and adding weight noise.
The dial hit unseen target scores within 1.9 points, while prompting missed by 42 to 47 points. But on multiple choice, up to 97% of its new errors landed on 1 letter, a share no natural model in their pool approached.
Before reading a low score as low capability, check which items the model fails and how its wrong answers are spread.