EMNLP Findings2025
Learning SQL Like a Human: Structure-Aware Curriculum Learning for Text-to-SQL Generation
Xiaohu Zhu, Qian Li, Lizhen Cui, Yuntao Du
Abstract
The Text-to-SQL capabilities of large language allow users to interact with databases using natural language. Despite recent advances, existing models continue to struggle with complex queries, particularly those requiring multitable joins and reasoning. To address this gap, we propose to construct a model, namely SAC-SQL, with synthetic training samples followed by a structure-aware curriculum learning framework for enhancing SQL generation. Our approach begins with a supervised finetuning (SFT) stage, where we train open-source models on a synthetically constructed, crossdomain SQL dataset with diverse structural patterns. Moreover, we introduce a unified structure difficulty scoring function to partition the training samples into non-overlapping curriculum phases, guiding the model progressively learning from simpler to more complex SQL structures. Extensive experiments are conducted and the results show that SAC-SQL achieves better results than the baselines, and significantly narrows the performance gap between open-source and close-source models on Spider and Bird benchmarks.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- DIN-SQL: Decomposed In-Context Learning of Text-to-SQL with Self-Correction cites Graphix-T5: Mixing Pre-trained Transformers with Graph-Aware Layers for Text-to-SQL Parsing
- Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation cites RESDSQL: Decoupling Schema Linking and Skeleton Parsing for Text-to-SQL
- Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation cites DIN-SQL: Decomposed In-Context Learning of Text-to-SQL with Self-Correction
- Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation cites Graphix-T5: Mixing Pre-trained Transformers with Graph-Aware Layers for Text-to-SQL Parsing
Builds on12
- Direct Preference Optimization: Your Language Model is Secretly a Reward ModelRafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D. Manning et al.NeurIPS 2023 · 10,924 citations
- DIN-SQL: Decomposed In-Context Learning of Text-to-SQL with Self-CorrectionMohammadreza Pourreza, Davood RafieiNeurIPS 2023 · 909 citations
- Text-to-SQL Empowered by Large Language Models: A Benchmark EvaluationDawei Gao, Haibin Wang, Yaliang Li, Xiuyu Sun et al.VLDB 2024 · 609 citations
- RESDSQL: Decoupling Schema Linking and Skeleton Parsing for Text-to-SQLHaoyang Li, Jing Zhang, Cuiping Li, Hong ChenAAAI 2023 · 343 citations
- Graphix-T5: Mixing Pre-trained Transformers with Graph-Aware Layers for Text-to-SQL ParsingJinyang Li, Binyuan Hui, Reynold Cheng, Bowen Qin et al.AAAI 2023 · 164 citations
Related papers
- OpenSQL: Data-Efficient Text-to-SQL for Open-Source LLMs via Synthesized Intermediate SupervisionRuilin Hu, Yuyu Luo, Guoliang Li, Shuangqiao Wu et al.VLDB 2026 · 4 citations
- G³R: A Graph-Guided Generate-and-Rerank Framework for Complex and Cross-domain Text-to-SQL GenerationYanzheng Xiang, Qian-Wen Zhang, Xu Zhang, Zejie Liu et al.ACL Findings 2023 · 3 citations
- SchemaRAG: A Schema-aware Retrieval-Augmented Generation Framework for Text-to-SQLDi Wu, Zetong Tang, Yi He, Xin LuoSIGMOD 2026 · 9 citations
- Synthesizing Text-to-SQL Data from Weak and Strong LLMsJiaxi Yang, Binyuan Hui, Min Yang, Jian Yang et al.ACL 2024
- SQLForge: Synthesizing Reliable and Diverse Data to Enhance Text-to-SQL Reasoning in LLMsYu Guo, Dong Jin, Shenghao Ye, Shuangwu Chen et al.ACL Findings 2025 · 3 citations
