Skip to content

Datasets

Hugging Face datasets with plain-language descriptions: what's inside and what models to train on them. 4,788 in the catalog so far, with new ones added every hour.

Data type

For which models

Find a dataset for your goal

Answered by Gemini 4 · the chat is kept only in this tab

Datasetgoogle/xtreme

XTREME multilingual benchmark

Multilingual pairs for NLI and other tasks: 112.5k annotated pairs across 14 languages.

  • Multiple choice
  • Question answering
  • Token classification
  • Afrikaans
  • Arabic
  • Bulgarian
  • +9
29.1K 1191M–10M rows
Datasetdronefreak/SeaDronesSee

Marine drone imagery for object detection

An unofficial YOLO-format build of the SeaDronesSee v2 dataset for water object detection.

  • Object detection
  • English
29.1K 310K–100K rows
DatasetSpawning/pd12m-full

12 million public domain image-text pairs

A ready-to-stream version of the PD12M corpus in webdataset format for training generative and multimodal models.

  • English
29.1K 2110M–100M rows
DatasetComplexDataLab/OpenFake

Deepfake and AI photo detection dataset and benchmark

Millions of real and generated image pairs on socio-political topics for training and testing AI content detectors.

  • Image classification
  • English
29.1K 431M–10M rows
Datasetwildintelproject/donadataset

Wildlife camera trap images in YOLO format

Nearly 34 thousand camera trap frames from Doñana National Park (Spain) annotated with 15 animal categories.

  • Object detection
29K 010K–100K rows
Datasetopenai/mrcrTop pick

Multi-needle long context benchmark

A dataset for testing LLMs on finding and distinguishing multiple needles in a long context.

    28.9K 2351K–10K rows
    DatasetADSKAILab/Zero-To-CAD-1mTop pick

    One million synthetic CAD sequences

    1M executable CAD programs synthesized by an agent without real data, for text-to-3D and image-to-3D.

    • Text → 3d
    • Image → 3d
    • English
    • Code
    28.8K 145100K–1M rows
    Datasetnyu-visionx/VSI-Bench

    Video visual-spatial VQA benchmark

    Video question-answering dataset for testing visual-spatial understanding in multimodal models.

    • Visual question answering
    • English
    28.8K 7810K–100K rows
    DatasetKimTaeHyun1/MAGO_share

    Multi-hour Korean speech corpus and annotations for TTS

    Large collection of Korean audio recordings (thousands of hours) with metadata, tags, and instructions for speech synthesis.

    • Text → speech
    • Korean
    28.8K 01M–10M rows🔒 on request
    Datasetaustin-starks/congressional-stock-trades

    US Congressional stock trades registry

    A verified tabular dataset of financial disclosure reports on stock trades by US senators and congressmen.

    • English
    28.7K 6100K–1M rows
    Datasetcoastalcph/lex_glue

    LexGLUE legal benchmark

    A collection of legal text tasks for evaluating language understanding models in English.

    • Question answering
    • Text classification
    • English
    28.7K 85100K–1M rows
    DatasetYunncheng/Mirage-Test

    AI-generated image detection test benchmark

    A collection of real and photorealistic AI-generated images across five domains for deepfake detector evaluation.

    • Image classification
    • English
    28.7K 310K–100K rows
    Datasetthuml/Time-Series-Library

    Time-series analysis and forecasting benchmark

    Time-series collection from the authors of Time-Series-Library for training and evaluating models across five core tasks.

    • Time series forecasting
    • English
    28.7K 141M–10M rows
    Datasettsivakar/polysemous-words

    Contextual usage examples of ambiguous English words

    A massive web text corpus for lexical sample word sense disambiguation (WSD) and contextual understanding evaluation in language models.

    • English
    28.7K 1100M–1B rows
    DatasetHuggingFaceM4/ChartQA

    ChartQA: image and text pairs

    ChartQA dataset of 10k-100k examples with image and text modalities; card details are sparse.

      28.5K 7010K–100K rows
      DatasetmypThu/SportsSlomo-CVS

      SportsSloMo-CVS dataset for deblurring

      A dataset for spatio-temporal motion deblurring featuring synchronized RGB frames and sensors.

      • Image → image
      28.4K 4up to 1K rows
      Datasetgoogle/WaxalNLPTop pick

      Waxal: multilingual African speech corpus

      Audio and text for ASR and TTS across dozens of African languages—a large open speech corpus from Google.

      • Automatic speech recognition
      • Text → speech
      • Acoli
      • Akan
      • Amharic
      • +9
      28.1K 3131M–10M rows
      Datasetm-a-p/PIN-14MTop pick

      PIN-14M Multimodal Document Dataset

      Large corpus of paired and alternating English and Chinese multimodal documents with quality labels.

      • English
      • Chinese
      28K 3810K–100K rows
      Datasetbenchmark-anon-2026/RoadmapBench

      Complex multi-step benchmark for coding agents

      115 real-world version upgrade tasks for open-source projects across 5 programming languages.

      • Text generation
      • English
      28K 1up to 1K rows