nassimjp/pashto-reasoning-children-story-crafting-dataset

Pashto Reasoning Children Story Crafting Dataset

Welcome to the Pashto Reasoning Children Story Crafting Dataset! This dataset is designed to empower Large Language Models (LLMs) with the capability to craft engaging, moral, and logically structured children's stories in the Pashto language, integrating explicit reasoning steps.

Dataset Overview & Methodology

• —Language: Pashto (ps)
• —Base Prompts: 100 unique core story prompts.
• —Total Samples: 500 diverse story variations (each core prompt was expanded 5 times to maximize creative diversity, narrative style, and reasoning patterns).
• —Task: Text Generation, Creative Writing, Logical Reasoning
• —Domain: Children's Literature, Educational Content, Synthetic Reasoning

Tip for Developers: To achieve the best fine-tuning results, it is recommended to shuffle the data pairs thoroughly before initiating the model training process to prevent sequential bias.

Dataset Structure

Data Fields

• —prompt / instruction: The creative prompt or reasoning challenge provided to the model.
• —reasoning_steps: The step-by-step logical planning (character development, plot progression, moral integration) before crafting the story.
• —story: The final, beautifully crafted children's story written in fluent Pashto.

Example Entry (JSON)

{
  "prompt": "د یوې واړه مرغۍ په اړه کیسه ولیکئ چې د لومړي ځل لپاره یې پرواز زده کړ.",
  "reasoning_steps": "1. Character: A small, timid bird named Zarak. 2. Conflict: Afraid to jump out of the nest. 3. Action: Mother bird encourages him, and friends cheer him on. 4. Resolution: He takes a leap, discovers the joy of flying. 5. Moral: Courage begins with a single step.",
  "story": "په یوه شنه ځنګل کې، د بلوط په یوه لوړه ونه کې، یوې کوچنۍ مرغۍ ژوند کاوه چې نوم یې زارک وو..."
}

Use Cases

1. 1.LLM Fine-Tuning: Fine-tuning open-source models (like Llama, Mistral, or Qwen) using shuffled pairs to improve low-resource creative generation in Pashto.
2. 2.Chain-of-Thought (CoT) Reasoning: Teaching models how to structure narrative arcs and moral frameworks systematically before generating text.
3. 3.Educational Tools: Powering local apps, storytelling tools, and educational content for Pashto-speaking children.

Licensing Information

This dataset is distributed under the Apache 2.0 License.

Citation & Acknowledgments

If you use this dataset in your research or projects, please cite it as:

@dataset{pashto_reasoning_children_story_crafting,
  author = {Nassim JP},
  title = {Pashto Reasoning Children Story Crafting Dataset},
  year = {2026},
  publisher = {Hugging Face},
  url = {[https://huggingface.co/datasets/nassimjp/pashto-reasoning-children-story-crafting-dataset](https://huggingface.co/datasets/nassimjp/pashto-reasoning-children-story-crafting-dataset)}
}