TPC-H data generated with DuckDB's TPC-H extension and exported as Parquet.
This dataset repository is organized by scale factor. Raw Parquet files are
grouped by scale under data/, and each TPC-H table is exposed as a separate
Hugging Face dataset config because the tables have different schemas.
lineitem = load_dataset("liangyc/tpch-sf-0_1", "sf-0_1_lineitem")
lineitem = lineitem["train"]… See the full description on the dataset page:
https://huggingface.co/datasets/liangyc/tpch-sf-0_1.