The prediction results of ChatGPT on various datasets of Text-to-SQL.
For the specific evaluation methods and results, please refer to this paper "A comprehensive evaluation of ChatGPT's zero-shot Text-to-SQL capability" (https://arxiv.org/pdf/2303.13547.pdf).
python generate_txt.py
python evaluation.py --gold gold_example.txt --pred pred_example.txt --etype exec --db spider/database --table spider/tables.json
The result should be following: