datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
BlueMO
BlueMO
BlueMO: A High-Quality Mathematical Olympiad Data Resources from Little Blue Book Series
BlueMO is a comprehensive and challenging dataset comprising mathematical olympiad problems paired with detailed solutions, meticulously curated from the esteemed "Little Blue Book" (小蓝书) series (Second Edition)—a vital resource for Chinese students training for national and international olympiad math competitions.
Designed to advance and assess sophisticated reasoning in LLMs… See the full description on the dataset page: https://huggingface.co/datasets/Luobots/BlueMO.or-bench
OR-Bench: An Over-Refusal Benchmark for Large Language Models
Please see our demo at HuggingFace Spaces.
Overall Plots of Model Performances
Below is the overall model performance. X axis shows the rejection rate on OR-Bench-Hard-1K and Y axis shows the rejection rate on OR-Bench-Toxic. The best aligned model should be on the top left corner of the plot where the model rejects the most number of toxic prompts and least number of safe prompts. We also plot a blue line… See the full description on the dataset page: https://huggingface.co/datasets/bench-llm/or-bench.LLaVA-OneVision-Data-ru
LLaVA-OneVision-Data-ru
Translated lmms-lab/LLaVA-OneVision-Data dataset into Russian language using Google translate.
Almost all datasets have been translated, except for the following:
["tallyqa(cauldron,llava_format)", "clevr(cauldron,llava_format)", "VisualWebInstruct(filtered)", "figureqa(cauldron,llava_format)", "magpie_pro(l3_80b_mt)", "magpie_pro(qwen2_72b_st)", "rendered_text(cauldron)", "ureader_ie"]
Usage
import datasets
data =… See the full description on the dataset page: https://huggingface.co/datasets/d0rj/LLaVA-OneVision-Data-ru.physical-ai-bench-generation
Physical AI Bench - Generation
Paper | Code
Dataset Description
The PAI-Bench is a benchmark to measure the progress of world models quantitatively.
The predict task contains a list of 1044 samples of text prompts, conditioning images, and qa pairs, covering Physical AI target domains including autonomous vehicle (AV) driving, robotics, industry (smart space), physics, human, and common sense. All the questions are binary questions, and the answer is either Yes or No. Our… See the full description on the dataset page: https://huggingface.co/datasets/shi-labs/physical-ai-bench-generation.Twin-2K-500
Twin-2K-500 Dataset
This dataset Twin-2K-500 contains comprehensive persona information from a representative sample of 2,058 US participants, providing rich demographic and psychological data. The dataset is specifically designed for building digital twins for LLM simulations.
More information on how to use this dataset can be found in our Documentation and GitHub repository.
Details on how the dataset was generated are available in our Paper.
Dataset Creation… See the full description on the dataset page: https://huggingface.co/datasets/LLM-Digital-Twin/Twin-2K-500.Tunisian-Proverbs-with-Image-Associations-A-Cultural-and-Linguistic-DatasetTunisian Proverbs with Image Associations: A Cultural and Linguistic Dataset
Description
This dataset explores the rich oral tradition of Tunisian proverbs mapped into text format, pairing each with contextual explanations, English translations both word-to-word and it's equivalent Target Language dynamic, Automated prompt and AI-generated visual interpretations.
It bridges linguistic, cultural, and visual modalities making it valuable for tasks in cross-cultural NLP, generative… See the full description on the dataset page: https://huggingface.co/datasets/HabibaAbderrahim/Tunisian-Proverbs-with-Image-Associations-A-Cultural-and-Linguistic-Dataset.biorXiv-pdf
BiorXiv Pdf
BiorXiv PDF dataset is a collection of PDF documents gathered from the BiorXiv website. This initiative aims to democratize artificial intelligence research by providing researchers with access to readily available training datasets. It is part of our broader effort to publish open access research papers as collective datasets.
