datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ArabicMMLU
Fajri Koto, Haonan Li, Sara Shatnawi, Jad Doughman, Abdelrahman Boda Sadallah, Aisha Alraeesi, Khalid Almubarak, Zaid Alyafeai, Neha Sengupta, Shady Shehata, Nizar Habash, Preslav Nakov, and Timothy Baldwin
MBZUAI, Prince Sattam bin Abdulaziz University, KFUPM, Core42, NYU Abu Dhabi, The University of Melbourne
Introduction
We present ArabicMMLU, the first multi-task language understanding benchmark for Arabic language, sourced from school exams across diverse… See the full description on the dataset page: https://huggingface.co/datasets/MBZUAI/ArabicMMLU.arabic-stem-lexicon
Arabic Diacritized-Stem Lexicon
An undiacritized Arabic surface form → its most frequent diacritized stem.
Standard Arabic writes no short vowels, so anything that has to pronounce Arabic
must first put them back. A neural diacritizer does that well on rare words, where
inference is the only thing there is. On common words it is the wrong tool:
which vowels كتاب carries is not a thing to be inferred, it is a thing to be looked
up — and models get exactly these wrong, reading… See the full description on the dataset page: https://huggingface.co/datasets/TigreGotico/arabic-stem-lexicon.arabic-dialects-gold20
arabic-dialects-gold20
660 sentences: 33 Arabic lects × 20 sentences, each with fully diacritized
dialectal orthography, the undiacritized surface form, gold IPA, an engine
draft, an English gloss, machine-verified phonetic feature tags, per-row
verification metadata, and notes citing the dialectological literature that
grounds the row.
Columns (TSV, UTF-8, one file per lect):
id, sentence, raw, ipa, ipa_o2i, gloss_en, features, notes, fable_corrections, verification… See the full description on the dataset page: https://huggingface.co/datasets/Salesteq/arabic-dialects-gold20.gpqa_diamondarabic-qna
Sadeem QnA: An Arabic QnA Dataset 🌍✨
Welcome to the Sadeem QnA dataset, a vibrant collection designed for the advancement of Arabic natural language processing, specifically tailored for Question Answering (QnA) systems. Sourced from the rich and diverse content of Arabic Wikipedia, this dataset is a gateway to exploring the depths of Arabic language understanding, offering a unique challenge to both researchers and AI enthusiasts alike.
About Sadeem QnA
The Sadeem… See the full description on the dataset page: https://huggingface.co/datasets/sadeem-ai/arabic-qna.Arabic-Dialects
Arabic Dialects Dataset (Bivalency & Code-Switching)
The Arabic Dialects Dataset is a specialised corpus designed for automatic dialect identification, with a focus on the linguistic phenomena of bivalency and written code-switching between major Arabic dialects and Modern Standard Arabic (MSA).It covers five varieties:
EGY – Egyptian Arabic
GLF – Gulf Arabic
LAV – Levantine Arabic
NOR – North African / Tunisian Arabic
MSA – Modern Standard Arabic
The dataset was created… See the full description on the dataset page: https://huggingface.co/datasets/drelhaj/Arabic-Dialects.Shifaa_Arabic_Medical_Consultations
Shifaa Arabic Medical Consultations 🏥📊
Overview 🌍
Shifaa is revolutionizing Arabic medical AI by addressing the critical gap in Arabic medical datasets. Our first contribution is the Shifaa Arabic Medical Consultations dataset, a comprehensive collection of 84,422 real-world medical consultations covering 16 Main Specializations and 585 Hierarchical Diagnoses.
🔍 Why is this dataset important?
First large-scale Arabic medical dataset for AI applications.… See the full description on the dataset page: https://huggingface.co/datasets/Ahmed-Selem/Shifaa_Arabic_Medical_Consultations.Arabic_sentimentArabic-NLi-Triplet
Arabic NLI Triplet
Dataset Summary
The Arabic Version of SNLI and MultiNLI datasets. (Triplet Subset)
Originally used for Natural Language Inference (NLI),
Dataset may be used for training/finetuning an embedding model for semantic textual similarity.
Triplet Subset
Columns: "anchor", "positive", "negative"
Column types: str, str, str
Examples:
{
"anchor": "شخص على حصان يقفز فوق طائرة معطلة",
"positive": "شخص في الهواء الطلق، على حصان.",
"negative":… See the full description on the dataset page: https://huggingface.co/datasets/Omartificial-Intelligence-Space/Arabic-NLi-Triplet.Arabic-stsb
Arabic STSB Structure
The Arabic Version of the the Semantic Textual Similarity Benchmark (Cer et al., 2017)
it is a collection of sentence pairs drawn from news headlines, video and image captions, and natural language inference data.
