CogniSQL-R1-Zero: Lightweight Reinforced Reasoning for Efficient SQL Generation

TL;DR

CogniSQL-R1-Zero introduces a lightweight reinforcement learning framework that significantly improves SQL generation accuracy from natural language, achieving state-of-the-art results with less computational resources.

Contribution

The paper presents a novel RL-based approach for Text-to-SQL translation that avoids complex supervision and reward shaping, demonstrating superior performance with a smaller model and training setup.

Findings

Achieves state-of-the-art execution accuracy on Text2SQL benchmark.

Outperforms larger supervised models despite smaller size.

Efficient training on limited hardware resources.

Abstract

Translating natural language into SQL (Text-to-SQL) remains a core challenge at the intersection of language understanding and structured data access. Although large language models (LLMs) have improved fluency, generating correct and executable SQL, especially for complex queries, continues to be challenging. We introduce CogniSQL-R1-Zero, a reinforcement learning (RL) framework and model that produces accurate SQL using a lightweight reward signal based on execution correctness and format-tag compliance. By avoiding intermediate supervision, hybrid pipelines and complex reward shaping, our method encourages stable learning and stronger alignment with the ultimate task objective-producing executable programs. CogniSQL-R1-Zero achieves state-of-the-art execution accuracy on Text2SQL benchmark; BIRD bench, outperforming prior supervised and instruction-tuned baselines including SFT CodeS-7B, DeepSeek-Coder 236B, and Mistral 123B-despite being trained on a significantly smaller 7B backbone. This result underscores the scalability and efficiency of our RL-based approach when trained on just four NVIDIA A100 GPUs (40 GB VRAM each). To support further research in efficient and interpretable Text-to-SQL modeling, we release two curated datasets: (i) a collection of 5,024 reasoning traces with varying context lengths, and (ii) a positive-sampled corpus of 36,356 corpus of weakly supervised queries, each annotated with six semantically diverse reasoning paths. Together, these contributions advance scalable, execution-aligned Text-to-SQL generation.