tasal9/ZamAI-Pashto-MegaDataset-v1

ZamAI Pashto Mega Dataset v1

Dataset Details

Dataset Description

The ZamAI Pashto Mega Dataset v1 is a high-quality instruction tuning dataset for Pashto language models containing 10,137 examples. It covers diverse task categories such as language fluency, news summarization, tutor chat, and more.

Curated by

[ZamAI]

Funded by

[Add funding source if any]

Shared by

[Add sharing organization or person]

Language(s) (NLP)

Pashto (ps)

License

Apache 2.0 License

Dataset Sources

Repository: [Add repository URL if applicable]
Paper: [Add paper URL or DOI]
Demo: [Add demo link]

---

Uses

Direct Use

[Add direct use cases here, e.g., fine-tuning Pashto instruction-tuned models]

Out-of-Scope Use

[Add any inappropriate or unsupported uses]

---

Dataset Structure

The dataset is composed of a training split with 10,137 samples.
Each sample contains the following fields:

Name	Type	Description
instruction	string	Task prompt or instruction for the model
input	string	Input text or context related to the task
response	string	Expected output or answer from the model
category	string	Task category (e.g., language_fluency, news_summarization)

Data files are stored under the path pattern: data/train-*

---

Dataset Creation

Curation Rationale

[Add rationale behind dataset creation]

Source Data

[Describe original data sources]

Data Collection and Processing

[Explain collection and preprocessing steps]

Who are the source data producers?

[Add info about producers]

---

Annotations

Annotation process

[Explain annotation methodology]

Who are the annotators?

[Describe annotators’ profiles]

---

Personal and Sensitive Information

[Add info on PII if any]

---

Bias, Risks, and Limitations

[Discuss any known biases, risks, or limitations of the dataset]

---

Recommendations

Users should be aware of potential biases and limitations in this dataset and exercise caution in sensitive applications.

---

Citation

Please cite this dataset as:

@misc{tasal2025zamai,
  title={ZamAI Pashto Mega Dataset v1},
  author={Yaqoob Tasal},
  year={2025},
  howpublished={\url{https://huggingface.co/datasets/tasal9/ZamAI-Pashto-MegaDataset-v1}},
}