datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
saudi-dialect-conversations
Saudi Najdi Dialect Conversations
A curated dataset of 3,545 multi-turn conversations in Saudi Najdi Arabic dialect (the dialect spoken in Riyadh, Qassim, and central Najd region). Designed for Supervised Fine-Tuning (SFT) of Arabic language models.
Dataset Details
Metric
Value
Total conversations
3,545
Total turns
22,536
Average turns per conversation
6.4
Complexity distribution
Simple: 31%, Intermediate: 38%, Advanced: 31%
Topics covered
18 categories… See the full description on the dataset page: https://huggingface.co/datasets/HeshamHaroon/saudi-dialect-conversations.saudi-arabic-laws-and-regulations-corpus
Saudi Arabic Laws and Regulations Corpus
A structured, article-level Arabic legal corpus containing 22,593 records from 578 official Saudi legal documents, prepared for Arabic legal information retrieval, Retrieval-Augmented Generation (RAG), grounded generation, and LLM evaluation.
Release: 1.0.0Language: ArabicDomain: Saudi laws and regulationsGranularity: Legal articleTotal records: 22,593Archival DOI: 10.5281/zenodo.21265180
Quick Start
The corpus is… See the full description on the dataset page: https://huggingface.co/datasets/SaudiArabicLaws/saudi-arabic-laws-and-regulations-corpus.saudi-arabic-cs-conversations
Saudi Arabic Customer Service Conversations — Free 100 Sample
100 synthetic multi-turn conversations in authentic Saudi Arabic dialects
Built for LLM fine-tuning, chatbot training, and Arabic NLP research
Overview
This is a free 100-conversation sample from a production-quality dataset of 50,000 Saudi Arabic customer service conversations. Every conversation is fully synthetic — no real user data — and safe for commercial use.
Each conversation simulates a… See the full description on the dataset page: https://huggingface.co/datasets/dev-hussein/saudi-arabic-cs-conversations.voho-saudi-dialogues
Voho Saudi Dialogues
13,156 multi-turn conversations in spoken Saudi Arabic (Najdi), 105,808 turns, 694,240 words, from Voho. Apache 2.0.
Two halves. 7,603 service calls across the eight sectors Voho's voice agents work in — a technician handing over a rig shift, a customer disputing a SADAD charge, a permit-to-work request — and 5,553 everyday conversations between people who know each other: family, food, driving, weddings, the Hilal–Nassr match. Nothing like the first half… See the full description on the dataset page: https://huggingface.co/datasets/VohoAI/voho-saudi-dialogues.saudi-arabian-passports
Disclaimer: All passport images and associated data in this dataset are synthetically generated and do not correspond to real individuals. Any names, numbers, or personal details are fictional and used solely for research and development purposes.
Introduction - Saudi Arabia
The Synthetic Saudi Arabia Passports Dataset compiles more than 1,000 AI-generated passport images built for training OCR and computer vision models on identity documents. Every record is fully synthetic… See the full description on the dataset page: https://huggingface.co/datasets/ud-synthetic/saudi-arabian-passports.Saudi-Arabic-Alzheimers-Conversational-Dataset-Parameterized
Saudi Arabic Alzheimer's Patient QA Dataset (Conversational)
Overview
This dataset contains parameterized question-answer pairs designed for conversational AI assistants supporting Alzheimer's patients. The questions are written in the Saudi Arabic dialect and cover common memory-related interactions.
Features
Saudi Arabic dialect
Parameterized answers
Alzheimer's memory support
Conversational QA
RAG-ready
Language
Arabic (Saudi… See the full description on the dataset page: https://huggingface.co/datasets/ShahadAljohani/Saudi-Arabic-Alzheimers-Conversational-Dataset-Parameterized.sera-phase2-saudi-dialect-rag
SERA Saudi Dialect RAG Dataset (Phase 2 - Domain)
Domain-specific RAG fine-tuning dataset for Saudi Arabic dialect,
focused on Saudi Electricity Regulatory Authority (SERA) documents.
Format
LlamaFactory Alpaca format:
Field
Description
instruction
System prompt + real document chunk as context + question in Saudi dialect
input
Always empty
output
Answer in Saudi dialect
Usage with LlamaFactory
Copy the JSON files into your LlamaFactory… See the full description on the dataset page: https://huggingface.co/datasets/Rabe3/sera-phase2-saudi-dialect-rag.saudi-dialect-rag
Saudi Dialect RAG Fine-Tuning Dataset
A RAG-formatted fine-tuning dataset for Saudi Arabic dialect, built from
HeshamHaroon/saudi-dialect-conversations.
Format
Each example follows the LlamaFactory Alpaca format:
Field
Description
instruction
System prompt + MSA context paragraph + optional conversation history + question
input
Always empty string
output
Assistant reply in Saudi dialect
How it was built
Loaded source multi-turn Saudi… See the full description on the dataset page: https://huggingface.co/datasets/Rabe3/saudi-dialect-rag.SaudiLawEnglishsaudi-data-eou.jsonl
Saudi Arabic End-of-Utterance (EOU) Dataset
This dataset is designed for End-of-Utterance (EOU) detection in Arabic conversational AI, with a focus on the Saudi dialect (ar-SA).
Task
Binary classification:
0 → Incomplete utterance (speaker likely to continue)
1 → Complete utterance (end of turn)
Data Format
Each entry is a JSON object with:
text: Arabic conversational utterance
label: 0 or 1
Example:
{"text":"ابي احجز موعد بس ...","label":0}
arabic_eou_saudi_10000
Arabic End-of-Utterance (EOU) Dataset — Saudi Dialect
This dataset contains 10,000 synthetic and semi-realistic conversational Arabic utterances labeled for end-of-utterance detection.
Format
Each line is a JSON object:
{"text": "...", "label": 1}