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

Datasetr0b0tlab/qwen3.8-max-glm5.2-kimi-k3-distillation

Multi-Teacher Distillation for SFT

57,937 filtered trajectories from three models for SFT in math, code, reasoning, and multilingual tasks.

  • Text generation
  • English
  • Chinese
  • Spanish
  • +3
6.3K 27910M–100M rows
DatasetIFM/SFT-Reasoning

SFT data for reasoning and instructions

Dataset for supervised fine-tuning in instruction following and logical reasoning.

  • Text generation
6.2K 2310M–100M rows
Datasetfacebook/flores

FLORES-200 machine translation benchmark

Parallel texts for evaluating translation from English into multiple languages, including low-resource ones.

  • Text generation
  • Translation
  • Acehnese
  • acm
  • acq
  • +9
6.2K 1351M–10M rows🔒 on request
Datasetgretelai/gretel-pii-masking-en-v1

Synthetic documents with PII annotations

Generated texts from various domains with PII and PHI entity annotations for training anonymization models.

  • Text classification
  • Text generation
  • English
6.2K 4710K–100K rows
Datasetmithulaartigala/reasoning-corpus-4K-5M-v1

Millions of reasoning chains from various LLMs

Filtered reasoning traces from DeepSeek, Qwen, Gemma, and other models for small model distillation.

  • Text generation
  • English
6.1K 2131M–10M rows
Datasetallenai/WildChat

650k Human-ChatGPT Dialogues

Multilingual real user dialogues with ChatGPT for chat model training and analysis.

  • Text generation
  • Question answering
6.1K 242100K–1M rows
Datasetklue/klue

Korean NLU benchmark KLUE

8 Korean NLU tasks for evaluating language models: from NER and NLI to MRC and DST.

  • Fill mask
  • Question answering
  • Text classification
  • Korean
6K 99100K–1M rows
Datasetstanford-crfm/air-bench-2024Top pick

Stanford AIR-Bench 2024 AI safety benchmark

A set of adversarial English prompts for testing LLMs for safety policy compliance.

  • Text generation
  • English
6K 2710K–100K rows
DatasetByteDance-Seed/EdgeBenchTop pick

Autonomous agent training benchmark

134 real-world tasks where agents learn in executable environments via multi-level feedback over 12+ hours.

  • Text generation
  • English
5.9K 86up to 1K rows
Datasetvblagoje/cc_news

English CC-News articles

708k English news articles from 2017–2019 sourced from Common Crawl.

  • Text generation
  • Fill mask
  • English
5.9K 70100K–1M rows
DatasetAletheiaResearch/GLM-5.2-Agent

Tool call logs from the GLM-5.2 agent

A small collection of JSONL traces of AI agent operation with saved tool schemas for SFT.

  • Text generation
5.9K 60up to 1K rows
Datasetproject-numina/NuminaMath-TIR

70k math problems with TIR

About 70k problems with reasoning traces and code execution for training TIR models.

  • Text generation
  • English
5.8K 16610K–100K rows
DatasetHelsinki-NLP/news_commentary

Parallel News Commentary Corpus

Texts from the OPUS News-Commentary collection in six languages for training and evaluating machine translation models.

  • Translation
  • Arabic
  • Czech
  • German
  • +9
5.8K 391M–10M rows
Datasetproj-persona/PersonaHubTop pick

PersonaHub: Personas for Synthetic Data

A collection of personas for generating diverse synthetic texts and instructions.

  • Text generation
  • Text classification
  • Token classification
  • English
  • Chinese
5.8K 834100K–1M rows
Datasetargilla/ifeval-like-data

Synthetic instructions for checking strict format adherence

550 thousand instruction-response pairs in the style of the IFEval benchmark from Argilla for training models to precisely follow constraints.

  • Text generation
  • English
5.8K 51100K–1M rows
Datasetnvidia/Nemotron-Personas-JapanTop pick

Synthetic personas for Japan

A large synthetic dataset of Japanese personas covering demographics, geography, and personality traits.

  • Text generation
  • Japanese
5.8K 1371M–10M rows
Datasetwdndev/webnovel-chinese

Chinese web novel corpus with 9 billion tokens

Cleaned texts of about 9,000 Chinese web novels, split into chapters in JSONL format for LLM training.

  • Text generation
  • Chinese
5.8K 52100K–1M rows
Datasetmath-ai/AutoMathTextTop pick

AutoMathText: 200 GB of mathematical texts

A corpus of English mathematical texts of about 200 GB for training and evaluating language models.

  • Text generation
  • Question answering
  • English
5.7K 1991M–10M rows
DatasetHelsinki-NLP/multiun

Parallel UN document corpus in 6 languages

A massive multilingual corpus of official UN document translations for training machine translation systems.

  • Translation
  • Arabic
  • German
  • English
  • +4
5.7K 15100M–1B rows
Datasetnvidia/Nemotron-Personas-USA

1M synthetic US personas

About 1M text descriptions of US personas generated for LLM training and evaluation.

  • Text generation
  • English
5.6K 3821M–10M rows
Datasetccdv/govreport-summarization

Long government report summarization

English long-report and short-summary pairs for training and evaluating summarization models.

  • Summarization
  • Text generation
  • English
5.5K 7010K–100K rows