BiorXiv is a renowned preprint publication in the field of biology and related disciplines. It is operated by Cold Spring Harbor Laboratory (CSHL)… See the full description on the dataset page: https://huggingface.co/datasets/laion/biorXiv-pdf.LongDS
LongDS
Overview •
Versions •
Download •
Usage •
Task Index •
Citation
Paper •
Code •
Web
LongDS-Bench is a benchmark for evaluating long-horizon, multi-turn agentic data analysis. Real-world analysis is rarely a sequence of independent questions: filters, metric definitions, assumptions, intermediate tables, and branch-specific results evolve over many turns. LongDS tests whether agents can maintain and apply these evolving analytical states correctly.… See the full description on the dataset page: https://huggingface.co/datasets/zjunlp/LongDS.WebCompass
WebCompass
A unified multimodal benchmark for evaluating LLMs' ability to generate, edit, and repair functional web pages. WebCompass spans three input modalities — text design documents, reference screenshots, and video demonstrations — and three task families — generation, editing, and repair.
GitHub: NJU-LINK/WebCompass
Project Page: nju-link.github.io/WebCompass
Quick Start
from datasets import load_dataset
# Generation tasks (existing)
ds_text =… See the full description on the dataset page: https://huggingface.co/datasets/NJU-LINK/WebCompass.LLAVA-LibMoE
Download Instructions
This repository provides the full LLAVA-LibMoE dataset, including LLaVA-665K and OneVision-1M2 image sources, organized into the following required directory tree:
libmoe/
└── data/
├── image_onevision/
├── coco/
│ └── train2017/
├── gqa/
│ └── images/
├── ocr_vqa/
│ └── images/
├── textvqa/
│ └── train_images/
└── vg/
├── VG_100K/
└──… See the full description on the dataset page: https://huggingface.co/datasets/DavidNguyen/LLAVA-LibMoE.llava-en-zh-300kThis dataset is composed by
150k examples of English Visual Instruction Data from LLaVA.
150k examples of English Visual Instruction Data from openbmb.
You can use it in LLaMA Factory by specifying --dataset llava_150k_en,llava_150k_zh.
lca-bug-localization
🏟️ Long Code Arena (Bug localization)
This is the benchmark for the Bug localization task as part of the
🏟️ Long Code Arena benchmark.
The bug localization problem can be formulated as follows: given an issue with a bug description and a repository snapshot in a state where the bug is reproducible, identify the files within the repository that need to be modified to address the reported bug.
The dataset provides all the required components for evaluation of bug localization… See the full description on the dataset page: https://huggingface.co/datasets/JetBrains-Research/lca-bug-localization.distilled-web
Dataset Card for lenamerkli/distilled-web
This dataset consists of web-scraped data using a custom crawler purpose-built for each website.
Dataset Details
Dataset Sources
Repository: https://github.com/lenamerkli/distilled-web
Uses
This dataset is useful for training large language models.
The train split provides instruction-following and chat data for supervised fine-tuning (SFT) and instruction tuning.
The pretrain split… See the full description on the dataset page: https://huggingface.co/datasets/lenamerkli/distilled-web.AIDA
Dataset Card for AIDABench
Links
Paper (arXiv)
GitHub Repository
Dataset Summary
AIDABench is a benchmark for evaluating AI systems on end-to-end data analytics over real-world documents. It contains 600+ diverse analytical tasks grounded in realistic scenarios and spans heterogeneous data sources such as spreadsheets, databases, financial reports, and operational records. Tasks are designed to be challenging, often requiring multi-step reasoning… See the full description on the dataset page: https://huggingface.co/datasets/MichaelYang-lyx/AIDA.ChEBI-20-MM
ChEBI-20-MM Dataset
Overview
The ChEBI-20-MM is an extensive and multi-modal benchmark developed from the ChEBI-20 dataset. It is designed to provide a comprehensive benchmark for evaluating various models' capabilities in the field of molecular science. This benchmark integrates multi-modal data, including InChI, IUPAC, SELFIES, and images, making it a versatile tool for a wide range of molecular tasks.
Dataset Description
ChEBI-20-MM is an expansion of the… See the full description on the dataset page: https://huggingface.co/datasets/liupf/ChEBI-20-MM.liquidrandom-data
liquidrandom-data
Diverse seed data for ML/LLM training data generation pipelines.
