stanford-star/relbench-v1

---

license: cc-by-sa-4.0 tags:

• relbench
• relational-deep-learning configs:
• config_name: databases data_files:
  • split: eval path: STATS/databases.parquet
• config_name: tasks data_files:
  • split: eval path: STATS/tasks.parquet---

RelBench v1

The RelBench v1 databases and tasks, one database per subdirectory. The v2-only databases and tasks are in stanford-star/relbench-v2-extra. regression_stds.json holds the train-split target standard deviations that turn MAE into NMAE.

<dataset>/
  manifest.yaml                 # tables, primary keys, foreign-key graph, val/test timestamps
  schema.svg                    # ER diagram
  db/*.parquet                  # relational tables
  tasks/<task>/manifest.yaml    # task spec
  tasks/<task>/{train,val,test}.parquet

import relbench
ds = relbench.load_dataset("rel-f1")
task = ds.load_task("driver-position")

License

CC BY-SA 4.0 for the collection. Each database keeps the license of its source; see STATS/databases.parquet.

Citation

@inproceedings{relbench,
  title={RelBench: A Benchmark for Deep Learning on Relational Databases},
  author={Robinson, Joshua and Ranjan, Rishabh and Hu, Weihua and Huang, Kexin and Han, Jiaqi and Dobles, Alejandro and Fey, Matthias and Lenssen, Jan Eric and Yuan, Yiwen and Zhang, Zecheng and He, Xinwei and Leskovec, Jure},
  booktitle={Advances in Neural Information Processing Systems},
  year={2024}
}