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

DatasetBiogod/PMC-ClinicalLogsTop pick

Clinical logs and images from 150k PubMed articles

Chronological clinical events and 480 thousand medical images from 150 thousand PubMed Central cases.

  • Text generation
  • Question answering
  • English
32.3K 0100K–1M rows🔒 on request
Datasetuzaymacar/math-rollouts

Math reasoning rollouts

Step-by-step reasoning logs of DeepSeek R1-Distill models solving MATH dataset problems.

  • Question answering
  • Text generation
  • English
32K 1110K–100K rows
Datasetbigcode/the-stack-smol

Small code slice across 30 languages

A subset of The Stack: 10K random code examples for each of 30 languages, 2.6 GB.

  • Text generation
  • Code
31.5K 107100K–1M rows🔒 on request
DatasetRawvY/stack-prompts

Code training prompt corpus

High-quality data selection for training small programming language models.

  • Text generation
  • English
31.4K 010M–100M rows
Datasetderek-thomas/ScienceQA

ScienceQA: Scientific questions with images

A multimodal English science question dataset with multiple-choice options and images.

  • Multiple choice
  • Question answering
  • Other
  • English
31.3K 24410K–100K rows
Datasetgoogle/frames-benchmark

FRAMES benchmark for RAG systems

824 complex multi-hop questions across 2–15 Wikipedia articles to test RAG factuality, retrieval, and reasoning.

  • Text classification
  • Token classification
  • Table question answering
  • English
31.1K 276up to 1K rows
Datasetnvidia/Nemotron-CC-Code-v1Top pick

NVIDIA Common Crawl code corpus for pre-training

Updated code text dataset with 100 million to 1 billion examples for pre-training language models.

  • Text generation
31K 38100M–1B rows🔒 on request
Datasettasksource/mmlu

Lightweight duplicate-free MMLU benchmark

Cleaned and fast version of the well-known MMLU benchmark in English for evaluating language model knowledge.

  • Text classification
  • Multiple choice
  • Question answering
  • English
30.8K 3710K–100K rows
DatasetVyber07/cyber-security

Cybersecurity knowledge base

A cybersecurity corpus for training and question-answering tasks in English.

  • Text generation
  • Question answering
  • English
30.3K 1.4Kup to 1K rows🔒 on request
Datasetnvidia/OpenCodeReasoningTop pick

OpenCodeReasoning: synthetic code reasoning

735K synthetic reasoning examples across 28K competitive programming tasks.

  • Text generation
30.1K 577100K–1M rows
DatasetLego-X/Lego-RL-2699

2699 programming tasks for training RL agents

Difficulty-filtered executable software development tasks for training and evaluating agentic models via RL.

  • Text generation
  • English
30K 21K–10K rows
Datasetcais/wmdp

LLM hazardous knowledge & unlearning benchmark

English multiple-choice tests for evaluating models on dangerous domain knowledge and unlearning methods.

  • Text generation
  • English
29.8K 321K–10K rows
Datasetjuletxara/mgsm

MGSM multilingual grade-school math benchmark

250 math problems from GSM8K translated by humans into 10 languages, including Russian, to evaluate reasoning.

  • Text generation
  • English
  • Spanish
  • French
  • +7
29.7K 451K–10K rows
DatasetSilversorrow/SpineBench

Medical VQA benchmark for spine diseases

64,878 English QA pairs for 40,263 spine images to evaluate multimodal models.

  • Visual question answering
  • Multiple choice
  • English
29.2K 010K–100K rows
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
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
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
Datasetopen-r1/Mixture-of-ThoughtsTop pick

350k verified DeepSeek-R1 reasoning traces

350k verified reasoning chains in math, code, and science for training step-by-step reasoning.

  • Text generation
  • English
27.9K 351100K–1M rows
DatasetHelsinki-NLP/opus-100

English-centric OPUS-100 parallel corpus

Parallel texts in 100 languages with English always serving as one side of the pair.

  • Translation
  • Afrikaans
  • Amharic
  • Aragonese
  • +9
27.7K 26010M–100M rows