Skip to content

Datasets

Hugging Face datasets with plain-language descriptions: what's inside and what models to train on them. 4,797 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

DatasetOpen-Bee/Honey-Data-15M

Bee multimodal image-text corpus

15M English image-text pairs for training and evaluating open multimodal models.

  • Image text → text
  • English
15K 12110M–100M rows
DatasetQwen/Qwen-Image-BenchTop pick

Qwen Image Generation Benchmark

An English and Chinese image-text dataset for deep evaluation of Text-to-Image models.

  • Image → text
  • English
  • Chinese
15K 531K–10K rows
DatasetPKU-Alignment/PKU-SafeRLHF

Preferences and Q-A for LLM safety

Data for training and evaluating safe responses: preferences, question-answer pairs, and English prompts.

  • Text generation
  • English
14.9K 207100K–1M rows
DatasetSkywork/SkyPile-150B

Large Chinese web corpus for LLMs

Large filtered and deduplicated Chinese web text dataset for language model pre-training.

  • Text generation
  • Chinese
14.8K 4211M–10M rows
DatasetSandflow/sand3d-simready

Procedural 3D objects as Python code and OpenUSD

2,334 parametric 3D models and mechanisms with sand3d code, OpenUSD files, and renders for 3D generation.

  • Text → 3d
  • Image → 3d
14.7K 151K–10K rows
Datasetmadrylab/gsm8k-platinum

Corrected GSM8K school math test

Revised English test split of GSM8K for precise evaluation of LLM reasoning.

  • English
14.6K 481K–10K rows
Datasetnatolambert/GeneralThought-430K-filtered

Non-commercial filtered reasoning corpus

Commercial-friendly version of the GeneralThought-430K dataset without restricted prompts.

    14.6K 35100K–1M rows
    Datasetucinlp/drop

    DROP: Discrete reading benchmark

    96k paragraph-based questions to test arithmetic, counting, and sorting.

    • Question answering
    • English
    14.6K 7210K–100K rows
    DatasetCohereLabs/include-base-44

    Multilingual benchmark of examination tests

    22,637 four-choice questions across 44 languages to evaluate LLM knowledge and logic.

    • Multiple choice
    • Albanian
    • Arabic
    • Armenian
    • +9
    14.4K 5310K–100K rows
    Datasetbasis-ai/basis-conversations-1500Top pick

    Multilingual multi-speaker conversations (48 kHz)

    Multilingual dataset of live dialogues of up to 4 people with separate audio channels for ASR.

    • Automatic speech recognition
    • Audio classification
    • Arabic
    • German
    • English
    • +9
    14.3K 6310K–100K rows🔒 on request
    Datasetfancyzhx/amazon_polarity

    Amazon English Reviews for Sentiment Analysis

    A large corpus of text product reviews with ratings for text classification tasks.

    • Text classification
    • English
    14.3K 531M–10M rows
    DatasetSpatialVID/SpatialVIDTop pick

    Large-scale video dataset with spatial annotations

    1 to 10 million entries with spatial annotations for video and 3D generation tasks.

    • Text → video
    • Text → 3d
    • Image → 3d
    • English
    14.2K 491M–10M rows🔒 on request
    Datasetgarage-bAInd/Open-Platypus

    Open-Platypus for logical reasoning

    Filtered text dataset for improving LLM logical reasoning capabilities.

    • English
    14.2K 43210K–100K rows
    Datasetgoogle/svq

    Short Spoken Questions in 17 Languages

    Audio questions across 26 locales to evaluate ASR, audio search, and audio embeddings.

    • Question answering
    • Automatic speech recognition
    • Text retrieval
    • Arabic
    • Bangla
    • English
    • +9
    14.1K 681M–10M rows
    Datasetwalledai/AdvBench

    500 malicious instructions for LLM attacks

    A set of 500 malicious instructions in English to test language model robustness against adversarial attacks.

    • English
    14.1K 132up to 1K rows🔒 on request
    Datasetbigcode/the-stack-dedup

    Deduplicated source code corpus

    Code in many programming languages with permissive licenses after near-deduplication, about 1.5 TB.

    • Text generation
    • Code
    14.1K 431100M–1B rows🔒 on request
    Datasetallenai/tulu-3-sft-personas-instruction-following

    SFT instructions with personas and constraints

    29,980 synthetic examples for training precise adherence to instructions with constraints from IFEval.

    • Text generation
    • English
    14.1K 7110K–100K rows
    Datasetopen-r1/OpenThoughts-114k-math

    OpenThoughts mathematical reasoning

    A math-filtered version of OpenThoughts-114k with metadata and verifiable answers.

      14.1K 10310K–100K rows
      Datasetibm-esa-geospatial/TerraMeshTop pick

      TerraMesh Geospatial Satellite Dataset

      Over 9 million co-registered satellite imagery and geospatial data patches by IBM and ESA.

      • Image feature extraction
      14.1K 351M–10M rows
      DatasetShareGPT4Video/ShareGPT4Video

      ShareGPT4Video: video captions

      4.8M multimodal GPT-4-Vision video descriptions for LVLM and text-to-video models.

      • Visual question answering
      • Question answering
      • English
      14K 21610K–100K rows
      DatasetAgent-Ark/Toucan-1.5MTop pick

      MCP tool agent trajectories

      1.5 million synthetic trajectories working with 2,000+ tools from 495 real MCPs.

        14K 2501M–10M rows