Each pair is human-annotated with a similarity score from 1 to 5. However, for this variant, the similarity scores are normalized to between 0 and 1.
Examples:
{
"sentence1": "طائرة ستقلع",
"sentence2": "طائرة جوية ستقلع",
"score": 1.0
}
{… See the full description on the dataset page: https://huggingface.co/datasets/Omartificial-Intelligence-Space/Arabic-stsb.Rasaif-Classical-Arabic-English-Parallel-texts
Introduction
This dataset represents a curated collection of parallel Arabic-English texts, featuring the translations of 24 historically and culturally significant books. These texts provide a portal to the intellectual and literary heritage of the Arabic-speaking world during its classical period.
Content Details
Contained within this dataset are English translations of the following texts, sourced from the Rasaif website:
A Muslim Manual of War
Al-Hanin Ila'l-Awtan… See the full description on the dataset page: https://huggingface.co/datasets/ImruQays/Rasaif-Classical-Arabic-English-Parallel-texts.AraDiCE
AraDiCE: Benchmarks for Dialectal and Cultural Capabilities in LLMs
Overview
The AraDiCE dataset is designed to evaluate dialectal and cultural capabilities in large language models (LLMs). The dataset consists of post-edited versions of various benchmark datasets, curated for validation in cultural and dialectal contexts relevant to Arabic.
As part of the supplemental materials, we have selected a few datasets (see below) for the reader to review. We will make the full… See the full description on the dataset page: https://huggingface.co/datasets/QCRI/AraDiCE.Arabic-NLi-Pair-Score
Arabic NLI Pair-Score
Dataset Summary
The Arabic Version of SNLI and MultiNLI datasets. (Pair-Score Subset)
Originally used for Natural Language Inference (NLI),
Dataset may be used for training/finetuning an embedding model for semantic textual similarity.
Pair-Class Subset
Columns: "sentence1", "sentence2", "score"
Column types: str, str, float
Arabic Examples:
{
"sentence1": "شخص على حصان يقفز فوق طائرة معطلة",
"sentence2": "شخص يقوم… See the full description on the dataset page: https://huggingface.co/datasets/Omartificial-Intelligence-Space/Arabic-NLi-Pair-Score.arabic-dialects-gold20-code-switch
gold20-code-switch
Code-switched Arabic sentences with IPA: 20 rows per lect across 33 Arabic
lects (the same roster as the sibling TigreGotico/arabic-dialects-gold20).
Each row embeds foreign material in a dialectal Arabic frame: inline
Latin-script English (and French, for the lects whose live contact language
is French), Arabic-script loanwords (سيرفس، كاش، موبايل-class), and Arabizi
(Latin-written Arabic with digit gutturals).
Columns (TSV, UTF-8, one file per lect):
id… See the full description on the dataset page: https://huggingface.co/datasets/Salesteq/arabic-dialects-gold20-code-switch.Saudi-Arabia-Stock-Symbols-and-Metadata
Saudi Arabia Stock Symbols & Company Metadata
This dataset contains stock symbols and basic company metadata for all listed companies in Saudi Arabia.It is updated weekly if new changes are there.
📊 Dataset Contents
The dataset is provided as a CSV file with the following columns:
Column
Description
name
Full company name
ticker
Stock ticker symbol (e.g., AAPL, MSFT)
market
The exchange/market where the stock is listed
sector
The primary business sector… See the full description on the dataset page: https://huggingface.co/datasets/kjhq/Saudi-Arabia-Stock-Symbols-and-Metadata.Arabic_Sentiment_DatasetShifaa_Arabic_Mental_Health_Consultations
🏥 Shifaa Arabic Mental Health Consultations 🧠
📌 Overview
Shifaa Arabic Mental Health Consultations is a high-quality dataset designed to advance Arabic medical language models.This dataset provides 35,648 real-world medical consultations, covering a wide range of mental health concerns.
📊 Dataset Summary
Size: 35,648 consultations
Main Specializations: 7
Specific Diagnoses: 123
Languages: Arabic (العربية)
Why This Dataset?
🔹 Lack of… See the full description on the dataset page: https://huggingface.co/datasets/Ahmed-Selem/Shifaa_Arabic_Mental_Health_Consultations.central-kurdish-tts4all
TTS4All Central Kurdish Speech Dataset
Dataset Summary
The TTS4All Central Kurdish Speech Dataset is a multi-speaker speech corpus developed for speech synthesis and speech technology research in Central Kurdish (Sorani Kurdish).
The dataset was created within the TTS4All initiative during the JSALT 2025 Workshop and provides more than 35 hours of transcribed speech from three native Central Kurdish speakers.
