datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
karsl-502-arabic-sign-language-v2silma-arabic-english-sts-dataset-v1.0
SILMA STS Arabic/English Dataset - v1.0
Overview
The SILMA STS Arabic/English Dataset - v1.0 is a dataset designed for training and evaluating sentence embeddings for Arabic and English tasks. It consists of five different splits that cover monolingual and multilingual sentence pairs, with human-annotated similarity scores. The dataset includes both Arabic-to-Arabic and English-to-English pairs, as well as cross-lingual Arabic-English pairs, making it a valuable resource… See the full description on the dataset page: https://huggingface.co/datasets/silma-ai/silma-arabic-english-sts-dataset-v1.0.karsl-502-arabic-sign-language-finalsitr-arabic-pii
سِتر · Sitr
An Arabic-first PII detection dataset for Saudi Arabia, the Gulf and the Levant
Character-level spans for 43 personal-data types across 17,899 Arabic examples — regional dialects, code-switching, call-centre transcripts, and the long documents people paste into AI assistants.
Why this exists
General-purpose PII filters underperform badly in Arabic. Independent evaluation of the widely used openai/privacy-filter reports… See the full description on the dataset page: https://huggingface.co/datasets/mabahboh/sitr-arabic-pii.samer-arabic-text-simplification
SAMER Arabic Text Simplification Dataset (Cleaned Version)
Description
This dataset is a cleaned and structured version of the SAMER Corpus (The SAMER Arabic Text Simplification Corpus). It is prepared specifically for training Seq2Seq models (e.g., AraT5) and fine-tuning Large Language Models (LLMs) on Arabic Text Simplification and Readability Assessment tasks.
Dataset Structure
The dataset contains the following fields:
clean_text: Cleaned… See the full description on the dataset page: https://huggingface.co/datasets/vn3er/samer-arabic-text-simplification.arabic-triplets-1m-curated-sims-len
Arabic 1 Million Triplets (curated):
This is a curated dataset to use in Arabic ColBERT and SBERT models (among other uses).
In addition to anchor, positive and negative columns, the dataset has two columns: sim_pos and sim_neg which are cosine
similarities between the anchor (query) and bothe positive and negative examples.The last 3 columns are lengths (words) for each of the anchor, positive and negative examples. Length uses simple split on space, not tokens.
The cosine… See the full description on the dataset page: https://huggingface.co/datasets/akhooli/arabic-triplets-1m-curated-sims-len.samer-arabic-simplification-onlysilma-arabic-triplets-dataset-v1.0
SILMA Arabic Triplets Dataset - v1.0
Overview
The SILMA Arabic Triplets Dataset - v1.0 is a high-quality, diverse dataset specifically curated
for training and training embedding models for semantic search tasks
in the Arabic language.
The dataset contains more than 2.25M records (2,280,319 records).
This dataset includes triplets in the form of anchor, positive, and negative samples, designed to enhance models in learning semantic similarity and dissimilarity.
The… See the full description on the dataset page: https://huggingface.co/datasets/silma-ai/silma-arabic-triplets-dataset-v1.0.Arabic-Call-Center-Audio-Dataset-Single-ChannelDataset Description:
This dataset is a large-scale collection of 64,027 hours of processed Arabic (AR) single-channel call center audio recordings, part of a broader multilingual conversational audio collection containing approximately 3,569,083 processed call center recordings across 54 languages, designed to support the development and training of advanced speech AI and conversational AI systems.
The dataset captures authentic speech characteristics such as tone variation, pauses, silence… See the full description on the dataset page: https://huggingface.co/datasets/InfoBayAI/Arabic-Call-Center-Audio-Dataset-Single-Channel.arabic-broad-benchmark
Arabic Broad Benchmark (ABB)
The Arabic Broad Benchmark is a unique dataset and an advanced benchmark created by SILMA.AI to assess the performance of Large Language Models in Arabic Language.
ABB consists of 470 high quality human-validated questions sampled from 64 Arabic benchmarking datasets, evaluating 22 categories and skills.
The advanced benchmarking script utilizes the dataset to evaluate models or APIs using a mix of 20+ Manual Rules and LLM as Judge variations customized… See the full description on the dataset page: https://huggingface.co/datasets/silma-ai/arabic-broad-benchmark.six_millions_instruction_dataset_for_arabic_llm_ftarabic-wiki-simple-datasetarabic-qa-dataset-sigir2024
Arabic QA Dataset | 10,000 Instruction-Tuning Pairs
High-quality Arabic question-answering dataset designed for fine-tuning and instruction-tuning LLMs on Arabic language tasks.
Dataset Details
Size
10,000 entries
Format
JSONL (instruction/input/output)
Language
Arabic
License
MIT
Provider
AlTal Datamining
Schema
Field
Type
Description
instruction
string
The question or task in Arabic
input
string
Additional context… See the full description on the dataset page: https://huggingface.co/datasets/bobez999/arabic-qa-dataset-sigir2024.Arabic_dataset_13M_translated_cleaned_v2_jsonl_format_ViT-B-16-SigLIP-512_validationArabic_3M_5M_ViT-B-16-SigLIP-512
Loading the training split as follows:
from datasets import load_dataset
ds_train = load_dataset("Arabic-Clip/Arabic_3M_5M_ViT-B-16-SigLIP-512", split="train")
ds_train
# Dataset({
# features: ['index', 'url', 'en_caption', 'embeddings_en', 'caption_ar'],
# num_rows: 2000000
# })
Loading the validation split as follows:
from datasets import load_dataset
ds_validation = load_dataset("Arabic-Clip/Arabic_3M_5M_ViT-B-16-SigLIP-512", split="validation")… See the full description on the dataset page: https://huggingface.co/datasets/Arabic-Clip/Arabic_3M_5M_ViT-B-16-SigLIP-512.Arabic_MSCOCO_1st_ViT-B-16-SigLIP-512mscoco_captions_ViT-B-16-SigLIP-512simple-benchmark-arabic-summarizationsquad_arabicarabic-simpsons-blip-captionsarabic-text-simplification
Arabic Text Simplification Dataset
Complex-to-simplified Arabic text pairs for accessibility fine-tuning.
Why This Matters
Arabic text simplification is critical for:
Low-literacy readers — 1 in 5 Arabic speakers struggle with complex text
Cognitive accessibility — dyslexia, intellectual disabilities, autism
Non-native speakers — Arabic learners and expatriate workers
Children's content — making educational material age-appropriate
Almost no Arabic accessibility… See the full description on the dataset page: https://huggingface.co/datasets/dispatchAI/arabic-text-simplification.Arabic-Text-to-Sign-Language-TranslationSALMA-Arabic_Sense_Annotated_Corpussilma-arabic-testarabic_dataset_translated_v2_ViT-B-16-SigLIP-512Arabic-Reflection-Tuning-Natural-Farming-Instruct-SINGLE_SEED_EXAMPLE
Dataset Card for "Arabic-Reflection-Tuning-Natural-Farming-Instruct"
More Information needed