DatasetCard

DatasetCard summarizes a dataset's task, scale, split, source, and license. Use it when introducing benchmarks or explaining experimental setup.

Basic Usage

<DatasetCard
  name="AcademicBench"
  description="Curated benchmark for efficient scientific model adaptation."
  task="Classification"
  samples="12k"
  features="128"
  split="70 / 15 / 15"
  source="Internal benchmark"
  license="Research use"
/>

With Notes

<DatasetCard name="RobustBench-Shift" task="Domain shift" samples="8.4k" split="5 folds">
All examples use the same preprocessing pipeline as the main benchmark.
</DatasetCard>

Props

Prop	Type	Default	Description
name	string	required	Dataset name
label	string	'Dataset'	Small label above the name
description	string	-	Short dataset description
task	string	-	Dataset task or modality
samples	string | number	-	Sample count or scale marker
features	string | number	-	Feature, modality, or field count
split	string	-	Train/validation/test split
source	string	-	Source benchmark or repository
license	string	-	License or access constraint
variant	string	'primary'	primary, success, warning, danger, or info
compact	boolean	false	Reduce spacing