通过优化互补技能提升文本转SQL集成效果，无需微调模型。
Residual Skill Optimization for Text-to-SQL Ensembles
- 新技能在当前模型失败的样本上优化，提升整体准确率
- 在Spider2-Lite上最高提升11.1点，且跨方言通用
- 减少幻觉引用3倍，真实提升可靠性而非表面差异
文本转SQL集成通过生成多个SQL候选并选择最优解来提升性能，但其上限受制于Pass@K（至少一个候选正确概率）。现有方法依赖随机解码或提示变体生成多样性，导致候选结果仍高度相关，常集体失败。本文提出DivSkill-SQL，一种残差技能优化框架，在不进行模型微调的前提下构建互补型智能体集成：每个新技能在当前技能集失败的样本上优化，理论上最大化其对Pass@K的边际贡献。在Spider2-Lite数据集上，相较于最强基线，DivSkill-SQL在Snowflake上提升最多达+11.1点，在BigQuery上提升+8.3点，且在两种基础模型（Opus-4.6和GPT-5.4）上均表现一致。单一方言优化的技能可无重训练迁移至Snowflake、BigQuery、SQLite等不同方言，并适用于BIRD-Critic任务（+2.6点）。错误诊断显示，幻觉模式引用与函数调用最多减少3倍，表明性能提升源于真正可靠的互补能力，而非表层形式变化。
原文摘要 · Abstract (English)
Text-to-SQL ensembles improve over single-candidate generation by drawing multiple SQL candidates and selecting one, but their effectiveness is bounded by Pass@K, the probability that at least one of K candidates is correct. Existing methods source diversity heuristically through stochastic decoding or prompt variants, leaving candidate sets dominated by correlated failures. We present DivSkill-SQL, a residual skill optimization framework that builds complementary agentic Text-to-SQL ensembles without model fine-tuning: each new skill is optimized on examples the current skill ensemble fails on, provably targeting its marginal contribution to Pass@K. On Spider2-Lite, DivSkill-SQL improves selected accuracy by up to +11.1 points on Snowflake and +8.3 on BigQuery over the strongest ensemble baseline, with consistent gains across two base models (Opus-4.6 and GPT-5.4). Skills optimized on a single dialect transfer without retraining across dialects (Snowflake, BigQuery, SQLite) and to a different task formulation, such as BIRD-Critic (+2.6 pts). Error diagnostics show up to 3x fewer hallucinated schema references and function calls, indicating that gains come from genuinely reliable complementary skills rather than surface-form variation.
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。