Translation in progress Some entries still include text in the original language.
Oct 13 · Tuesday
Another 201 papers do not have category tags yet and are not included in filtered results.
“
About this collection Collection in progress We are building a large AI safety paper collection to help researchers find work related to their interests. Papers are collected automatically andfiltered only for relevance to AI safety , without judging quality, novelty or reading value. The default “Annotated first” shows papers with tags or published paper analyses first. Each group is ordered by original publication date, newest first, with undated papers last. “Newest first” uses publication dates only. This order is not a quality ranking.
One-line summaries, tags and paper analyses are prepared by AI from the papers. Language models may have systematic biases and produce hallucinations.We plan to carry out human annotation and review to reduce errors in this content.
提出路由 TopK SAE，把语音编码器稀疏码拆成语言与副语言通路
Published Oct 8
Evaluated models SPEAR XLarge , WavLM Large Abstract 路由 SAE 在冻结 SPEAR 与 WavLM 上分开语言和副语言因子，探针与交换支持通路分工。TopK SAE 的稀疏码被分成语言通路与副语言通路，用来分开冻结语音编码器里同时存在的音素信息和说话人、情感、韵律。只重构不能得到作者选定的这一分解。语言通路由音素识别监督，副语言通路由说话人识别、情感分类和韵律预测监督，作用在对方通路上的梯度反转对抗用来压低交叉因子。
用均值差激活转向检验全双工语音模型的情感方向几何
Published Oct 6
Evaluated models kyutai/moshika-pytorch-q8 Abstract 转向跟方向几何走而不跟标签走：sad 可单独对准，高唤醒情感共享方向。Moshi 上的情感激活转向跟方向之间的几何关系走，不跟被请求的标签走。
提出多 SAE 对齐方法，从音乐模型表征中发现和弦与调性轨道
Published Oct 1
Evaluated models MuQ , MusicFM Abstract 用移调对齐的多 SAE 恢复和弦、调与旋律轨道，并用少量锚点做下游音乐理解。作者提出一套面向音乐基础模型的结构可解释性框架：不把 SAE 特征当作孤立概念，而用半音移调把它们连成有序轨道。
用激活交换分析音频语言模型的事件时间绑定
Published Sep 28
Evaluated models AF-Next , MOSS-Audio , Qwen3-Omni Abstract 事件名表示在中间层编码相对时间，可改先后判断，但不能稳定改 onset。这项工作用机制干预检查三个开源 LALM 如何把时间位置绑到声音事件。
用 SAE 分析神经音频编解码器如何编码年龄、性别与口音
Published Sep 28
Evaluated models EnCodec (24 kHz, 6 kbps) , DAC , SpeechTokenizer and others5 Abstract SAE steering 被用作探测 NAC 说话人属性编码的代理。作者用 SAE steering 探测 NAC 表示如何编码说话人属性，目标是可解释性，不是属性转换。
You have reached the end