🤗 Models | 📊 Dataset | 📄 Paper

📢 The paper associated with this dataset has been accepted to the AAAI-26 Workshop on Personalization in the Era of Large Foundation Models (PerFM).

Human-Like-DPO-Dataset

This dataset was created as part of research aimed at improving conversational fluency and engagement in large language models. It is suitable for formats like Direct Preference Optimization (DPO) to guide models toward generating more human-like responses.

The dataset includes 10,884 samples across 256 topics, including:

• Technology
• Daily Life
• Science
• History
• Arts

Each sample contains:

• Conversational Question: Natural, engaging questions that reflect everyday human dialogue.
• Human-Like Response: A natural, conversational answer generated to mimic human interaction.
• Formal Response: A structured, professional answer reflecting traditional AI responses.

Dataset Usage

This dataset can be used to fine-tune LLMs to:

• Improve conversational coherence.
• Reduce mechanical or impersonal responses.
• Enhance emotional intelligence in dialogue systems.

More details on dataset creation and usage can be found in the accompanying research paper.