Skip to content

Dataset · AuthenticIlm

Full Corpus of the Classical Islamic Library Shamela 4

AuthenticIlm/Shamela4_Full_DB

Digitized collection of 8,589 Arabic books across 40 categories of classical Islamic sciences.

What's inside

This dataset is a full dump from the al-Maktaba al-Shamela v4 database, created in April 2026. It contains 8,589 books (approx. 7.6 million pages or 19 GB of text) in Arabic, covering 40 thematic categories. Metadata is organized in parquet and JSONL formats.

What you can do with it

  • Pre-training and fine-tuning language models on classical Arabic
  • Building text-retrieval and question-answering systems for Islamic texts
  • Training models for fill-mask and token classification tasks
Tasks
Text generation, Fill mask, Token classification, Question answering, Text retrieval
Languages
Arabic
Size
10M–100M rows
Downloads this month
23.7K
Likes
38

How to load it

from datasets import load_dataset

ds = load_dataset("AuthenticIlm/Shamela4_Full_DB")