SQL-Trail: Multi-Turn Reinforcement Learning with Interleaved Feedback for Text-to-SQL

TL;DR

SQL-Trail introduces a multi-turn reinforcement learning framework for Text-to-SQL that iteratively refines queries through interaction and feedback, significantly improving accuracy and efficiency over traditional single-pass models.

Contribution

It presents a novel multi-turn RL approach with adaptive interaction and a composite reward, advancing state-of-the-art performance in Text-to-SQL tasks.

Findings

Sets new state-of-the-art results on benchmarks.

Achieves up to 18x higher data efficiency.

Outperforms larger proprietary models by 5% on average.

Abstract

While large language models (LLMs) have substantially improved Text-to-SQL generation, a pronounced gap remains between AI systems and human experts on challenging benchmarks such as BIRD-SQL. We argue this gap stems largely from the prevailing single-pass paradigm, which lacks the iterative reasoning, schema exploration, and error-correction behaviors that humans naturally employ. To address this limitation, we introduce SQL-Trail, a multi-turn reinforcement learning (RL) agentic framework for Text-to-SQL. Rather than producing a query in one shot, SQL-Trail interacts with the database environment and uses execution feedback to iteratively refine its predictions. Our approach centers on two key ideas: (i) an adaptive turn-budget allocation mechanism that scales the agent's interaction depth to match question difficulty, and (ii) a composite reward panel that jointly incentivizes SQL correctness and efficient exploration. Across benchmarks, SQL-Trail sets a new state of the art and delivers strong data efficiency--up to 18x higher than prior single-pass RL state-of-the-art methods. Notably, our 7B and 14B models outperform substantially larger proprietary systems by 5% on average, underscoring the effectiveness of interactive, agentic workflows for robust Text-to-SQL generation.