twinkle-ai/twinkle-sqlcoder
twinkle-ai/twinkle-sqlcoder is a 24 billion parameter full-parameter SFT (Supervised Fine-Tuning) checkpoint for SQL generation, built upon the mistralai/Devstral-Small-2505 base model. This model is specifically optimized for text-to-SQL tasks, leveraging a diverse training dataset including Spider, BIRD, and WikiSQL. It excels at converting natural language queries into SQL code, making it suitable for research and evaluation in SQL generation.
Loading preview...
Overview
twinkle-ai/twinkle-sqlcoder is a 24 billion parameter model developed by twinkle-ai, specifically fine-tuned for text-to-SQL generation. It is built on the mistralai/Devstral-Small-2505 base model, utilizing the MistralForCausalLM architecture. The model was trained using bf16 precision with a maximum sequence length of 4096 tokens.
Key Capabilities
- SQL Generation: Translates natural language into SQL queries.
- Full-Parameter SFT: Achieves specialized performance through comprehensive supervised fine-tuning.
- Diverse Training: Leverages a merged dataset including
spider,bird,synsql-2.5m,wikisql, and synthetic data for robust SQL understanding.
Intended Use Cases
- Text-to-SQL Research: Ideal for academic and experimental work in converting natural language to SQL.
- Benchmarking: Suitable for evaluating and comparing SQL generation performance against other models.
Limitations
- Generated SQL may require validation before production deployment due to potential inaccuracies.
- Performance is influenced by prompt formatting, the quality of schema context provided, and decoding settings.
- Users should assess safety and compliance requirements prior to deployment.