仅用9个情绪词+50段故事，就能找到跨模态情感轴。
Nine Emotion Centroids: A Label-Free Valence Axis That Transfers Across Four Modalities
- 用9类情绪故事平均嵌入，提取主成分方向作为情感轴。
- 在文本、图像、音频、脑电上均达93%监督性能，相关性高达0.636。
- 该方向可迁移至多模态且无需目标标签，适合跨模态情感分析研究者。
现代语言模型中存在一条内部方向，能追踪句子的情感极性。我们仅需9个情绪类别名称和每类50段简短叙事（共约1500条标注），即可定位该情感轴（V轴），且该方向在视觉、音频及人脑编码器中均出现，这些模型未联合训练。方法：将九组情绪锚定的故事嵌入冻结编码器，取其平均嵌入的主成分方向。投影新输入到该方向，在SST-2上达到93%监督性能（Llama-3-8B-Instruct，AUC 0.772 vs. 0.828），在11,811张EmoSet图像上与人类情感评分相关性r=0.636，ESC-50音频数据集上达AUC 0.906（p<2.2e-15），123名受试者脑电数据上为AUC 0.720±0.055（p<3.65e-8）。该方向具有机制活性：消融后情感准确率下降5.5–37.2个百分点，而随机方向最多下降0.88个百分点（z>12）。基于文本标签训练的2参数分类器可无标签迁移至图像（AUC 0.961）、音频（0.764）和脑记录（0.828）；通用16维子空间则表现随机（0.525）。该方法限于连续属性——七项分类概念测试返回近随机结果；且方向迁移具家族特异性（Llama/Mistral可行，Qwen/Gemma不可行）。
原文摘要 · Abstract (English)
Inside a modern language model sits a single internal direction that tracks how positive or negative a sentence feels. We show how to find this valence axis (V-axis) from just 9 emotion category names plus 50 short narrative paragraphs per emotion -- about 1,500 fewer labels than the usual supervised approach -- and that the same direction appears in vision, audio, and human-brain encoders never jointly trained. The recipe: embed nine emotion-anchored story sets in a frozen encoder, take the top principal direction of the nine averaged embeddings. Projecting new inputs onto it captures 93% of supervised performance on SST-2 (Llama-3-8B-Instruct, AUC 0.772 vs. 0.828), correlates with human valence ratings on 11,811 EmoSet images at r=0.636, reaches AUC 0.906 on ESC-50 audio (p<2.2e-15), and AUC 0.720+/-0.055 on EEG from 123 subjects (p<3.65e-8). The direction is mechanistically active: ablating it collapses sentiment accuracy by 5.5-37.2 pp across three LLMs vs. at most 0.88 pp for matched random directions (z>12). A 2-parameter classifier trained on text labels transfers to images (AUC 0.961), audio (0.764), and brain recordings (0.828) without target-modality labels; a generic 16-D subspace stays at chance (0.525). The recipe is bounded to continuous attributes -- seven tests on categorical concepts return near-chance -- and steering is family-specific (Llama/Mistral yes, Qwen/Gemma no).
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。