BeyondArena Datasets

Datasets from BeyondArena, a unified, holistic benchmark for tabular data that supports diverse task types (IID, temporal, grouped), across sample size and feature dimensionality scales, with diverse feature types (with text, with high cardinality) from a broad range of disciplines.

We introduce BeyondArena and its datasets in Beyond IID: How General Are Tabular Foundation Models, Really?.

Click for BibTeX!

@misc{purucker2026iidgeneraltabularfoundation,
      title={Beyond IID: How General Are Tabular Foundation Models, Really?}, 
      author={Lennart Purucker and Andrej Tschalzev and Nick Erickson and Gioia Blayer and David Holzmüller and Alan Arazi and Alexander Pfefferle and Mustafa Tajjar and Gaël Varoquaux and Frank Hutter},
      year={2026},
      eprint={2606.30410},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2606.30410}, 
}

More details:

• Project page and leaderboard: http://tabarena.ai/
• Code / Benchmark repository: https://tabarena.ai/code

Quickstart

We recommend using the datasets via Data Foundry, which resolves a curated container (table + dtypes + task metadata + outer CV splits) by name and caches it locally:

pip install data-foundry

from data_foundry.collections import BEYOND_ARENA

container = BEYOND_ARENA.get_dataset("airfoil_self_noise")
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"

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 ...

To pre-download the entire collection in a single network round-trip:

from data_foundry.collections import BEYOND_ARENA

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)

See Data Foundry's examples for a full benchmarking walkthrough, the three split regimes (IID / temporal / grouped), and the curation flow.

Datasets

BeyondArena comes with 142 datasets. BeyondArena covers tabular classification and regression tasks. And the following types of datasets:

• IID tabular data
• Non-IID temporal tabular data
• Non-IID grouped tabular data
• IID and non-IID tabular data with text features
• Tabular data with high-cardinality categoricals

Dataset Selection Overview

We build on top of the dataset curation protocol of TabArena-v0.1 (https://arxiv.org/abs/2506.16791) and curate 142 tiny to large-sized, tabular IID and non-IID tasks. For details, see the paper.

Dataset Dashboard

We curated a diverse set of datasets. We share the dataset sizes (w.r.t. rows, columns, and cells), their age distribution, the distribution of feature types per dataset, and the share of datasets from a specific problem type, task type, dataset source, or application domain.

Per-Dataset Index

Per-dataset metadata for the BeyondArena benchmark, sorted by number of rows (N).

Click for expand all 142 Datasets!

Columns. N = rows · d = columns (before preprocessing) · C = classes (regression: —) · Prob. = problem type (Binary classification / Multiclass / Regression) · Task = task type (IID / Temporal / Grouped) · Age = years since publication at release time.

Domain abbreviations. M & H = Medical & Healthcare · B & M = Business & Marketing · B & L = Biology & Life Sciences · T & I = Technology & Internet · I & M = Industry & Manufacturing · C & M = Chemistry & Material Science · E & C = Environmental Science & Climate · P & A = Physics & Astronomy.

Each dataset has an academic_reference_bibtex_key in its dataset_metadata.dataset-mold-v1.json; the matching BibTeX entries are collected in dataset_references.bib. The BibKey(s) column below lists the keys to look up in that file (some datasets cite multiple sources).

Dataset	Domain	Source	Year	Age	N	d	C	Prob.	Task	BibKey(s)
hepatitis_survival_prediction	M & H	UCI	1981	45	155	19	2	Binary	IID	efron1981statistical
cirrhosis_patient_survival_prediction	M & H	UCI	1984	42	161	17	—	Reg	IID	dickson1989prognosis
clock_protein_toxicity	B & L	UCI	2021	5	171	1,117	2	Binary	IID	gul2021structure
pancreatic_cancer_mouse_detection	M & H	Other	2003	23	181	6,771	2	Binary	Grouped	hingorani2003preinvasive
lung_cancer_epithelial_genexp	M & H	GOV Website	2006	20	187	22,215	2	Binary	IID	spira2007airway
parkinsons_biomedical_voice_measurements	M & H	UCI	2007	19	195	23	2	Binary	Grouped	little2007exploiting
lung_cancer	M & H	Other	2001	25	197	12,600	4	Multi	IID	bhattacharjee2001classification
audiology_diagnosis	M & H	UCI	1987	39	199	68	3	Multi	IID	bareiss1990protos
heart_disease_va_long_beach	M & H	UCI	1989	37	200	13	2	Binary	IID	detrano1989international
forensic_glass_identification	C & M	UCI	1987	39	214	9	6	Multi	IID	German1987glass
early_stage_diabetes_risk_prediction	M & H	UCI	2019	7	251	16	2	Binary	IID	islam2019likelihood
body_density_prediction	M & H	Kaggle	1985	41	252	13	—	Reg	IID	penrose1985generalized
ljubljana_breast_cancer	M & H	UCI	1988	38	286	9	2	Binary	IID	Zwitter1988BreastCancer
heart_disease_hungary	M & H	UCI	1989	37	294	13	2	Binary	IID	detrano1989international
heart_failure_followup_survival	M & H	UCI	2020	6	299	12	2	Binary	IID	chicco2020machine
ljubljana_primary_tumor	M & H	UCI	1987	39	302	17	11	Multi	IID	Zwitter1987primarytumor
heart_disease_cleveland	M & H	UCI	1989	37	303	13	2	Binary	IID	detrano1989international
biomechanical_orthopaedic_prediction	M & H	UCI	2011	15	310	6	3	Multi	IID	Barreto2005Vertebral
gallstone_disease	M & H	UCI	2023	3	319	38	2	Binary	IID	esen2024early
prostate_cancer_detection	M & H	Other	2002	24	322	15,154	2	Binary	IID	petricoin2002serum
ecoli_proteins	B & L	UCI	1996	30	327	6	5	Multi	IID	horton1996probabilistic
horse_colic_survival	B & L	UCI	1989	37	344	20	3	Multi	IID	McLeish1989HorseColic
blood_tests_drink_prediction	M & H	UCI	1996	30	345	5	—	Reg	IID	UCILiverDisorders2016
eryhemato_squamous_disease	M & H	UCI	1997	29	366	34	6	Multi	IID	guvenir1998learning
dementia_prediction	M & H	Other	2010	16	370	8	3	Multi	Grouped	marcus2010open
south_africa_coronary_heart_disease	M & H	Kaggle	1983	43	462	9	2	Binary	IID	rossouw1983coronary
obesity_estimation	M & H	UCI	2019	7	498	14	—	Reg	IID	palechor2019dataset
telemonitoring_parkinsons_biomedical_voice_measurements	M & H	UCI	2007	19	502	19	—	Reg	Grouped	tsanas2009accurate
forest_fires	E & C	UCI	2008	18	517	12	—	Reg	IID	cortez2007data
qsar_aquatic_toxicity	B & L	UCI	2014	12	546	8	—	Reg	IID	cassotti2014prediction
micro_mass	B & L	UCI	2013	13	571	1,082	20	Multi	Grouped	mahe2014automatic
indian_liver_patient_dataset	M & H	UCI	2012	14	583	10	2	Binary	IID	ramana2012critical
drug_induced_autoimmunity_prediction	M & H	UCI	2025	1	597	177	2	Binary	IID	huang2025interdia
hepatitis_c_prediction	M & H	UCI	2018	8	608	12	4	Multi	IID	hoffmann2018using
biogeographical_ancestry_prediction	B & L	GitHub	2025	1	635	104	10	Multi	IID	heinzel2025advancing, ruiz2023development, xavier2020development
student_portuguese_performance	Education	UCI	2008	18	649	30	—	Reg	IID	silva2008using
credit_approval	Finance	UCI	1987	39	690	15	2	Binary	IID	quinlan1987simplifying
blood_transfusion	M & H	UCI	2008	18	748	4	2	Binary	IID	yeh2009knowledge
regensburg_pediatric_appendicitis	M & H	Other	2021	5	763	51	2	Binary	IID	marcinkevivcs2024interpretable
mutual_funds_india	Finance	Kaggle	2023	3	793	12	—	Reg	IID	Barnawal2022MutualFundsIndiaDetailed
qsar_fish_toxicity	B & L	UCI	2015	11	908	6	—	Reg	IID	cassotti2015similarity
tour_travels_churn	B & M	Kaggle	2021	5	954	6	2	Binary	IID	Tejashvi2023TourTravelsCustomerChurnPrediction
credit_g	Finance	UCI	1994	32	1,000	20	2	Binary	IID	hofmann1994statlog
maternal_health_risk	M & H	UCI	2020	6	1,014	6	3	Multi	IID	ahmed2020review
concrete_compressive_strength	C & M	UCI	1998	28	1,030	8	—	Reg	IID	yeh1998modeling
qsar_biodeg	B & L	UCI	2013	13	1,054	41	2	Binary	IID	mansouri2013quantitative
mice_protein_trisomy_discriminant	B & L	UCI	2015	11	1,080	76	8	Multi	Grouped	higuera2015self
garments_worker_productivity	I & M	UCI	2020	6	1,197	15	—	Reg	Temporal	imran2021mining
asp_potassco_classification	T & I	ASlib	2014	12	1,212	136	11	Multi	Grouped	hoos2014claspfolio, bischl_aslib_2016
wine_world_cost	B & M	Kaggle	2023	3	1,279	14	—	Reg	IID	Rustamov2023WineDataset
healthcare_insurance_expenses	M & H	Kaggle	2023	3	1,338	6	—	Reg	IID	arunjangir2452023insurance
website_phishing	T & I	UCI	2014	12	1,353	9	3	Multi	IID	abdelhamid2014phishing
fitness_club	B & M	Kaggle	2023	3	1,500	6	2	Binary	IID	ddosad2023fitness
airfoil_self_noise	P & A	UCI	2014	12	1,503	5	—	Reg	IID	brooks1989airfoil
fiat_500	T & I	Kaggle	2020	6	1,538	7	—	Reg	IID	paolocons2020fiat
mic	M & H	UCI	2020	6	1,699	111	8	Multi	IID	golovenkin2020trajectories
bad_customer_detection	B & M	Kaggle	2020	6	1,723	13	2	Binary	IID	Podsyp2020IsThisAGoodCustomer
cardiotocography	M & H	UCI	2010	16	2,126	22	3	Multi	Grouped	campos2010cardiotocography
marketing_campaign	B & M	Kaggle	2020	6	2,240	25	2	Binary	IID	saldanha2020marketing
coffee_rating_prediction	B & M	Kaggle	2023	3	2,369	12	—	Reg	Temporal	AlIrsyad2023CoffeeDataCoffeeReview
hazelnut_spread_contaminant_detection	B & L	OpenML	2020	6	2,400	30	2	Binary	IID	ricci2021machine
seismic_bumps	E & C	UCI	2013	13	2,584	15	2	Binary	IID	sikora2010application
iranian_churn	B & M	UCI	2011	15	2,850	13	2	Binary	IID	keramati2011churn
sat11_hand_algo_runtime	T & I	ASlib	2011	15	2,960	169	—	Reg	Grouped	xu-sat12a, sat12, bischl_aslib_2016
splice	B & L	UCI	1991	35	3,190	60	3	Multi	IID	towell1994knowledge
thyroid_discordant	M & H	UCI	1986	40	3,711	26	2	Binary	IID	quinlan1987simplifying
bioresponse	B & L	Kaggle	2012	14	3,751	1,776	2	Binary	IID	bioresponse2012hamner
hiva_agnostic	C & M	Other	2007	19	3,845	1,518	2	Binary	IID	guyon2007agnostic
mercedes_benz_greener_manufacturing	I & M	Kaggle	2017	9	4,204	371	—	Reg	Temporal	Novy2017MercedesBenzGreenerManufacturing
predict_students_dropout_and_academic_success	Education	UCI	2021	5	4,424	36	3	Multi	IID	martins2021early
santander_transaction_value	Finance	Kaggle	2018	8	4,447	540	—	Reg	IID	McDonald2018SantanderValuePredictionChallenge
churn	T & I	OpenML	2005	21	5,000	19	2	Binary	IID	marcoulides2005churn
homeq_default_prediction	B & M	Other	2016	10	5,708	12	2	Binary	IID	baesens2016credit
qsar_tid_11	C & M	OpenML	2015	11	5,741	1,024	—	Reg	IID	olier2018meta
polish_companies_bankruptcy	Finance	UCI	2010	16	5,790	64	2	Binary	IID	zikeba2016ensemble
wine_quality	C & M	UCI	2009	17	6,497	12	—	Reg	IID	cortez2009modeling
musk	C & M	UCI	1994	32	6,598	166	2	Binary	Grouped	dietterich1993comparison
taiwanese_bankruptcy_prediction	Finance	UCI	2009	17	6,819	92	2	Binary	IID	liang2016financial
naticusdroid_android_permissions_dataset	T & I	UCI	2021	5	7,491	85	2	Binary	IID	mathur2021naticusdroid
coil_2000	B & M	UCI	2000	26	9,822	85	2	Binary	IID	van2000coil
bank_customer_churn	B & M	Kaggle	2020	6	10,000	10	2	Binary	IID	Topre2022BankCustomerChurn
immoscout_german_house_prices	B & M	Kaggle	2019	7	10,317	23	—	Reg	IID	Shritech2019GermanHousingPricePrediction, OpenML43342Dataset
heloc	Finance	Kaggle	2021	5	10,459	23	2	Binary	IID	averkiyoliabev2021heloc
jm1	T & I	OpenML	2004	22	10,885	21	2	Binary	IID	menzies2004good
ghanas_indigenous_intel	E & C	Zindi	2025	1	10,928	10	4	Multi	Temporal	zindi_ghana_indigenous_intel_2025
ecommerce_shipping	B & M	Kaggle	2021	5	10,999	10	2	Binary	IID	gopalani2021ecommerce
video_game_fps_prediction	T & I	OpenML	2020	6	12,288	38	—	Reg	Grouped	peeters2021performance
online_shoppers_purchasing_intention_dataset	B & M	UCI	2017	9	12,330	17	2	Binary	IID	sakar2019real
in_vehicle_coupon_recommendation	B & M	UCI	2017	9	12,684	24	2	Binary	IID	wang2017bayesian
miami_housing	Finance	Kaggle	2016	10	13,776	15	—	Reg	IID	bourassa2021big
emscad	B & M	Other	2014	12	17,460	17	2	Binary	IID	vidros2017automatic
early_learning_predictors	Education	Other	2023	3	18,874	743	—	Reg	Grouped	DataDrive2030_2024_elom_thrivebyfive
hr_analytics	B & M	Kaggle	2021	5	19,158	12	2	Binary	IID	arashnic2021hr
houses	B & M	Other	1990	36	19,675	8	—	Reg	IID	pace1997sparse
superconductivity	P & A	UCI	2018	8	21,263	81	—	Reg	IID	hamidieh2018data
sberbank_housing_market_forecasting	B & M	Kaggle	2017	9	27,195	386	—	Reg	Temporal	Herman2024HomeCreditCreditRiskModelStability
credit_card_clients_default	Finance	UCI	2009	17	30,000	23	2	Binary	IID	yeh2009comparisons
amazon_employee_access	B & M	Kaggle	2010	16	32,769	9	2	Binary	IID	hamner2013amazon
california_house_prices_2020	B & M	Kaggle	2021	5	41,528	41	—	Reg	Temporal	d2lcourse2021california_house_prices
bank_marketing	Finance	UCI	2012	14	45,211	13	2	Binary	IID	moro2014bank-marketing
food_delivery_time	B & M	Kaggle	2023	3	45,451	9	—	Reg	IID	rajatkumar302023food
physiochemical_protein	C & M	UCI	2013	13	45,730	9	—	Reg	IID	rana2013protein
anes_voting_2026	Social Science	Other	2026	0	48,587	318	2	Binary	Temporal	anes2026timeseries
kdd_cup_09_appetency	B & M	Other	2008	18	50,000	212	2	Binary	IID	guyon2009analysis
diamonds	B & M	Other	2015	11	53,940	9	—	Reg	IID	wickham2016data
otto_group_product_classification_challenge	B & M	Kaggle	2015	11	61,878	93	9	Multi	IID	Bossan2015OttoGroupProductClassificationChallenge
labour_inspection_compliance	I & M	Other	2019	7	63,634	376	2	Binary	IID	flogard2022dataset
video_transcoding_time_prediction	T & I	UCI	2015	11	68,784	18	—	Reg	Grouped	deneke2014video
santander_customer_satisfaction	B & M	Kaggle	2016	10	71,080	307	2	Binary	IID	Jimenez2016SantanderCustomerSatisfaction
diabetes_130_us	M & H	UCI	2014	12	71,518	44	2	Binary	IID	strack2014impact
kick	B & M	Kaggle	2011	15	72,983	32	2	Binary	Temporal	DontGetKicked
aps_failure	I & M	UCI	2016	10	76,000	170	2	Binary	IID	ida2016challenge
sdss_17	P & A	Kaggle	2022	4	78,053	11	3	Multi	IID	accetta2022seventeenth
hotel_booking_demand	B & M	Other	2019	7	81,418	31	2	Binary	Temporal	antonio2019hotel
5g_energy_consumption	T & I	HuggingFace	2023	3	92,629	20	—	Reg	Grouped	huawei_netop_5g_energy_consumption
sepsis_survival_minimal_clinical_records	M & H	UCI	2020	6	110,204	3	2	Binary	IID	chicco2020survival
sf_permit_time	B & M	GOV Website	2025	1	116,954	37	—	Reg	Temporal	SanFrancisco2026BuildingPermits
wids_diabetes_mellitus	M & H	Kaggle	2021	5	127,358	181	2	Binary	IID	Matthys2021WiDSDatathon2021
customer_satisfaction_in_airline	B & L	Kaggle	2023	3	129,880	21	2	Binary	IID	yakhyojon2023airlinesatisfaction
pva_revenue_prediction_kddcup98	B & M	Other	1997	29	144,095	477	2	Binary	IID	Parsa1998KDDCup1998
give_me_some_credit	Finance	Kaggle	2011	15	150,000	10	2	Binary	IID	cukierski2011credit
acquire_valued_shoppers_challenge	B & M	Kaggle	2014	12	160,057	111	2	Binary	Temporal	DMDave2014AcquireValuedShoppersChallenge
kickstarter	B & M	Other	2025	1	187,118	15	2	Binary	Temporal	webrobots2026kickstarter
allstate_claims_severity	Insurance	Kaggle	2016	10	188,317	130	—	Reg	IID	Ferguson2016AllstateClaimsSeverity
santander_customer_transaction_prediction	Finance	Kaggle	2019	7	200,000	600	2	Binary	IID	Piedra2019SantanderCustomerTransactionPrediction
homesite_quote_conversion	Insurance	Kaggle	2015	11	260,753	295	2	Binary	IID	Darrel2015HomesiteQuoteConversion
home_credit_default_risk	Finance	Kaggle	2018	8	307,507	504	2	Binary	IID	Montoya2018HomeCreditDefaultRisk
covertype	E & C	UCI	1998	28	512,625	13	3	Multi	Grouped	blackard1999comparative
ieee_fraud_detection	Finance	Kaggle	2019	7	590,540	435	2	Binary	Temporal	ieee-fraud-detection
porto_seguro	Insurance	Kaggle	2017	9	595,206	37	2	Binary	IID	Howard2017PortoSegurosSafeDriverPrediction
rossmann_store_sales	B & M	Kaggle	2015	11	844,392	15	—	Reg	Temporal	kaggle_rossmann_store_sales
lending_club_1m	Finance	Kaggle	2018	8	1,064,751	96	2	Binary	Temporal	sanz2025credit
home_credit_default_stability_1m	Finance	Kaggle	2024	2	1,224,927	711	2	Binary	Temporal	Herman2024HomeCreditCreditRiskModelStability
consumer_complaints_1m	Finance	GOV Website	2025	1	1,226,140	12	3	Multi	Temporal	cfpb2025ConsumerComplaintDatabase
sepsis_prediction_1m	M & H	Other	2019	7	1,228,686	42	2	Binary	Grouped	reyna2020early
amex_non_iid_1m	Finance	Kaggle	2022	4	1,249,605	189	2	Binary	Grouped	howard2022amex
delivery_eta_1m	I & M	Kaggle	2024	2	1,250,000	225	—	Reg	Temporal	rubachev2025tabred
cooking_time_1m	I & M	Kaggle	2024	2	1,250,000	196	—	Reg	Temporal	rubachev2025tabred
climate_model_weather_forecasting_1m	E & C	Kaggle	2024	2	1,250,000	100	—	Reg	Temporal	rubachev2025tabred
maps_router_eta_1m	I & M	Kaggle	2024	2	1,250,000	988	—	Reg	Temporal	rubachev2025tabred
mercari_price_suggestion_1m	B & M	Kaggle	2018	8	1,250,000	6	—	Reg	IID	Howard2017MercariPriceSuggestionChallenge
electric_motor_temperature_prediction	I & M	Kaggle	2021	5	1,296,316	109	—	Reg	Grouped	kirchgassner2020estimating

Dataset Structure

The release ships as a flat bundle of 142 datasets. Each dataset lives in its own top-level directory named by unique_name, with a UUID-named version subdirectory holding all artifacts. Two layout variants exist:

<dataset_name>/<uuid>/...                 # default (132 datasets)
<dataset_name>/versions/<uuid>/...        # versioned wrapper (10 large non-IID datasets)

Each directory contains six core files, plus an optional tabarena_text_cache.parquet for the 16 datasets with text features:

<uuid>/
├── dataset.parquet                                       # the table (rows × columns)
├── dtypes.json                                           # column name → pandas dtype
├── container_metadata.json                               # uuid + sha256 checksum
├── dataset_metadata.dataset-mold-v1.json                 # provenance & curation notes
├── task_metadata.predictive-ml-task-mold-v1.json         # target, problem type, metric, split keys
├── experiment_metadata.predictive-ml-splits-mold-v1.json # CV fold indices
└── tabarena_text_cache.parquet                           # (optional) precomputed sentence embeddings keyed by text

For details on files and the metadata structure, checkout DataFoundry!

Text embedding cache (tabarena_text_cache.parquet)

Shipped for the 16 text-bearing datasets listed below. The file is a pandas DataFrame written via SemanticTextFeatureGenerator.save_embedding_cache (see TabArena's text_feature_generators.py):

• Index — a string column named text containing every unique text value observed across all text columns of dataset.parquet.
• Columns — 0, 1, …, D-1, holding the precomputed sentence embedding for each text value (default model: 32-dim embeddings).

Reload with:

import pandas as pd
df = pd.read_parquet("<dataset_name>/<uuid>/tabarena_text_cache.parquet")
cache = dict(zip(df.index, df.to_numpy()))   # {text: np.ndarray}

This lets you skip the embedding step at fit time. Datasets with a tabarena_text_cache.parquet: coffee_rating_prediction, consumer_complaints (1m variant), california_house_prices_2020, drug_induced_autoimmunity_prediction, emscad, immoscout_german_house_prices, kickstarter, labour_inspection_compliance, lending_club (1m variant), mercari_price_suggestion (1m variant), mutual_funds_india, pva_revenue_prediction_kddcup98, regensburg_pediatric_appendicitis, sf_permit_time, wids_diabetes_mellitus, wine_world_cost.

Loading a single dataset directly

Each per-dataset config in this card's frontmatter routes only dataset.parquet, which is enough to get the table but not the sibling metadata files (dtypes.json, task_metadata.*, experiment_metadata.* with the CV folds, dataset_metadata.*, container_metadata.json). Because the benchmark protocol depends on those files, the recommended path is to download the whole dataset folder with huggingface_hub:

from huggingface_hub import snapshot_download

local_dir = snapshot_download(
    repo_id="TabArena/BeyondArena",
    repo_type="dataset",
    allow_patterns=["churn/**"],                # one or more <dataset_name>/** globs
)
# local_dir/<dataset_name>/<uuid>/ now contains all six files for that dataset.

For the 10 datasets that use the versions/ wrapper (see Dataset Structure), the layout is <dataset_name>/versions/<uuid>/... — the <dataset_name>/** glob already covers both layouts.

If you only need the table (no folds, no metadata), the datasets library shortcut works:

from datasets import load_dataset

ds = load_dataset("<org>/BeyondArena", name="churn")   # any per-dataset config_name

Downloading the full bundle

from huggingface_hub import snapshot_download

local_dir = snapshot_download(
    repo_id="<org>/BeyondArena",
    repo_type="dataset",
)

Licensing

This collection is released under the terms in LICENSE (copyright-at-original-authors). Individual datasets retain their original licenses; see each dataset metadata for their source-specific terms.

Citation

If you use BeyondArena, please cite:

Beyond IID: How General Are Tabular Foundation Models, Really? Lennart Purucker, Andrej Tschalzev, Nick Erickson, Gioia Blayer, David Holzmüller, Alan Arazi, Alexander Pfefferle, Mustafa Tajjar, Gaël Varoquaux, Frank Hutter arXiv:2606.30410

📄 arXiv

BibTeX:

@misc{purucker2026iidgeneraltabularfoundation,
      title={Beyond IID: How General Are Tabular Foundation Models, Really?}, 
      author={Lennart Purucker and Andrej Tschalzev and Nick Erickson and Gioia Blayer and David Holzmüller and Alan Arazi and Alexander Pfefferle and Mustafa Tajjar and Gaël Varoquaux and Frank Hutter},
      year={2026},
      eprint={2606.30410},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2606.30410}, 
}

Per-Dataset References

If you use individual datasets, please also cite their original authors. BibTeX for every dataset in the benchmark is shipped alongside this card in dataset_references.bib (one entry per unique academic_reference_bibtex_key referenced by the dataset metadata files).

Changelog

• [27th May 2026] — Initial release: 142 curated IID and non-IID tasks.