Pashto Datasets

Curated and processed Pashto datasets published by ZamAI Labs.
This repository includes custom cleaning, normalization, and consolidation work, plus documentation and attribution to original sources.

• Website: https://zamai.dev
• ZamAI Labs: https://github.com/ZamAI-ORG
• Repo: https://github.com/ZamAI-ORG/pashto-datasets

Repository layout

• DATASETS/ contains dataset folders. Each dataset includes:
  • SOURCE.md (where it came from)
  • LICENSE.md (original license or terms, if applicable)
  • raw/ (optional; small samples for testing when permitted)
  • processed/ (normalized/cleaned outputs ready for training)
  • notes.md (what ZamAI changed)

Sources & attribution

All datasets in this repository originate from sources that allow redistribution. Each dataset folder includes source links and license/terms to preserve attribution and compliance.

Raw data note

Some datasets include small raw/ samples used for testing and validation. Processed datasets live under processed/.