SQLCritic: Correcting Text-to-SQL Generation via Clause-wise Critic
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913849267978240 |
|---|---|
| author | Chen, Jikai Gan, Leilei Zhao, Ziyu Wang, Zechuan Wang, Dong Zhuang, Chenyi |
| author_facet | Chen, Jikai Gan, Leilei Zhao, Ziyu Wang, Zechuan Wang, Dong Zhuang, Chenyi |
| contents | Existing refinement methods in LLM-based Text-to-SQL systems exhibit limited effectiveness. They often introduce new errors during the self-correction process and fail to detect and correct semantic inaccuracies. To address these gaps, we first introduce a clause-wise critique generation task along with a benchmark, SQLCriticBench, which performs fine-grained error localization including both syntax and semantic errors at the clause level. Furthermore, we introduce a variant of DPO for training our SQLCritic model, where the $β$ coefficient is adaptively changed according to the clause-level inconsistencies between the preferred and dispreferred critiques. We also propose an automatically training dataset curation pipeline which annotate clause-wise critique at scale in a cost-effective way. Experiments demonstrate that the SQLCritic model significantly improves SQL accuracy on the BIRD and Spider datasets, and the results on SQLCriticBench further reveals its superior critique capabilities compared to existing models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_07996 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SQLCritic: Correcting Text-to-SQL Generation via Clause-wise Critic Chen, Jikai Gan, Leilei Zhao, Ziyu Wang, Zechuan Wang, Dong Zhuang, Chenyi Artificial Intelligence Computation and Language Existing refinement methods in LLM-based Text-to-SQL systems exhibit limited effectiveness. They often introduce new errors during the self-correction process and fail to detect and correct semantic inaccuracies. To address these gaps, we first introduce a clause-wise critique generation task along with a benchmark, SQLCriticBench, which performs fine-grained error localization including both syntax and semantic errors at the clause level. Furthermore, we introduce a variant of DPO for training our SQLCritic model, where the $β$ coefficient is adaptively changed according to the clause-level inconsistencies between the preferred and dispreferred critiques. We also propose an automatically training dataset curation pipeline which annotate clause-wise critique at scale in a cost-effective way. Experiments demonstrate that the SQLCritic model significantly improves SQL accuracy on the BIRD and Spider datasets, and the results on SQLCriticBench further reveals its superior critique capabilities compared to existing models. |
| title | SQLCritic: Correcting Text-to-SQL Generation via Clause-wise Critic |
| topic | Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2503.07996 |