---
license: mit
task_categories:
- text-classification
language:
- en
---

## Paper

You can access the paper on here: [StereoDetect: Detecting Stereotypes and Anti-stereotypes the Correct Way Using Social Psychological Underpinnings](https://aclanthology.org/2025.findings-emnlp.216/)

This work is accepted at the Findings of the Association for Computational Linguistics: EMNLP 2025.

## Dataset Overview

The dataset is divided into three splits:
- **train.csv** — used for model training  
- **val.csv** — used for validation and hyperparameter tuning  
- **test.csv** — used for final evaluation  

### Label Definitions

| Label | Description |
|:------:|-------------|
| **0** | Anti-stereotype |
| **1** | Stereotype |
| **2** | Neutral without target term |
| **3** | Neutral with target term |
| **4** | Bias |

-----

### Contact
For queries related to this work, please contact:  
[kaustubhshejole@cse.iitb.ac.in](mailto:kaustubhshejole@cse.iitb.ac.in)