将多标签分类拆成独立的二元判断，提升大模型推理效率。
Divide, Cache, Conquer: Dichotomic Prompting for Efficient Multi-Label LLM-Based Classification
- 把多标签任务转为一系列是/否问答，逐维独立判断。
- 在24个情感维度上，微调后模型性能显著优于零样本基线。
- 适合需要高效推理且需多标签输出的应用场景。
我们提出一种基于大语言模型（LLMs）的高效多标签文本分类方法，将分类任务重构为一系列二元（是/否）决策。不一次性生成所有标签，而是独立查询每个目标维度，结合前缀缓存机制，在短文本推理中实现显著效率提升，且不损失准确率。以情感文本分析为例，涵盖24个维度（包括情绪与情感）。通过LLM到小模型（SLM）的蒸馏，强标注模型（DeepSeek-V3）对每条文本生成多个标注，经聚合后用于微调小型模型（HerBERT-Large、CLARIN-1B、PLLuM-8B、Gemma3-1B）。微调后的模型在训练时见过的维度上表现显著优于零样本基线。结果表明，将多标签分类分解为二元查询，并结合蒸馏与缓存感知推理，可构建一个可扩展、高效的LLM分类框架。尽管在情感状态上验证，该方法具有广泛适用性。
原文摘要 · Abstract (English)
We introduce a method for efficient multi-label text classification with large language models (LLMs), built on reformulating classification tasks as sequences of dichotomic (yes/no) decisions. Instead of generating all labels in a single structured response, each target dimension is queried independently, which, combined with a prefix caching mechanism, yields substantial efficiency gains for short-text inference without loss of accuracy. To demonstrate the approach, we focus on affective text analysis, covering 24 dimensions including emotions and sentiment. Using LLM-to-SLM distillation, a powerful annotator model (DeepSeek-V3) provides multiple annotations per text, which are aggregated to fine-tune smaller models (HerBERT-Large, CLARIN-1B, PLLuM-8B, Gemma3-1B). The fine-tuned models show significant improvements over zero-shot baselines, particularly on the dimensions seen during training. Our findings suggest that decomposing multi-label classification into dichotomic queries, combined with distillation and cache-aware inference, offers a scalable and effective framework for LLM-based classification. While we validate the method on affective states, the approach is general and applicable across domains.
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。