XGenerationLab/XiYanSQL-QwenCoder-7B-2504
The XiYanSQL-QwenCoder-7B-2504 is a 7.6 billion parameter SQL generation model developed by XGenerationLab, built upon the QwenCoder architecture. It specializes in generating SQL queries across multiple dialects (SQLite, PostgreSQL, MySQL) and features a unique combination of fine-tuning and GRPO training for enhanced efficiency and accuracy. This model demonstrates strong generalization capabilities, performing well on both in-domain and out-of-domain datasets, including a real-world DW test set.
Loading preview...
XiYanSQL-QwenCoder-7B-2504 Overview
The XiYanSQL-QwenCoder-7B-2504, developed by XGenerationLab, is a 7.6 billion parameter model specifically designed for SQL generation. This version optimizes previous iterations by integrating fine-tuning with GRPO training, a post-training strategy that enhances both efficiency and accuracy in SQL query generation without requiring a thinking process.
Key Capabilities
- Multi-Dialect Support: Supports mainstream SQL dialects including SQLite, PostgreSQL, and MySQL.
- Enhanced Generalization: Demonstrates improved performance on diverse dialects and out-of-domain datasets.
- Robust Evaluation: Evaluated against standard benchmarks like BIRD and Spider, and a real-world DW test set comprising thousands of complex queries from PostgreSQL and MySQL scenarios.
- Performance: Achieves 62.13% on BIRD Dev@M-Schema and 85.97% on Spider Test@M-Schema, outperforming several larger models in specific SQL generation tasks.
When to Use This Model
This model is ideal for applications requiring accurate and efficient text-to-SQL conversion, especially in environments with varied SQL dialects. Its strong performance on real-world and out-of-domain SQL benchmarks makes it suitable for complex database interaction tasks.