EMNLP Findings2025
ExeSQL: Self-Taught Text-to-SQL Models with Execution-Driven Bootstrapping for SQL Dialects
Jipeng Zhang, Haolin Yang, Kehao Miao, Ruiyuan Zhang, Renjie Pi, Jiahui Gao, Xiaofang Zhou
Abstract
Recent text-to-SQL models have achieved strong performance, but their effectiveness remains largely confined to SQLite due to dataset limitations. However, real-world applications require SQL generation across multiple dialects with varying syntax and specialized features, which remains a challenge for current models.
The main obstacle in building a dialect-aware model lies in acquiring high-quality dialectspecific data. Data generated purely through static prompting-without validating SQLs via execution-tends to be noisy and unreliable.Moreover, the lack of real execution environments in the training loop prevents models from grounding their predictions in executable semantics, limiting generalization despite surface-level improvements from data filtering. This work introduces ExeSQL, a text-to-SQL framework with execution-driven, agentic bootstrapping. The method consists of iterative query generation, execution-based filtering (e.g., rejection sampling), and preferencebased training, enabling the model to adapt to new SQL dialects through verifiable, feedbackguided learning. Experiments show that Ex-eSQL bridges the dialect gap in text-to-SQL, achieving average improvements of 15.2%, 10.38%, and 4.49% over GPT-4o on Post-greSQL, MySQL, and Oracle, respectively, across multiple datasets of varying difficulty.
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.
4 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- WizardLM: Empowering Large Pre-Trained Language Models to Follow Complex Instructions cites Language Models are Few-Shot Learners
- WizardLM: Empowering Large Pre-Trained Language Models to Follow Complex Instructions cites Training language models to follow instructions with human feedback
- Direct Preference Optimization: Your Language Model is Secretly a Reward Model cites Language Models are Few-Shot Learners
- Direct Preference Optimization: Your Language Model is Secretly a Reward Model cites Training language models to follow instructions with human feedback
- MARS-SQL: A Multi-Agent Reinforcement Learning Framework For Text-To-SQL cites On Stable Long-Form Generation: Benchmarking and Mitigating Length Volatility
- On Stable Long-Form Generation: Benchmarking and Mitigating Length Volatility cites MARS-SQL: A Multi-Agent Reinforcement Learning Framework For Text-To-SQL
- LIMA: Less Is More for Alignment cites Training language models to follow instructions with human feedback
- Training language models to follow instructions with human feedback cites Language Models are Few-Shot Learners
Cited by top-tier papers4
- MARS-SQL: A Multi-Agent Reinforcement Learning Framework For Text-To-SQLHaolin Yang, Jipeng Zhang, Zhitao He, Alexander Zhou et al.ICML 2026 · 12 citations
- Dial: A Knowledge-Grounded Dialect-Specific NL2SQL SystemXiang Zhang, Le Zhou, Hongming Xu, Wei Zhou et al.VLDB 2026 · 2 citations
- On Stable Long-Form Generation: Benchmarking and Mitigating Length VolatilityZhitao He, Haolin Yang, Rui Min, Zeyu Qin et al.ICML 2026
- Dialect-Agnostic SQL Parsing via LLM-Based SegmentationJunwen An, Kabilan Mahathevan, Manuel RiggerSIGMOD 2026
Builds on22
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- Training language models to follow instructions with human feedbackLong Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida et al.NeurIPS 2022 · 24,707 citations
- 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
- LIMA: Less Is More for AlignmentChunting Zhou, Pengfei Liu, Puxin Xu, Srinivasan Iyer et al.NeurIPS 2023 · 1,486 citations
- WizardLM: Empowering Large Pre-Trained Language Models to Follow Complex InstructionsCan Xu, Qingfeng Sun, Kai Zheng, Xiubo Geng et al.ICLR 2024 · 1,206 citations
Related papers
- Dialect-SQL: An Adaptive Framework for Bridging the Dialect Gap in Text-to-SQLJie Shi, Xi Cao, Bo Xu, Jiaqing Liang et al.EMNLP 2025 · 2 citations
- OmniSQL: Synthesizing High-quality Text-to-SQL Data at ScaleHaoyang Li, Shang Wu, Xiaokang Zhang, Xinmei Huang et al.VLDB 2025 · 90 citations
- Knowledge-to-SQL: Enhancing SQL Generation with Data Expert LLMZijin Hong, Zheng Yuan, Hao Chen, Qinggang Zhang et al.ACL Findings 2024 · 31 citations
- ReSQL: Self-Improving Framework for Reasoning-Aware Text-to-SQL Dataset GenerationMinjun Park, Yongju Seong, Myoseop Sim, Kyungkoo Min et al.ACL Findings 2026
- SAFE-SQL: Self-Augmented In-Context Learning with Fine-grained Example Selection for Text-to-SQLJimin Lee, Ingeol Baek, Byeongjeong Kim, Hyunkyung Bae et al.EMNLP 2025 · 1 citation
