Qwill RP CreativeWriting Reasoning Dataset

📝 Dataset Summary

Qwill-RP-CreativeWriting-Reasoning is a creative writing dataset focused on structured reasoning. Each row contains a fictional or narrative prompt sourced from nothingiisreal/Reddit-Dirty-And-WritingPrompts, along with an AI-generated response that includes:

• Reasoning, wrapped in <think>...</think>
• Final Answer, wrapped in <answer>...</answer>

The goal is to train or evaluate models on chain-of-thought prompting within creative or storytelling domains.

---

📊 Dataset Preview

• Total Rows: ~3,000
• Columns:
  • prompt: The prompt given to the model.
  • model_used: The model that generated the response.
  • gemini: The structured response, with reasoning and final answer.
  • temperature: Temperature used in generation (randomly sampled, 0.1–1.2).
  • top_k: Top-k value used (sampled from: 20, 40, 50, 64, 80, 90, 100).
  • top_p: Top-p value used (sampled from: 0.8, 0.85, 0.9, 0.95, 1.0).

---

⚙️ Generation Methodology

We employed random sampling of generation parameters to avoid overly deterministic or excessively creative outputs:

• Temperature: Sampled from a continuous range between 0.1 and 1.3.
• Top-k: Randomly chosen from [20, 40, 50, 64, 80, 90, 100].
• Top-p: Randomly chosen from [0.8, 0.85, 0.9, 0.95, 1.0].

This encourages stylistic and logical variety across responses while maintaining narrative relevance and coherence.

---

🎯 Purpose

The dataset aims to:

• Explore creative generation grounded in step-by-step reasoning.
• Provide training material for models that simulate internal thought before output.
• Serve as a benchmark or augmentation resource for CoT (Chain of Thought) based generation tasks.

---

⚠️ Limitations

• Prompts were inherited from a public Reddit-based dataset, and may include adult or NSFW content. Filtering is advised for sensitive applications.

---

💡 Use Cases

• Fine-tuning or evaluating models on reasoning-augmented creative generation.
• Developing storytelling models with logical structure.
• Research in narrative reasoning and step-by-step AI output planning.

---