Implementation:Huggingface Datasets Get Dataset Config Names

Knowledge Sources	Huggingface Datasets HF Datasets Docs
Domains	Data_Engineering, NLP
Last Updated	2026-02-14 18:00 GMT

Overview

Concrete tool for inspecting dataset configurations available on the Hub before loading, provided by the HuggingFace Datasets library.

Description

get_dataset_config_names is a top-level inspection function that returns the list of available configuration names for a given dataset. It resolves the dataset module via dataset_module_factory, retrieves the builder class, and reads the keys of builder_configs. If no explicit configurations are defined, it falls back to the default configuration name or the literal string "default".

Usage

Use get_dataset_config_names when you need to enumerate the available configurations of a dataset before loading. This is particularly useful for multi-configuration datasets like GLUE, SuperGLUE, or any dataset that bundles multiple tasks or subsets under one repository.

Code Reference

Source Location

• Repository: datasets
• File: src/datasets/inspect.py
• Lines: 109-172

Signature

def get_dataset_config_names(
    path: str,
    revision: Optional[Union[str, Version]] = None,
    download_config: Optional[DownloadConfig] = None,
    download_mode: Optional[Union[DownloadMode, str]] = None,
    data_files: Optional[Union[dict, list, str]] = None,
    **download_kwargs,
):

Import

from datasets import get_dataset_config_names

I/O Contract

Inputs

Name	Type	Required	Description
path	str	Yes	Path to the dataset repository. Can be a local path to the dataset directory or a dataset identifier on the Hugging Face Hub (e.g. 'nyu-mll/glue').
revision	Optional[Union[str, Version]]	No	Version of the dataset module to load. Defaults to the local library version, falling back to the main branch.
download_config	Optional[DownloadConfig]	No	Specific download configuration parameters (caching, proxies, etc.).
download_mode	Optional[Union[DownloadMode, str]]	No	Download/generate mode. Defaults to REUSE_DATASET_IF_EXISTS.
data_files	Optional[Union[dict, list, str]]	No	Defines the data files of the dataset configuration.
**download_kwargs	keyword arguments	No	Optional attributes for DownloadConfig which override attributes in download_config, for example token.

Outputs

Name	Type	Description
config_names	list[str]	A list of configuration name strings available for the dataset. For single-config datasets this is typically ["default"].

Usage Examples

Basic Usage

from datasets import get_dataset_config_names

# List all configurations for the GLUE benchmark
configs = get_dataset_config_names("nyu-mll/glue")
print(configs)
# ['cola', 'sst2', 'mrpc', 'qqp', 'stsb', 'mnli', 'mnli_mismatched',
#  'mnli_matched', 'qnli', 'rte', 'wnli', 'ax']

With Authentication

from datasets import get_dataset_config_names

# Access a private/gated dataset
configs = get_dataset_config_names(
    "my-org/private-dataset",
    token="hf_xxxxxxxxxxxxx",
)

Related Pages

Implements Principle

• Principle:Huggingface_Datasets_Dataset_Metadata_Inspection