arXiv:2603.03742v2 Announce Type: replace 
Abstract: Despite the remarkable performance of large language models (LLMs) in text-to-SQL (SQL generation), correctly producing SQL queries remains challenging during initial generation. The SQL refinement task is subsequently introduced to correct syntactic and semantic errors in generated SQL queries. However, existing paradigms face two major limitations: (i) self-debugging becomes increasingly ineffective as modern LLMs rarely produce explicit execution errors that can trigger debugging signals; (ii) self-correction exhibits low detection precision due to the lack of explicit error modeling grounded in the question and schema, and suffers from severe hallucination that frequently corrupts correct SQLs. In this paper, we propose ErrorLLM, a framework that explicitly models text-to-SQL Errors within a dedicated LLM for text-to-SQL refinement. Specifically, we represent the user question and database schema as structural features, employ static detection to identify execution failures and surface mismatches, and extend ErrorLLM's semantic space with dedicated error tokens that capture categorized implicit semantic error types. Through a well-designed training strategy, we explicitly model these errors with structural representations, enabling the LLM to detect complex implicit errors by predicting dedicated error tokens. Guided by the detected errors, we perform error-guided refinement on the SQL structure by prompting LLMs. Extensive experiments demonstrate that ErrorLLM achieves the most significant improvements over backbone initial generation. Further analysis reveals that detection quality directly determines refinement effectiveness, and ErrorLLM addresses both sides by high detection F1 score while maintain refinement effectiveness.

يمثل إدخال ErrorLLM تقدمًا كبيرًا في مجال تحسين النص إلى SQL، حيث يتناول التحديات المستمرة في إنتاج استعلامات SQL دقيقة من مدخلات اللغة الطبيعية. يقوم هذا الإطار بنمذجة أخطاء SQL بشكل صريح، مما يعزز قدرة النماذج اللغوية الكبيرة (LLMs) على تصحيح الأخطاء النحوية والدلالية خلال عملية توليد الاستعلامات.

La introducción de ErrorLLM marca un avance significativo en el campo del refinamiento de texto a SQL, abordando los desafíos persistentes de generar consultas SQL precisas a partir de entradas en lenguaje natural. Este marco modela explícitamente los errores de SQL, mejorando la capacidad de los grandes modelos de lenguaje (LLMs) para corregir errores sintácticos y semánticos durante el proceso de generación de consultas.

L'introduction d'ErrorLLM représente une avancée significative dans le domaine du raffinement du texte en SQL, en s'attaquant aux défis persistants de la génération de requêtes SQL précises à partir d'entrées en langage naturel. Ce cadre modélise explicitement les erreurs SQL, améliorant la capacité des grands modèles de langage (LLMs) à corriger les erreurs syntaxiques et sémantiques lors du processus de génération de requêtes.

The introduction of ErrorLLM marks a significant advancement in the field of text-to-SQL refinement, addressing the persistent challenges of generating accurate SQL queries from natural language inputs. This framework explicitly models SQL errors, enhancing the ability of large language models (LLMs) to correct syntactic and semantic mistakes during the query generation process.