The corpus was designed to support:
Text-to-Speech… See the full description on the dataset page: https://huggingface.co/datasets/aranemini/central-kurdish-tts4all.arabic-sentiments2Arabic-news-and-management-corpus
Arabic Management, Economics & Financial News Corpus (1,200 Articles)
This corpus contains 1,200 Arabic news and management articles drawn from three distinct domains. It was originally compiled as part of research into Arabic Corpus Linguistics, management communication, financial discourse and domain-specific NLP. Both plain text and POS-tagged versions are available.
The dataset has been widely used in teaching and research, including the King Saud University book Corpus… See the full description on the dataset page: https://huggingface.co/datasets/drelhaj/Arabic-news-and-management-corpus.Gutenberg-Arabic-OCR-HTML-Pages
Gutenberg Arabic HTML-Page Dataset
📖 Dataset Description
The Gutenberg Arabic HTML-Page Dataset is a large-scale, synthetically generated dataset designed for training and evaluating document understanding and Optical Character Recognition (OCR) models. The primary goal of this project is to provide a comprehensive resource of page images paired with their corresponding structured HTML ground truth, with a focus on the Arabic language.
The dataset was created by… See the full description on the dataset page: https://huggingface.co/datasets/FatimahEmadEldin/Gutenberg-Arabic-OCR-HTML-Pages.Multi-Arabic-dialectsKazakhLawCorpus-clean
KazakhLawCorpus-clean
Dataset Summary
KazakhLawCorpus-clean is a cleaned, Kazakh-only corpus of legislative documents from the Republic of Kazakhstan. It is a processed derivative of the original Arailym-tleubayeva/KazakhLawCorpus dataset.
The original dataset repository was downloaded from Hugging Face and used as the source for this release. Its laws_metadata.csv file contained 223,245 legislative records with multilingual fields and source-oriented metadata.… See the full description on the dataset page: https://huggingface.co/datasets/Arailym-tleubayeva/KazakhLawCorpus-clean.sada2022-arabic-tts
SADA 2022 - Saudi Arabic Dataset for TTS
مجموعة بيانات صوتية سعودية للنص إلى كلام (Text-to-Speech)
المصدر الأصلي
Kaggle: sdaiancai/sada2022
الاستخدام
# طريقة 1: Git Clone
!git clone https://huggingface.co/datasets/aalshalfi/sada2022-arabic-tts /content/saudi_dataset
# طريقة 2: مكتبة datasets
from datasets import load_dataset
dataset = load_dataset("aalshalfi/sada2022-arabic-tts")
الملفات
valid.csv - ملف البيانات الرئيسي
wavs/ - ملفات الصوت… See the full description on the dataset page: https://huggingface.co/datasets/aalshalfi/sada2022-arabic-tts.saudipedia-arabic-qa
Saudipedia Q&A Dataset
Dataset Description
Summary
This dataset contains question-answer pairs scraped from Saudipedia, a comprehensive Arabic encyclopedia focused on Saudi Arabia. The dataset includes 1,082 Q&A entries covering various topics related to Saudi culture, history, economy, government, society, geography, religion, and notable personalities.
The data was collected by scraping the website's question-answer section, which provides detailed answers to… See the full description on the dataset page: https://huggingface.co/datasets/AhmadHakami/saudipedia-arabic-qa.xnli2.0_train_arabicGulf-Arabic-Tweets-2018-2020
Dataset Summary
This is a pre-processed (cleaned) Twitter Gulf Arabic dialect 2018-2020 dataset. Pleasre refer to the source, and data cleaning code and algorithm Github.
Languages
Arabic
Source Data
Twitter
arabic-news
Arabic News Hierarchical Classification Dataset
A dataset containing 265,860 Arabic news articles designed for Arabic news classification and hierarchical text classification.
Dataset Structure
The dataset contains the following fields:
content: Arabic news article text.
url:the article url.
date:Publication Date.
category: Original article category.
MainCategory: Main-level category.
SubCategoryLevel1: First-level subcategory.
SubCategoryLevel2: Second-level… See the full description on the dataset page: https://huggingface.co/datasets/ahmed-altaweel/arabic-news.hatecheck-arabic
Dataset Card for Multilingual HateCheck
Dataset Description
Multilingual HateCheck (MHC) is a suite of functional tests for hate speech detection models in 10 different languages: Arabic, Dutch, French, German, Hindi, Italian, Mandarin, Polish, Portuguese and Spanish.
For each language, there are 25+ functional tests that correspond to distinct types of hate and challenging non-hate.
This allows for targeted diagnostic insights into model performance.
For more details… See the full description on the dataset page: https://huggingface.co/datasets/Paul/hatecheck-arabic.arabic-enterprise
Arabic Enterprise
This is a proposed dataset for evaluating enterprise use cases for LLMs in Arabic.