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

Datasetieasybooks-org/prophet-mosque-libraryTop pick

Prophet's Mosque Library: 48k Arabic Books

Digitized Arabic Islamic books from the Prophet's Mosque Library for OCR and language models.

  • Image → text
  • Arabic
545.7K 2710K–100K rows
DatasetHuggingFaceM4/FineVisionMaxTop pick

Giant Collection of Images and Dialogues for VLMs

24.3 million interleaved image-text examples in English and Chinese for training multimodal models.

  • Image text → text
  • English
  • Chinese
291.9K 4810M–100M rows
DatasetHubLabs/HubEgo-100KTop pick

First-person real manual labor video dataset

Over 10,000 hours of real-world recordings of craftsmanship, cooking, and construction work with IMU sensor data.

  • Robotics
  • Video classification
  • Video text → text
  • English
257.2K 931B–10B rows🔒 on request
Datasetieasybooks-org/waqfeya-library

Digitized Islamic library in Arabic

Over 10 thousand Arabic books in PDF, TXT, and DOCX formats, recognized via Google Document AI.

  • Image → text
  • Arabic
210.6K 2310K–100K rows
Datasetmlfoundations/MINT-1T-HTMLTop pick

HTML portion of MINT-1T for multimodal pre-training

English MINT-1T HTML corpus: part of an open 1 trillion token and 3.4 billion image multimodal dataset.

  • Image → text
  • Text generation
  • English
209.3K 133100M–1B rows
Datasetieasybooks-org/shamela-waqfeya-library

Digitized Arabic Islamic book library

Corpus of over 4.6 thousand Arabic books (over 5 million pages) with extracted text.

  • Image → text
  • Arabic
170.4K 121K–10K rows
DatasetBenchCAD/BenchCAD

Multimodal CAD modeling dataset and benchmark

A three-configuration set featuring CadQuery 3D code, part renders, editing tasks, and CAD QA.

  • Image → text
  • Text generation
  • Question answering
126.5K 2310K–100K rows
Datasetpicbreeder-vlm/picbreeder-vlm-archive

VLM-agent image evolution archive

Generated images, genomes, and VLM agent reasoning logs from the Picbreeder study.

  • Image → text
  • English
125.1K 7100K–1M rows
DatasetInnovatorLab/Innovator-VL-Instruct-46MTop pick

Large-scale 46M-example SFT dataset for VLMs

A huge corpus of multimodal instructions (image + text) for full-parameter SFT of vision-language models.

  • Image text → text
  • English
103.7K 1810M–100M rows
DatasetMMMU/MMMU

MMMU multimodal benchmark for expert tasks

Benchmark for multimodal understanding and reasoning featuring questions with images across multiple disciplines.

  • Question answering
  • Visual question answering
  • Multiple choice
  • English
90.7K 34710K–100K rows
DatasetVLM2Vec/MVBench

Video Understanding Benchmark for Multimodal AI

An English set of 1k–10k examples for evaluating models on video question-answering tasks.

  • Visual question answering
  • Video classification
  • English
87K 01K–10K rows
Datasettumor-vqa/DeepTumorVQA_2.0Top pick

Medical Visual Question Answering Benchmark

3D CT abdominal benchmark for Visual Question Answering with 438K QA pairs and clinical subtypes.

  • Visual question answering
  • Image classification
  • English
85.1K 810K–100K rows
Datasetmlfoundations/dcvlm-baseline-200bTop pick

DCVLM-Baseline multimodal corpus with 200B tokens

Cleaned mixture of 103.9 million image-text pairs in 6 languages in WebDataset format for VLM pre-training.

  • Image text → text
  • Visual question answering
  • English
  • Chinese
  • German
  • +9
78.3K 910K–100K rows
Dataset1aurent/unsplash-lite

Unsplash photo descriptions and metadata

A lightweight version of the Unsplash dataset with texts and metadata for 25,000 English photographs.

  • Text → image
  • Image → text
  • English
77.7K 1210K–100K rows
DatasetFsoft-AIC/RobotDesign1MTop pick

Multimodal Robot Design Dataset

Millions of image-text pairs from research papers for robot design generation and analysis.

  • Image text → text
  • Text → image
72.7K 161M–10M rows
Datasetfpvlabs/stera-10mTop pick

Stera-10m: egocentric multimodal dataset

200 hours of synchronized first-person recordings with RGB, LiDAR depth, audio, and text for embodied AI and robotics.

  • Robotics
  • Video classification
  • Image → text
  • English
72.2K 881M–10M rows🔒 on request
Dataseteinrafh/hnm-fashion-recommendations-data

H&M transactions and product catalog for recommendations

Large array of anonymized purchases, customer profiles, and H&M clothing metadata for recommendation systems.

  • Tabular classification
  • Text classification
  • Image text → text
  • English
  • Indonesian
69.2K 510M–100M rows
Datasetmlfoundations/dcvlm-balanced-200bTop pick

Balanced multimodal DCVLM corpus of 200B tokens

Cleaned dataset of 112M image-text pairs in 6 languages in WebDataset format for pretraining multimodal models.

  • Image text → text
  • Visual question answering
  • English
  • Chinese
  • German
  • +9
65.2K 210K–100K rows
Datasetmvp-lab/LLaVA-OneVision-2-DataTop pick

LLaVA-OneVision-2 Multimodal and Video Data

Large collection of video clips, captions, and spatial reasoning tasks for training multimodal models.

  • Video text → text
  • Visual question answering
  • Image text → text
  • English
61.5K 51up to 1K rows
Datasetamolharsh/Ground3D_Dataset

3D scene question-answering with point grounding and metrics

Large English corpus of 3D scenes and dialogues based on ScanNet for training models in spatial and object size understanding.

  • Visual question answering
  • Question answering
  • English
60.5K 11M–10M rows
Datasetbshada/open-schematics

Electronic Schematics and PCB Layouts

Largest collection of schematics and PCBs: sources, images, metadata, and related board files.

  • Text generation
  • Image → text
  • Text → image
  • English
55.7K 20810K–100K rows
DatasetHoneyDataV2/Honey-Data-V2Top pick

Bee Corpus for End-to-End MLLM Training

Large English-Chinese corpus of images, texts, and documents for training multimodal models.

  • Visual question answering
  • Image text → text
  • English
  • Chinese
54.8K 8410M–100M rows