Cheaper, Better, Faster, Stronger: Robust Text-to-SQL without Chain-of-Thought or Fine-Tuning

2025-05-20

AI Summary

Large language models (LLMs) for text-to-SQL suffer from high inference costs and poor robustness due to reliance on Chain-of-Thought (CoT) prompting or fine-tuning. Method: We propose N-rep consistency—a zero-shot, CoT-free, and fine-tuning-free framework that generates N lightweight, multi-view structural representations of the database schema to enable cross-perspective result verification. Contribution/Results: Our method establishes the first ultra-low-budget lightweight inference framework for text-to-SQL. On the BIRD benchmark, it achieves state-of-the-art performance with a per-query cost of only $0.039—over 10× lower than mainstream approaches. Crucially, it significantly improves robustness against schema perturbations and natural language ambiguities without sacrificing accuracy, thereby introducing a new paradigm for cost-effective and reliable text-to-SQL generation.

Technology Category

• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Learning to Search

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

LLMs are effective at code generation tasks like text-to-SQL, but is it worth the cost? Many state-of-the-art approaches use non-task-specific LLM techniques including Chain-of-Thought (CoT), self-consistency, and fine-tuning. These methods can be costly at inference time, sometimes requiring over a hundred LLM calls with reasoning, incurring average costs of up to $0.46 per query, while fine-tuning models can cost thousands of dollars. We introduce"N-rep"consistency, a more cost-efficient text-to-SQL approach that achieves similar BIRD benchmark scores as other more expensive methods, at only $0.039 per query. N-rep leverages multiple representations of the same schema input to mitigate weaknesses in any single representation, making the solution more robust and allowing the use of smaller and cheaper models without any reasoning or fine-tuning. To our knowledge, N-rep is the best-performing text-to-SQL approach in its cost range.

Problem

• Reducing high costs of text-to-SQL methods
• Eliminating need for Chain-of-Thought or fine-tuning
• Improving robustness with multi-schema representations

Innovation

• Uses N-rep consistency for cost efficiency
• Leverages multiple schema representations for robustness
• Avoids reasoning and fine-tuning with smaller models