Blar-SQL: Faster, Stronger, Smaller NL2SQL

TL;DR

Blar-SQL introduces a task decomposition approach with fine-tuned open-source models and schema chunking, achieving high accuracy in NL2SQL tasks while being significantly smaller, faster, and cheaper than GPT-4.

Contribution

The paper presents a novel framework combining task decomposition, model fine-tuning, and schema chunking to improve NL2SQL performance with open-source models.

Findings

Comparable accuracy to GPT-4

135 times smaller model size

90 times faster and over 100 times cheaper

Abstract

Large Language Models (LLMs) have gained considerable notoriety in the field of natural language to SQL tasks (NL2SQL). In this study, we show how task decomposition can greatly benefit LLMs in database understanding and query generation in order to answer human questions with an SQL query. We fined-tuned open source models, specifically Llama-2 and Code Llama, by combining 2 different models each designated to focus on one of two tasks in order to leverage each model's core competency to further increase the accuracy of the final SQL query. We propose a new framework to divide the schema into chunks in order to fit more information into a limited context. Our results are comparable with those obtained by GPT-4 at the same time being 135 times smaller, 90 times faster and more than 100 times cheaper than GPT-4.