LLMSQL Organization on Hugging Face

Visit our Github page →

Description:
LLMSQL is a patched and improved version of WikiSQL for Text-to-SQL tasks. We provide fully prepared datasets, scripts for SQL generation, evaluation tools, and easy-to-use pipelines for Hugging Face models.

Highlights:

• Full benchmark: questions, tables, train/val/test splits (llmsql-bench/llmsql-benchmark)
• Ready-to-use evaluation scripts (https://github.com/LLMSQL/llmsql-benchmark)
• Compatible with Hugging Face causal LMs
• Optimized for direct inference and evaluation without fine-tuning
• Optional fine-tuning datasets available on Hugging Face for research or domain adaptation

Documentation:

• Full guides and manuals are available online (Documentation)

Citation:
Please cite LLMSQL if you use it in your work:

@inproceedings{llmsql_bench,
  title={LLMSQL: Upgrading WikiSQL for the LLM Era of Text-to-SQL},
  author={Pihulski, Dzmitry and Charchut, Karol and Novogrodskaia, Viktoria and Koco{'n}, Jan},
  booktitle={2025 IEEE International Conference on Data Mining Workshops (ICDMW)},
  year={2025},
  organization={IEEE}
}

Follow us: