SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2401.13822 · ICLR · 2024

Navigating Dataset Documentations in AI: A Large-Scale Analysis of Dataset Cards on Hugging Face

James Zou, Weixin Liang, Xinyu Yang

arXiv · PDF · Open in the Atlas

Code that ran

3 of 6 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 2 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2401.13822.json · badge: /api/ran/2401.13822.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
huggingface/datasets canonical (classifier label)
classifier basis: census
3 ran · 3 no recorded run
FunctionStatusWhere it lives
contains_wildcards Ran
ran, violated the contract we drafted
huggingface/datasets/src/datasets/data_files.py
code served (permissive licence) · get_code("c915bf78a27d08b3")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

# Setup code for contains_wildcards function

ARGS = {
    "pattern": "*test[abc]data"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_random_fingerprint Ran
ran; our contract draft was wrong, not the code
huggingface/datasets/src/datasets/fingerprint.py
code served (permissive licence) · get_code("fdde5c1fc2fe5556")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import random

fingerprint_rng = random.Random()

def generate_random_fingerprint(nbits: int=64) -> str:
    return f'{fingerprint_rng.getrandbits(nbits):0{nbits // 4}x}'

ARGS = {
    'nbits': 64
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
transmit_format Ran
ran; our contract draft was wrong, not the code
huggingface/datasets/src/datasets/arrow_dataset.py
code served (permissive licence) · get_code("91abb27e4b2dea99")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned function. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from functools import partial, wraps
from typing import Union

# Create a minimal Dataset class that matches what the function expects
class Dataset:
    def __init__(self, column_names, format_type='torch', format_kwargs=None, format_columns=None, output_all_columns=False):
        self.column_names = column_names
        self._format_type = format_type
        self._format_kwargs = format_kwargs or {}
        self._format_columns = format_columns
        self._output_all_columns = output_all_columns
        self._fingerprint = "test_fingerprint_123"
    
    def set_format(self, type=None, format_kwargs=None, columns=None, output_all_columns=False):
        self._format_type = type or self._format_type
        self._format_kwargs = format_kwargs or self._format_kwargs
        self._format_columns = columns
        self._output_all_columns = output_all_columns

class DatasetDict(dict):
    pass

# Create a test function that uses the decorator
def transmit_format(func):
    """Wrapper for dataset transforms that recreate a new Dataset to transmit the format of the original dataset to the new dataset"""

    @wraps(func)
    def wrapper(*args, **kwargs):
        if args:
            self = args[0]
            args = args[1:]
        else:
            self = kwargs.pop('self')
        unformatted_columns = set(self.column_names) - set(self._format_columns or [])
        self_format = {'type': self._format_type, 'format_kwargs': self._format_kwargs, 'columns': self._format_columns, 'output_all_columns': self._output_all_columns}
        out = func(self, *args, **kwargs)
        datasets = list(out.values()) if isinstance(out, dict) else [out]
        for dataset in datasets:
            new_format = self_format.copy()
            if new_format['columns'] is not None:
                new_format['columns'] = sorted(set(dataset.column_names) - unformatted_columns)
            out_format = {'type': dataset._format_type, 'format_kwargs': dataset._format_kwargs, 'columns': sorted(dataset._format_columns) if dataset._format_columns is not None else None, 'output_all_columns': dataset._output_all_columns}
            if out_format != new_format:
                fingerprint = dataset._fingerprint
                dataset.set_format(**new_format)
                dataset._fingerprint = fingerprint
        return out
    wrapper._decorator_name_ = 'transmit_format'
    return wrapper

# Create a simple test function to decorate
@transmit_format
def test_transform(dataset):
    """A test transform function that returns a new dataset"""
    new_dataset = Dataset(
        column_names=['col1', 'col2', 'col3'],
        format_type='torch',
        format_kwargs={},
        format_columns=['col1', 'col2'],
        output_all_columns=False
    )
    return new_dataset

# Create input dataset
input_dataset = Dataset(
    column_names=['col1', 'col2', 'col3'],
    format_type='torch',
    format_kwargs={},
    format_columns=['col1', 'col2'],
    output_all_columns=False
)

# Define ARGS mapping for transmit_format function
ARGS = {
    'func': test_transform.__wrapped__
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
delete_from_hub Not yet run
unverified (harvested, no recorded run)
huggingface/datasets/src/datasets/hub.py
code served (permissive licence) · get_code("2074bb03247a0985")
How it ran

no invocation recorded.

generate_fingerprint Not yet run
unverified (harvested, no recorded run)
huggingface/datasets/src/datasets/fingerprint.py
code served (permissive licence) · get_code("c7e8b6cb6ebf0f3d")
How it ran

no invocation recorded.

get_writer_batch_size_from_data_size Not yet run
unverified (harvested, no recorded run)
huggingface/datasets/src/datasets/arrow_writer.py
code served (permissive licence) · get_code("8e3fe703e41ff602")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 6 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Advances in machine learning are closely tied to the creation of datasets. While data documentation is widely recognized as essential to the reliability, reproducibility, and transparency of ML, we lack a systematic empirical understanding of current dataset documentation practices. To shed light on this question, here we take Hugging Face -one of the largest platforms for sharing and collaborating on ML models and datasets -as a prominent case study. By analyzing all 7,433 dataset documentation on Hugging Face, our investigation provides an overview of the Hugging Face dataset ecosystem and insights into dataset documentation practices, yielding 5 main findings: (1) The dataset card completion rate shows marked heterogeneity correlated with dataset popularity: While 86.0% of the top 100 downloaded dataset cards fill out all sections suggested by Hugging Face community, only 7.9% of dataset cards with no downloads complete all these sections. (2) A granular examination of each section within the dataset card reveals that the practitioners seem to prioritize Dataset Description and Dataset Structure sections, accounting for 36.2% and 33.6% of the total card length, respectively, for the most downloaded datasets. In contrast, the Considerations for Using the Data section receives the lowest proportion of content, accounting for just 2.1% of the text. (3) By analyzing the subsections within each section and utilizing topic modeling to identify key topics, we uncover what is discussed in each section, and underscore significant themes encompassing both technical and social impacts, as well as limitations within the Considerations for Using the Data section. (4) Our findings also highlight the need for improved accessibility and reproducibility of datasets in the Usage sections. (5) In addition, our human annotation evaluation emphasizes the pivotal role of comprehensive dataset content in shaping individuals' perceptions of a dataset card's overall quality. Overall, our study offers a unique perspective on analyzing dataset documentation through large-scale data science analysis and underlines the need for more thorough dataset documentation in machine learning research.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2401.13822")
get_code_for_paper("2401.13822")
have("2401.13822")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2401.13822.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2401.13822.svg)](https://syntology.ai/paper/2401.13822)

Connect an agent — have() is free.