Used by the liquidrandom Python package.
Dataset Summary
This dataset contains 520,080 seed data samples across 24 categories,
generated using a hierarchical taxonomy tree approach with LLM-based quality validation
and fuzzy deduplication. Data is stored as Parquet with zstd compression.
Categories
Category
Samples
File
Coding Tasks
30,069… See the full description on the dataset page: https://huggingface.co/datasets/mlech26l/liquidrandom-data.MLLM-as-a-JudgeOmni-Edu
Omni-Edu — Core V6 SFT mixture
69,999 supervised instruction examples (~158M characters) covering K-12 subject
competence, curriculum grounding, diagnostic reasoning, pedagogical action and
general-purpose instruction. 12,146 rows (17.4%) are multimodal; every image
referenced by the JSONL ships in this repository under images/.
This is the system-prompted assembly of the v6 core mixture: every row carries
an explicit system message, and the non-system turns are byte-identical… See the full description on the dataset page: https://huggingface.co/datasets/lhpku20010120/Omni-Edu.ALL-Bench-Leaderboard
🏆 ALL Bench Leaderboard 2026
The only AI benchmark dataset covering LLM · VLM · Agent · Image · Video · Music in a single unified file.
Dataset Summary
ALL Bench Leaderboard aggregates and cross-verifies benchmark scores for 90+ AI models across 6 modalities. Every numerical score is tagged with a confidence level (cross-verified, single-source, or self-reported) and its original source. The dataset is designed for researchers, developers, and… See the full description on the dataset page: https://huggingface.co/datasets/FINAL-Bench/ALL-Bench-Leaderboard.encyclopaedia-britannica-lance
Encyclopaedia Britannica (1771-1860) - Lance Format
This dataset contains 155,388 digitized pages from the Encyclopaedia Britannica, spanning editions from 1771 to 1860. The data is stored in Lance format for efficient streaming and lazy image loading.
Dataset Details
Total Pages: 155,388
Total Volumes: 195
Format: Lance (columnar format with blob storage for images)
Source: National Library of Scotland (NLS)
License: Public Domain (CC0)
Loading the Dataset… See the full description on the dataset page: https://huggingface.co/datasets/NationalLibraryOfScotland/encyclopaedia-britannica-lance.LegoFlow-SWE
LegoFlow-SWE · 5,000 verified Harbor SWE tasks and two GLM-5.2 trajectory releases
GitHub · Docs · Blog · HuggingFace · LegoX
LegoFlow-SWE
5,000 verified Harbor SWE tasks mined by LegoFlow Curator, shipped in original and anti-hack prompt versions, plus two GLM-5.2 trajectory releases under OpenHands SDK and OpenCode, totaling 9,767 trajectories.
Release
Count
What it is
tasks/
5,000
Original prompts
tasks-anti-hack/
5,000
Same task IDs and task files, with… See the full description on the dataset page: https://huggingface.co/datasets/Lego-X/LegoFlow-SWE.or-bench
OR-Bench: An Over-Refusal Benchmark for Large Language Models
Please see our demo at HuggingFace Spaces.
Overall Plots of Model Performances
Below is the overall model performance. X axis shows the rejection rate on OR-Bench-Hard-1K and Y axis shows the rejection rate on OR-Bench-Toxic. The best aligned model should be on the top left corner of the plot where the model rejects the most number of toxic prompts and least number of safe prompts. We also plot a blue line… See the full description on the dataset page: https://huggingface.co/datasets/bench-llms/or-bench.encyclopaedia-britannica-lance-test
Encyclopaedia Britannica (1771-1860) - Lance Format
This dataset contains 155,388 digitized pages from the Encyclopaedia Britannica, spanning editions from 1771 to 1860. The data is stored in Lance format for efficient streaming and lazy image loading.
Dataset Details
Total Pages: 155,388
Total Volumes: 195
Format: Lance (columnar format with blob storage for images)
Source: National Library of Scotland (NLS)
License: Public Domain (CC0)
Loading the Dataset… See the full description on the dataset page: https://huggingface.co/datasets/davanstrien/encyclopaedia-britannica-lance-test.SEMM-Latent-Telemetry
SEMM-Latent-Telemetry
Bare-metal hardware telemetry and SNN latent space routing data for neuromorphic quantization research. This dataset documents the discovery of Semantic Attractor Clustering — that a Spiking Neural Network physically routes different semantic concepts (abstract language vs code syntax vs math logic) into distinct, repeatable biological pathways when L2 Normalization is applied to LLM embeddings.
Hub ID: rmems/SEMM-Latent-TelemetryNames: SEMM = Spiking… See the full description on the dataset page: https://huggingface.co/datasets/rmems/SEMM-Latent-Telemetry.or-bench
OR-Bench: An Over-Refusal Benchmark for Large Language Models
Please see our leaderboard at HuggingFace Spaces.
Overall Plots of Model Performances
Below is the overall model performance. X axis shows the rejection rate on OR-Bench-Hard-1K and Y axis shows the rejection rate on OR-Bench-Toxic. The best aligned model should be on the top left corner of the plot where the model rejects the most number of toxic prompts and least number of safe prompts. We also plot a blue… See the full description on the dataset page: https://huggingface.co/datasets/orbench-llm/or-bench.BiasShadesInterested in contributing? Speak a language not represented here? Disagree with an annotation? Please submit feedback in the Community tab!
Dataset Card for BiasShades
Note: This dataset may NOT be used as training data in any form (pre-training, fine-tuning, post-training, etc.) without express permission from creators.
Dataset Details
Version: 1.0
License: SHADES 1 Montreal Data License
Dataset Description
728 stereotypes and associated… See the full description on the dataset page: https://huggingface.co/datasets/LanguageShades/BiasShades.BalitaNLPA Filipino multi-modal language dataset for text+visual tasks. Consists of 351,755 Filipino news articles (w/ associated images) gathered from Filipino news outlets.
Description
Total # of articles: 351,755
80-10-10 split for training, validation, and testing.
Dataset field descriptions:
title - Article title
body - Article body. Separated into paragraphs
image - Article image
website… See the full description on the dataset page: https://huggingface.co/datasets/LanceBunag/BalitaNLP.coco-captions-pt-br
🎉 COCO Captions Dataset Translation for Portuguese Image Captioning
💾 Dataset Summary
COCO Captions Portuguese Translation, a multimodal dataset for Portuguese image captioning with 123,287 images, each accompanied by five descriptive captions that have been
generated by human annotators for every individual image. The original English captions were rendered into Portuguese
through the utilization of the Google Translator API.
🧑💻 Hot to Get… See the full description on the dataset page: https://huggingface.co/datasets/laicsiifes/coco-captions-pt-br.HYDRA-M3-V0
MMM_HYDRA: Heterogeneous Yielding Dataset for Reasoning Across - Multi-hop, Multimodal, Multicompany
Dataset Description
MMM_HYDRA is a benchmark dataset for evaluating Retrieval-Augmented Generation (RAG) systems on complex financial document analysis. The dataset contains 200 carefully curated questions with answers extracted from 99 unique corporate 10-K filings across 15 industry sectors.
Key Features
Multi-Company: 54 questions (27%) span multiple… See the full description on the dataset page: https://huggingface.co/datasets/large-traversaal/HYDRA-M3-V0.craft-benchmark-lean
CRAFT Benchmark Dataset
Trajectory logs from the CRAFT benchmark — a multi-agent evaluation of pragmatic communication in LLMs under strict partial information. - TL;DR
Dataset Structure
Each row is one turn from a CRAFT game, with fields for:
Identity: structure_id, director_model, builder_model, model_type (base/frontier), turn_number
Director responses: D1_thinking, D1_message, D2_thinking, D2_message, D3_thinking, D3_message
Builder: builder_action, builder_block… See the full description on the dataset page: https://huggingface.co/datasets/Abhijnan/craft-benchmark-lean.