Guardado en:
Detalles Bibliográficos
Autores principales: Li, Anni, Attar, Aria, Dong, Paul
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2510.00186
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912623865364480
author Li, Anni
Attar, Aria
Dong, Paul
author_facet Li, Anni
Attar, Aria
Dong, Paul
contents Transforming natural-language requests into reliable, production-ready data transformations remains challenging: correctness depends on precise schema linking and warehouse-specific SQL dialects, while the strongest supervision available during training--execution success and result matching--are provided only at the sequence level. At the same time, assembling large, execution-validated corpora is costly, and token-level objectives misalign with these global signals, yielding unstable optimization and limited portability. We introduce Thinkquel, a fine-tuned model for producing robust, portable, and execution-validated database queries. Methodologies in Thinkquel integrates a novel synthetic data pipeline, TS-SQL, that leverages dbt as a portable intermediate representation with a span-aware reinforcement learning objective, and Token-Sequence GRPO (TS-GRPO), specifically designed to bridge the gap between token-level training signals and sequence-level execution rewards when finetuning LLMs. On the 500-example TS-SQL test set, Thinkquel (32B) reaches 93.2% execution success and 61.8% exact-result match with a two-stage SFT curriculum, improving over the base model by 67.2% (exec.) and 44.4% (match). In Spider (14B) experiments, TS-GRPO increases training stability and speeds convergence of the execution-match reward relative to GRPO and GSPO.
format Preprint
id arxiv_https___arxiv_org_abs_2510_00186
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Thinkquel: A Model Dedicated to Text-to-dbt Using Synthetic Data and a Span-Aware Objective
Li, Anni
Attar, Aria
Dong, Paul
Artificial Intelligence
Machine Learning
Transforming natural-language requests into reliable, production-ready data transformations remains challenging: correctness depends on precise schema linking and warehouse-specific SQL dialects, while the strongest supervision available during training--execution success and result matching--are provided only at the sequence level. At the same time, assembling large, execution-validated corpora is costly, and token-level objectives misalign with these global signals, yielding unstable optimization and limited portability. We introduce Thinkquel, a fine-tuned model for producing robust, portable, and execution-validated database queries. Methodologies in Thinkquel integrates a novel synthetic data pipeline, TS-SQL, that leverages dbt as a portable intermediate representation with a span-aware reinforcement learning objective, and Token-Sequence GRPO (TS-GRPO), specifically designed to bridge the gap between token-level training signals and sequence-level execution rewards when finetuning LLMs. On the 500-example TS-SQL test set, Thinkquel (32B) reaches 93.2% execution success and 61.8% exact-result match with a two-stage SFT curriculum, improving over the base model by 67.2% (exec.) and 44.4% (match). In Spider (14B) experiments, TS-GRPO increases training stability and speeds convergence of the execution-match reward relative to GRPO and GSPO.
title Thinkquel: A Model Dedicated to Text-to-dbt Using Synthetic Data and a Span-Aware Objective
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2510.00186