The package ships a toy CuratedContainer so you can poke at the full API — schema, dtypes, splits, describe() — without touching the network. Identical interface to a downloaded BeyondArena container.

from data_foundry.curation_container import CuratedContainer
from data_foundry.examples import get_toy_container_path

container = CuratedContainer.load(get_toy_container_path())
print(container.describe())          # full identity + dtypes + task + splits
print(container.dataset.shape)       # the actual DataFrame
print(container.task_metadata.split_regime)  # "iid", "temporal_non_iid", or "grouped_non_iid"

Full inspection script (every metadata field printed): examples/load_curated_container.py.

Download a single BeyondArena container by name (or UUID) and iterate its outer CV splits. The collection resolves the container against your local cache; subsequent runs hit disk, not the network.

from data_foundry.collections import BEYOND_ARENA

container = BEYOND_ARENA.get_dataset("airfoil_self_noise")
df = container.dataset
target = container.task_metadata.target_column_name

for repeat_id, folds in container.experiment_metadata.splits.items():
    for fold_id, (train_idx, test_idx) in folds.items():
        X_train, y_train = df.iloc[train_idx].drop(columns=target), df.iloc[train_idx][target]
        X_test,  y_test  = df.iloc[test_idx].drop(columns=target),  df.iloc[test_idx][target]
        # ... fit, evaluate ...

Full worked example (Random Forest, RMSE per fold, full metadata via container.describe()): examples/benchmark_on_beyond_arena.py.

Split regimes. BeyondArena ships datasets from three regimes — which one a dataset is in shows up directly on task_metadata:

Side-by-side regime printout (one IID, two grouped variants — per_group vs per_sample — and one temporal): examples/data_foundry_data_regimes.py.

BEYOND_ARENA.prefetch(...) batches every container into a single Hugging Face snapshot_download call (one network round-trip for the whole collection). On a warm cache it skips importing huggingface_hub entirely.

from data_foundry.collections import BEYOND_ARENA

paths = BEYOND_ARENA.prefetch()          # warms the cache once
for container in BEYOND_ARENA.iter_containers():  # now hits disk only
    print(container.dataset_metadata.unique_name, container.dataset.shape)

Cache management:

BEYOND_ARENA.clear_cache()                 # nuke this collection's subdir
BEYOND_ARENA.get_dataset(name, force_download=True)  # re-fetch a single container

Full worked example with tqdm progress + checksum verification: examples/download_all_beyond_arena_datasets.py. For a single dataset round-trip with checksum verification, see examples/download_beyond_arena_dataset.py.

End-to-end pipeline, condensed (the full runnable version is examples/curate_a_dataset.py):

from data_foundry.schema import DatasetMetadata, PredictiveMLTaskMetadata

# --- Basic metadata
dataset_mold = DatasetMetadata(
    unique_name="blood_transfusion",
    dataset_year="2008",
    domain_str="medical & healthcare",
    dataset_source="UCI",
    original_dataset_source_download_link="https://doi.org/10.24432/C5GS39",
    download_description="""
We download the data from the UCI repository and unzip it to a predefined folder.

mkdir -p local-data-warehouse/blood_transfusion/ \\
  && wget -P local-data-warehouse/blood_transfusion/ \\
       https://archive.ics.uci.edu/static/public/176/blood+transfusion+service+center.zip \\
  && unzip local-data-warehouse/blood_transfusion/blood+transfusion+service+center.zip \\
       -d local-data-warehouse/blood_transfusion/
""",
    academic_reference_bibtex="""@article{yeh2009knowledge,
  title={Knowledge discovery on RFM model using Bernoulli sequence},
  author={Yeh, I-Cheng and Yang, King-Jang and Ting, Tao-Ming},
  journal={Expert Systems with applications},
  volume={36}, number={3}, pages={5866--5871},
  year={2009}, publisher={Elsevier},
}
""",
    academic_reference_bibtex_key="yeh2009knowledge",
    license="CC BY 4.0",
    data_tags=["IID"],
    curation_comments="Renamed features for clarity; mapped target 0/1 → No/Yes; ~29% duplicate rows kept.",
)
task_mold = PredictiveMLTaskMetadata(
    target_column_name="DonatedBloodInMarch2007",
    problem_type="binary_classification",
    objective_metric_name="roc_auc",
    stratify_on="DonatedBloodInMarch2007",
)

