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

Datasetm-a-p/COIG-CQIA

High-quality Chinese instructions for SFT

A high-quality Chinese instruction dataset for SFT sourced from web Q&A and articles.

  • Question answering
  • Text classification
  • Text generation
  • Chinese
11.9K 79610K–100K rows
DatasetTsinghuaC3I/MedXpertQA

MedXpertQA expert medical questions

4,460 challenging medical questions spanning text and multimodal tasks with clinical images.

  • Table question answering
  • Question answering
  • Text generation
  • English
11.8K 681K–10K rows
DatasetBAAI/Infinity-Instruct

Infinity-Instruct: a large instruction dataset

Millions of English and Chinese instructions for LLM fine-tuning and evaluation.

  • Text generation
  • English
  • Chinese
11.7K 77110M–100M rows🔒 on request
Datasetopencompass/AIME2025

AIME 2025 math olympiad problems

Collection of English math problems from the AIME 2025 mathematical olympiad (rounds I and II) to evaluate LLM reasoning.

  • Question answering
  • English
11.7K 57up to 1K rows
DatasetOpen-Orca/OpenOrca

Augmented FLAN for instruction tuning

A collection of augmented FLAN data aligned with the Orca distribution for instruction tuning.

  • Text classification
  • Token classification
  • Table question answering
  • English
11.6K 1.6K1M–10M rows
DatasetQwen/DeepPlanning

DeepPlanning benchmark for agents

Multi-day travel and shopping planning tasks with verifiable constraints for evaluating LLM agents.

  • Text generation
  • English
  • Chinese
11.3K 2281K–10K rows
DatasetMedRAG/pubmed

PubMed corpus for medical RAG

English PubMed text snippets for medical Retrieval-Augmented Generation.

  • Question answering
  • English
11.2K 1171M–10M rows
Datasetnvidia/Nemotron-SFT-Math-v4Top pick

Math problems with step-by-step reasoning

A large English-language set of complex math problems with reasoning chains from DeepSeek-V4-Pro for SFT training.

  • Text generation
  • English
11.1K 58100K–1M rows
Datasetai-hyz/MemoryAgentBench

MemoryAgentBench for AI Agent Memory Evaluation

An English dataset (up to 1K examples) for testing agent long-term memory.

  • Question answering
  • Zero shot classification
  • Summarization
  • English
11.1K 44up to 1K rows
DatasetSimpleStories/SimpleStories

Over 2 million short English stories

A synthetic corpus of simple stories for training small and interpretable language models.

  • Text generation
  • English
11K 441M–10M rows
Datasetnvidia/Nemotron-Terminal-Corpus

SFT corpus for terminal agents

English SFT dataset from NVIDIA with trajectories for training LLMs to work in the terminal.

  • Question answering
  • English
11K 155100K–1M rows
Datasetnvidia/AceReason-1.1-SFT

NVIDIA math and code SFT dataset

4M SFT examples for math and code reasoning with answers generated by DeepSeek-R1.

  • Text generation
  • English
11K 1051M–10M rows
Datasetargilla/ultrafeedback-binarized-preferences-cleaned

UltraFeedback binarized preferences

Cleaned preference pairs from UltraFeedback for language model fine-tuning.

  • Text generation
  • English
10.7K 17610K–100K rows
DatasetCanCLID/zoengjyutgaai

Cantonese audiobooks and transcriptions

Recordings by famous storyteller Zhang Yuekai with verified texts for Cantonese TTS and ASR.

  • Automatic speech recognition
  • Text → speech
  • Text generation
  • Cantonese
10.5K 35100K–1M rows
DatasetDKYoon/SlimPajama-6B

SlimPajama 6 billion token sample

An English corpus of approximately 6 billion tokens for training and testing language models.

  • Text generation
  • English
10.4K 671M–10M rows
DatasetEricLu/SCP-116K

Scientific tasks with images SCP-VL

41.8k pairs of scientific tasks and solutions in English and Chinese, each with one or more images.

  • Text generation
  • Question answering
  • English
10.4K 125100K–1M rows
Datasetflytech/python-codes-25k

Python instructions for 25k tasks

Cleaned set of ~25k Python instructions with input/output for training and evaluating code models.

  • Text classification
  • Question answering
  • Token classification
10.3K 19310K–100K rows
Datasettokyotech-llm/swallow-math-v2

Math text corpus of 32 billion tokens

English math dataset for training and fine-tuning language models on problem solving.

  • Text generation
  • English
10.1K 3810M–100M rows
Datasetvtllms/sealqa

Complex benchmark for models with web search

Set of evaluation questions (under 1000 pieces) where standard web search yields contradictory or noisy answers.

  • Question answering
10.1K 34up to 1K rows
DatasetPeacefulData/RMS-AQA

Spatial audio and household scene questions

English benchmark for evaluating spatial sound understanding and multi-hop reasoning.

  • Question answering
  • English
10.1K 15🔒 on request