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

Datasetnvidia/Nemotron-SFT-CUDA-v1Top pick

NVIDIA SFT dataset for CUDA code generation

Synthetic dataset of 1k-10k tasks with tests from NVIDIA for training language models to write CUDA kernels.

  • Text generation
  • English
846 231K–10K rows
DatasetO1-OPEN/OpenO1-SFT

SFT data for activating the chain of thought

Dataset for language model SFT fine-tuning to generate coherent reasoning chains in English and Chinese.

  • Text generation
  • English
  • Chinese
836 39910K–100K rows
DatasetSakanaAI/JFinBench

JFinBench Japanese financial benchmark

An exam-based benchmark for evaluating LLMs in the Japanese financial domain.

  • Question answering
  • Visual question answering
  • Japanese
836 3up to 1K rows🔒 on request
DatasetKeepaceAI/MoveBench

Chinese sports and health benchmark

Set of 809 questions to test model knowledge in fitness, sports, and health.

  • Question answering
  • Text generation
  • Chinese
831 18up to 1K rows
Datasetbigcode/the-stack-v2-train-smol-ids

Cleaned code identifiers from The Stack v2

Subsample of deduplicated source code identifiers for training generative models.

  • Text generation
  • Code
830 6510M–100M rows🔒 on request
Datasetstorytracer/German-PD-NewspapersTop pick

German historical newspaper archive (13B words)

Massive corpus of OCR texts from public domain old German newspapers for training language models.

  • Text generation
  • German
820 81M–10M rows
DatasetMultilingual-Multimodal-NLP/McEval-Instruct

SFT instructions from the McEval benchmark

English instruction set for the McEval paper for SFT training and text generation evaluation.

  • Text generation
  • English
813 3910K–100K rows
DatasetSylvanL/Traditional-Chinese-Medicine-Dataset-SFT

Traditional Chinese medicine SFT dataset

Traditional Chinese medicine clinical cases, classic texts, and QA in Chinese.

  • Table question answering
  • Chinese
793 1091M–10M rows
DatasetComplexDataLab/BluePrintTop pick

Social media threads for LLM agents

A large-scale social media dialogue corpus for training and evaluating language agents with persona clustering.

  • Text generation
  • English
786 610M–100M rows🔒 on request
DatasetMoreThought/SmolPyEdu

SmolPyEdu clean Python script corpus

Large array of cleaned short Python scripts for language model pre-training.

  • Text generation
  • Question answering
  • English
782 41M–10M rows
DatasetHFforLegal/case-law

Standardized case law corpus

Multi-jurisdiction court decision collection in English and French for legal AI.

  • Question answering
  • Text generation
  • Table question answering
  • French
  • English
781 32100K–1M rows
Datasetr0b0tlab/qwen3.8-max-distillation-50k

49,772 Qwen3.8-Max distillation traces

About 49,772 traces from qwen3.8-max-preview for SFT and off-policy distillation.

  • Text generation
  • Question answering
  • English
778 12210K–100K rows
Datasetethanolivertroy/nist-cybersecurity-training

NIST cybersecurity document dataset

Large open text and QA pair collection based on NIST standards for cybersecurity LLM specialization.

  • Text generation
  • Question answering
  • English
775 61100K–1M rows
Datasetmarin-community/token-counts

Marin pretraining token statistics

Token count information for all datasets used in pretraining Marin models.

  • Text generation
771 3up to 1K rows
DatasetGOAL-Lab/FraudBenchTop pick

LLM fraud propensity benchmark (FraudBench)

159 legally vetted scenarios from real US courts to test models for deception and fraud propensity.

  • Question answering
  • Text generation
  • English
765 3up to 1K rows🔒 on request
DatasetJanosAudran/financial-reports-sec

SEC financial reports with sentiment

US companies 10-K annual reports, split into sections and sentences, with sentiment labels.

  • Fill mask
  • Text classification
  • English
764 7810M–100M rows
DatasetsimpleG2023/chinese-materials-science-open-intelligence

Chinese Materials Science Research Summaries and Tables

Structured reviews of Chinese research in materials science, semiconductors, and metallurgy with tables and QA pairs.

  • Text retrieval
  • Question answering
  • Table question answering
  • English
  • Chinese
763 3up to 1K rows🔒 on request
Datasetkatielink/healthsearchqa

HealthSearchQA medical questions dataset

A collection of consumer medical search queries released by Google.

  • Question answering
  • English
761 231K–10K rows
Datasetoyildirim/cyberstrike-sft-120kTop pick

Offensive cybersecurity (Red Team) instructions

A large corpus of 121 thousand instruction-response pairs on pentesting and attacks in English for SFT.

  • Text generation
  • Question answering
  • English
760 46100K–1M rows
Datasetsaidutta69/CyberSec-Reasoning-Premium

Cybersecurity reasoning for SFT

A curated dataset of 3,067 English cybersecurity reasoning chains for SFT training.

  • Text generation
  • Question answering
  • English
759 111K–10K rows