# --- Preprocessing
import pandas as pd
df = pd.read_csv(f"{dataset_mold.path}/transfusion.data")
df.columns = [
    "MonthsSinceLastDonation", "NumberOfDonations", "TotalBloodDonated",
    "MonthsSinceFirstDonation", "DonatedBloodInMarch2007",
]
df["DonatedBloodInMarch2007"] = df["DonatedBloodInMarch2007"].map({1: "Yes", 0: "No"})
df["DonatedBloodInMarch2007"] = df["DonatedBloodInMarch2007"].astype("category")
df = df.sample(frac=1, random_state=42).reset_index(drop=True)

# --- Sanity checks
from data_foundry import dataset_checks
df_head, summary, numeric_stats, cat_stats, target_df = dataset_checks.run_all_checks(
    data=df,
    target_feature=task_mold.target_column_name,
    problem_type=task_mold.problem_type,
)

# --- Outer CV splits
from data_foundry.curation_recommendations import (
    get_recommended_iid_splits,
    get_recommended_splits_dimensions,
)

n_repeats, n_splits, test_size = get_recommended_splits_dimensions(dataset=df)
splits = get_recommended_iid_splits(
    dataset=df,
    n_repeats=n_repeats,
    n_splits=n_splits,
    test_size=test_size,
    stratify_on=task_mold.stratify_on,
)

# --- Split metadata + container
from data_foundry.schema import PredictiveMLSplitsMetadata
from data_foundry.curation_container import CuratedContainer

splits_mold = PredictiveMLSplitsMetadata(
    splits_comment="Default splits for IID data.",
    splits=splits,
)
curated_data = CuratedContainer(
    dataset=df,
    dataset_metadata=dataset_mold,
    task_metadata=task_mold,
    experiment_metadata=splits_mold,
)
curated_data.save()
print(curated_data.uuid, curated_data.checksum)

For the contributor flow (where to put the notebook, how to open the PR, the /new-dataset Claude Code skill, best practices around versioning, anomaly tracking, and dtype handling), see CONTRIBUTING_DATASETS.md.

Before a dataset becomes a notebook, it lives in the curation log: one markdown record per candidate dataset under curation/records/. Each <unique_name>.md has YAML front-matter (the structured / dropdown fields) plus a free-text body (## Comments, ## Reference). Add or triage a dataset by creating/editing its file — by hand, with an agent, or through the dashboard.

pip install "data-foundry[curation]"   # or the editable dev install
data-foundry-curation serve            # → http://127.0.0.1:8765

The local, Sheets-like dashboard edits those records in place (filter by status, pin a working row, add dropdown options, …) and ships a built-in Guidelines tab describing the selection criteria and processing conventions. Other CLI subcommands:

data-foundry-curation validate                 # check records against the dropdown vocab
data-foundry-curation export --format xlsx out.xlsx   # flat snapshot (csv|parquet|xlsx|gsheet)
data-foundry-curation build-site site/          # read-only static copy (e.g. GitHub Pages)
data-foundry-curation import-sheet sheet.csv    # one-time migration from the old Google Sheet

Browse it online. A read-only copy of the backlog is published to GitHub Pages — tabarena.github.io/data-foundry — and regenerated automatically from curation/records/ on every push to main (no install, no network round-trip to Hugging Face; search, sort, filter, and pin all run in the browser). Note this makes every record's comments, reviewer names, and decision notes public.

Working with Claude Code? The /curate slash command starts the dashboard and loads the curation guidelines into context so the agent can help you decide and process datasets.

pip install data-foundry

git clone https://github.com/TabArena/data-foundry.git
cd data-foundry
uv pip install -e .

git clone https://github.com/TabArena/data-foundry.git
cd data-foundry
uv pip install -e ".[dev,tests]"
pytest                                 # run the test suite
ruff check . && ruff format --check .  # lint + format

The dev extra adds curation-time deps (openml, kaggle, seaborn, polars, etc.); tests adds pytest and scikit-learn (needed for the recommended-split helpers and examples).