用统一损失联合优化文本转SQL的链接与生成，提升抗噪声能力。
JOLT-SQL: Joint Loss Tuning of Text-to-SQL with Confusion-aware Noisy Schema Sampling
- 单阶段微调框架，统一优化模式链接与SQL生成。
- 在Spider和BIRD上达到开源模型最优执行准确率。
- 通过噪声模式采样增强鲁棒性，训练推理更高效。
文本转SQL任务近年来受益于大语言模型（LLM）的发展。尽管提示和监督微调（SFT）等范式已广泛应用，但传统SFT方法仍面临多阶段流程复杂、对噪声模式信息敏感等问题。为此，我们提出JOLT-SQL，一种简化的一阶段SFT框架，通过统一损失联合优化模式链接与SQL生成。该方法采用判别式模式链接机制，结合局部双向注意力，并引入混淆感知的噪声模式采样策略与选择性注意力，显著提升在噪声模式条件下的鲁棒性。在Spider和BIRD基准上的实验表明，JOLT-SQL在同类开源模型中达到了最先进的执行准确率，同时大幅提高训练与推理效率。代码已开源：https://github.com/Songjw133/JOLT-SQL。
原文摘要 · Abstract (English)
Text-to-SQL, which maps natural language to SQL queries, has benefited greatly from recent advances in Large Language Models (LLMs). While LLMs offer various paradigms for this task, including prompting and supervised fine-tuning (SFT), SFT approaches still face challenges such as complex multi-stage pipelines and poor robustness to noisy schema information. To address these limitations, we present JOLT-SQL, a streamlined single-stage SFT framework that jointly optimizes schema linking and SQL generation via a unified loss. JOLT-SQL employs discriminative schema linking, enhanced by local bidirectional attention, alongside a confusion-aware noisy schema sampling strategy with selective attention to improve robustness under noisy schema conditions. Experiments on the Spider and BIRD benchmarks demonstrate that JOLT-SQL achieves state-of-the-art execution accuracy among comparable-size open-source models, while significantly improving both training and inference efficiency. Our code is available at https://github.com/Songjw133/JOLT-SQL.
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。