The XiYanSQL-QwenCoder-3B-2502 model, developed by XGenerationLab, is a 3.1 billion parameter language model specifically fine-tuned for text-to-SQL tasks. With a context length of 32768 tokens, it excels at generating SQL queries from natural language prompts across multiple dialects including SQLite, PostgreSQL, and MySQL. This model is designed to serve as an effective solution for direct SQL generation or as a robust base for further fine-tuning in the text-to-SQL domain.
Loading preview...
XiYanSQL-QwenCoder-3B-2502: Specialized Text-to-SQL Model
The XiYanSQL-QwenCoder-3B-2502 is part of the XiYanSQL-QwenCoder series by XGenerationLab, specifically engineered to advance large language models in the text-to-SQL domain. This 3.1 billion parameter model is optimized for generating SQL queries from natural language, supporting a 32768 token context length.
Key Capabilities
- High Performance in SQL Generation: Achieves competitive results on benchmarks like BIRD and Spider, with the 32B variant setting a new SOTA EX score of 69.03% on the BIRD TEST set.
- Multi-Dialect Support: Capable of generating SQL for mainstream dialects including SQLite, PostgreSQL, and MySQL.
- Flexible Schema Formats: Supports both M-Schema and original DDL formats for database schema descriptions.
Good for
- Direct Text-to-SQL Applications: Ideal for converting natural language questions into executable SQL queries.
- Fine-tuning Base: Serves as an excellent starting point for developing more specialized SQL generation models.
- Database Interaction: Automating data retrieval and manipulation through natural language interfaces.