PExA: Parallel Exploration Agent for Complex Text-to-SQL

2026-04-24

AI Summary

This work addresses the challenge of balancing accuracy and latency in large language models for Text-to-SQL tasks by proposing a test coverage–driven iterative generation approach. The method reframes SQL generation as a software testing coverage problem, leveraging parallel execution of atomic SQL test cases to dynamically verify semantic coverage and guide a large language model agent toward producing the final query through informed exploration. By introducing, for the first time, a parallel test case exploration mechanism, the approach significantly enhances accuracy on complex queries while maintaining low latency. Evaluated on the Spider 2.0 benchmark, the method achieves a state-of-the-art execution accuracy of 70.2%.

Technology Category

• Natural Language Processing: (Large) Language Models
• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

LLM-based agents for text-to-SQL often struggle with latency-performance trade-off, where performance improvements come at the cost of latency or vice versa. We reformulate text-to-SQL generation within the lens of software test coverage where the original query is prepared with a suite of test cases with simpler, atomic SQLs that are executed in parallel and together ensure semantic coverage of the original query. After iterating on test case coverage, the final SQL is generated only when enough information is gathered, leveraging the explored test case SQLs to ground the final generation. We validated our framework on a state-of-the-art benchmark for text-to-SQL, Spider 2.0, achieving a new state-of-the-art with 70.2% execution accuracy.

Problem

• text-to-SQL
• latency-performance trade-off
• LLM-based agents
• query generation

Innovation

• Parallel Exploration
• Test Coverage
• Text-to-SQL
• LLM-based Agent
• Execution Accuracy