Skip to content

Dataset · MohamedRashad

Arabic books text corpus

MohamedRashad/arabic-books

A collection of 8,500 full Arabic book texts totaling 1.1 billion tokens for language models.

What's inside

The dataset contains 8,500 rows with Arabic book texts extracted using the arabic-large-nougat model. The total volume is 1.1 billion tokens (according to the GPT-4 tokenizer). The data demonstrates the capabilities of Arabic OCR and text extraction quality.

What you can do with it

  • Training and fine-tuning language models in Arabic
  • Evaluating the quality of OCR and text extraction systems
Tasks
Text generation
Languages
Arabic
Size
1K–10K rows
Downloads this month
24.1K
Likes
3

How to load it

from datasets import load_dataset

ds = load_dataset("MohamedRashad/arabic-books")