Current State of Arabic NLP datasets

[Zaid Alyafeai]

In the past few weeks, I worked on improving the accessibility of Arabic NLP datasets in the Masader interface. Here are some numbers:

Total datasets: 650

Total datasets and subsets: 899

Freely accessible datasets: 376

HuggingFace datasets: 311

Upon-Request datasets: 66

Paid datasets: 208

If you know a missing dataset, please add it through this form.

Zaid