This dataset contains a large collection of high-quality Arabic text data with their metadata.
To access the full data please visit Token Haven
Creation
The dataset was created by filtering all English common crawl data for high-quality text using the FineWeb-Edu classifier with education score of 4 or higher over 5.
The data is source from the v1.0.0 of the HuggingFaceFW/fineweb-edu dataset which corresponds to CC-MAIN-2024-10 from… See the full description on the dataset page: https://huggingface.co/datasets/TokenHaven/FineWeb-Edu-Arabic.
