XGenerationLab/XiYanSQL-QwenCoder-7B-2504
The XiYanSQL-QwenCoder-7B-2504 is a 7.6 billion parameter SQL generation model developed by XGenerationLab, building upon the QwenCoder architecture. It is specifically optimized for converting natural language to SQL queries across multiple dialects, including SQLite, PostgreSQL, and MySQL. This model incorporates fine-tuning and GRPO training strategies to achieve high efficiency and accuracy in SQL generation, demonstrating strong performance on both in-domain and out-of-domain datasets, including a real-world DW test set.
Loading preview...
What is XiYanSQL-QwenCoder-7B-2504?
XiYanSQL-QwenCoder-7B-2504 is a 7.6 billion parameter model from XGenerationLab, designed for advanced Text-to-SQL generation. It represents an optimized iteration of the XiYanSQL-QwenCoder series, focusing on enhancing performance and generalization capabilities for SQL query creation from natural language.
Key Capabilities
- Optimized SQL Generation: Leverages a unique combination of fine-tuning and GRPO (Generative Reinforcement Pre-training Optimization) training strategies for efficient and accurate SQL output without a complex thinking process.
- Multi-Dialect Support: Out-of-the-box support for mainstream SQL dialects including SQLite, PostgreSQL, and MySQL, making it versatile for various database environments.
- Enhanced Generalization: Demonstrates improved performance on out-of-domain datasets and real-world complex queries, as evidenced by its evaluation on the DW test set.
- Strong Benchmarking: Achieves competitive results across various Text-to-SQL benchmarks like BIRD and Spider, and notably on the internal DW test set which includes thousands of complex real-world queries.
When to Use This Model
- Natural Language to SQL Conversion: Ideal for applications requiring the translation of user queries into executable SQL statements.
- Multi-Database Environments: Suitable for projects that interact with different SQL databases (SQLite, PostgreSQL, MySQL) and need consistent SQL generation.
- Complex Query Generation: Recommended for scenarios involving intricate database schemas and demanding complex, real-world SQL queries.
- Research and Development: Useful for researchers exploring advanced Text-to-SQL techniques, particularly those interested in GRPO training and multi-dialect performance.
Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.