HAL-9001/spider-databases

Spider Databases (SQLite)

A re-host of the SQLite databases from the Spider text-to-SQL benchmark (Yu et al., 2018), packaged as spider_data.zip so it can be pinned by commit SHA and verified by checksum.

The standard xlangai/spider parquet ships only the question/SQL pairs — not the databases needed to execute queries. This repo fills that gap for reproducible execution-based evaluation.

• Contents: spider_data/database/<db_id>/<db_id>.sqlite, spider_data/tables.json, and the split files (dev.json, train_spider.json, gold SQL, etc.).
• Provenance: downloaded from the canonical Yale distribution (Google Drive id 1403EGqzIDoHMdQF4c9Bkyl7dZLZ5Wt6J).
• Integrity: spider_data.zip SHA256 = 00636695dabed6b5f4b8328a16b13e069a2f16591d5efcce57660669c85b121b.

License & attribution

Spider is released under CC-BY-SA-4.0; this re-host keeps the same license (share-alike). Please cite the original work:

@inproceedings{yu-etal-2018-spider,
  title  = {{Spider}: A Large-Scale Human-Labeled Dataset for Complex and Cross-Domain Semantic Parsing and Text-to-{SQL} Task},
  author = {Yu, Tao and Zhang, Rui and Yang, Kai and others},
  booktitle = {EMNLP},
  year   = {2018}
}

Used by text2sql-rlvr as its pinned execution substrate.