用答案发散度筛选指令数据，提升模型性能。
Instruction Data Selection via Answer Divergence
- 基于多生成答案的几何结构选择指令数据
- 仅用1万条选中数据，性能超越多个强基线
- 适合需要高质量指令数据的训练场景
指令微调依赖大规模指令-响应语料库，其质量和构成显著影响下游表现。本文提出答案发散度引导选择（ADG），依据多样本输出的几何结构筛选指令数据。对每条指令生成多组高温采样结果，映射到嵌入空间后计算输出发散度得分，该得分同时反映分散程度和形状各向异性。高分对应答案既分散又多模态的指令，而非单一方向上的近似改写。在两种骨干模型和三个公开指令数据集上，仅使用1万条经ADG选出的数据进行微调，在六个涵盖推理、知识和编码的基准测试中均优于多个强基线。分析表明，分散程度与形状各向异性均不可或缺，支持答案发散度作为实际有效的数据选择信号。代码与附录见补充材料。
原文摘要 · Abstract (English)
Instruction tuning relies on large instruction-response corpora whose quality and composition strongly affect downstream performance. We propose Answer Divergence-Guided Selection (ADG), which selects instruction data based on the geometric structure of multi-sample outputs. ADG draws several high-temperature generations per instruction, maps responses into an embedding space, and computes an output divergence score that jointly encodes dispersion magnitude and shape anisotropy. High scores correspond to instructions whose answers are both far apart and multi-modal, rather than clustered paraphrases along a single direction. Across two backbones and three public instruction pools, fine-tuning on only 10K ADG-selected examples consistently outperforms strong selectors on six benchmarks spanning reasoning, knowledge, and coding. Analyses further show that both dispersion magnitude and shape anisotropy are necessary, supporting answer divergence as a practical signal for instruction data selection. Code and appendix are included in the supplementary materials.
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。