用盒子嵌入分析提示词的细微差异，更准发现大模型弱点。
PROMPT2BOX: Uncovering Entailment Structure among LLM Prompts
- 用训练好的编码器将提示转为盒子嵌入，捕捉语义与具体程度关系
- 在17个大模型上比向量方法多发现8.9%的弱点，相关性提升33%
- 适合想深入分析模型短板或设计评测集的研究者
为发现大模型弱点，研究者常将提示词嵌入向量空间并聚类以提取模式。但向量嵌入主要反映主题相似性，导致主题相同但具体程度不同的提示被视作相似，难以进行细粒度分析。为此，我们提出PROMPT2BOX，使用训练过的编码器将提示嵌入到盒子嵌入空间。该编码器在现有及合成数据集上训练，输出的盒子嵌入不仅能捕捉语义相似性，还能体现提示间的具体程度关系（如“写一个冒险故事”比“写一个故事”更具体）。我们还开发了一种新型盒子嵌入降维技术，便于数据集可视化与比较。实验表明，盒子嵌入在捕捉提示具体程度方面持续优于向量基线。在基于UltraFeedback数据集对17个大模型构建分层聚类树的下游任务中，PROMPT2BOX比向量基线多识别出8.9%的模型弱点，并使分层深度与指令具体程度之间的相关性提升约33%。
原文摘要 · Abstract (English)
To discover the weaknesses of LLMs, researchers often embed prompts into a vector space and cluster them to extract insightful patterns. However, vector embeddings primarily capture topical similarity. As a result, prompts that share a topic but differ in specificity, and consequently in difficulty, are often represented similarly, making fine-grained weakness analysis difficult. To address this limitation, we propose PROMPT2BOX, which embeds prompts into a box embedding space using a trained encoder. The encoder, trained on existing and synthesized datasets, outputs box embeddings that capture not only semantic similarity but also specificity relations between prompts (e.g., "writing an adventure story" is more specific than "writing a story"). We further develop a novel dimension reduction technique for box embeddings to facilitate dataset visualization and comparison. Our experiments demonstrate that box embeddings consistently capture prompt specificity better than vector baselines. On the downstream task of creating hierarchical clustering trees for 17 LLMs from the UltraFeedback dataset, PROMPT2BOX can identify 8.9\% more LLM weaknesses than vector baselines and achieves an approximately 33\% stronger correlation between hierarchical depth and instruction specificity.
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。