Metraipolis › archive › dc_17da682ba211
SQL-Zero: Self-Evolving Text-to-SQL — arXiv:2609.04697
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.04697
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T22:20:51.127Z
- SHA-256 of the archived text
08db64966a3af428ec99351e3b5bee83a0dce5a96a6a09b12677a9b31f0e9520- Stable URL of this page
https://metraipolis.com/documento/dc_17da682ba211
How to cite
SQL-Zero: Self-Evolving Text-to-SQL — arXiv:2609.04697 — arXiv (en), https://arxiv.org/abs/2609.04697 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_17da682ba211 (Metraipolis archive dc_17da682ba211, sha256 08db64966a3af428ec99351e3b5bee83a0dce5a96a6a09b12677a9b31f0e9520)
BibTeX
@misc{dc17da682ba211,
title = {SQL-Zero: Self-Evolving Text-to-SQL — arXiv:2609.04697},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.04697},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_17da682ba211; sha256 08db64966a3af428ec99351e3b5bee83a0dce5a96a6a09b12677a9b31f0e9520},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.04697. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Training a competitive Text-to-SQL agent usually depends on human-annotated natural-language/SQL pairs, which are expensive, domain-specific, and a bottleneck for scaling to new databases. We show it is possible to train a competitive solver with zero annotated pairs. We introduce SQL-Zero, a proposer-solver self-play in which a challenger and a solver start from the same base LLM and the only ground truth is execution against the database itself. The challenger generates SQL pairs calibrated to the solver's current difficulty (targeting "hard but solvable"), and both roles are updated with GRPO in alternating turns, with a template-level repetition penalty on the challenger to prevent diversity collapse. Training on BIRD databases with no labels, self-play improves over the zero-shot base on BIRD dev by 6.6 points at 3B and 7.3 points at 7B. It also scores higher than a matched control trained under the same recipe on human BIRD gold over the same databases, although an exact paired test does not resolve that margin. Transfer depends on scale: at 3B every iteration outperforms the base on unseen Spider databases and under lexical perturbation (Spider-Syn), where it also degrades less than the matched BIRD-gold control, whereas at 7B only the first iteration preserves transfer. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-07 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.04697 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T22:18:11.379Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).