lipichandrakar/Prompt-Framing-Bias-LLM
🧠 Prompt Framing and Socio-Demographic Bias in LLMs Dataset by: Lipi ChandrakarMSc Data Science and Analytics, University of Hertfordshire (2025) 📄 Associated Paper More Responses, More Harm: How Prompt Framing and Decoding Choices Shape Socio-Demographic Bias in Open-Source LLMs📎 https://doi.org/10.5281/zenodo.20169051 If you use this dataset please cite:Chandrakar, L. (2025). More Responses, More Harm. Zenodo. https://doi.org/10.5281/zenodo.20169051… See the full description on the dataset page: https://huggingface.co/datasets/lipichandrakar/Prompt-Framing-Bias-LLM.
🧠 Prompt Framing and Socio-Demographic Bias in LLMs
Dataset by: Lipi Chandrakar MSc Data Science and Analytics, University of Hertfordshire (2025)
📄 Associated Paper
More Responses, More Harm: How Prompt Framing and Decoding Choices Shape Socio-Demographic Bias in Open-Source LLMs 📎 https://doi.org/10.5281/zenodo.20169051
If you use this dataset please cite: Chandrakar, L. (2025). More Responses, More Harm. Zenodo. https://doi.org/10.5281/zenodo.20169051
📘 Overview
This dataset investigates how prompt framing and decoding strategy influence socio-demographic bias in LLM outputs. It contains ~5,000 annotated prompt-response pairs with toxicity, offensiveness, and identity bias scores across five demographic categories.
Key finding: naive multi-sampling (k=3) amplifies bias rather than reducing it — race-related scores increased tenfold compared to single-output generation.
📂 Dataset Structure
📁 Files Included
🧩 Usage
from datasets import load_dataset
dataset = loaddataset( "lipichandrakar/Prompt-Framing-Bias-LLM" ) df = dataset["train"].topandas()
📊 Categories Covered
- Race
- Religion
- Gender
- Age
- Profession
🤖 Models Used
- GPT-2 (Radford et al., 2019)
- Qwen-3B (Bai et al., 2023)
⚖️ License
CC BY 4.0 — free to use with attribution