This repository contains a Python script that utilizes the SFR-Embedding-Mistral model from Salesforce to calculate the diversity of opinions in a set of text statements. The script generates embeddings for each statement and applies a diversity measurement based on the Jensen-Shannon Divergence.
Before running the script, ensure you have the following installed:
- Python 3.6 or higher
- PyTorch
- Hugging Face's Transformers library
You can install the required libraries using pip:
pip install torch transformers
- Clone this repository to your local machine using:
git clone https://github.com/johannescastner/CognitiveDiversity.git
- Navigate to the repository directory:
cd your-repository-name
- Run the script:
python diversity_measurement.py
The script will output the "Diversity of opinions" value based on the predefined statements within the script.
To customize the statements, modify the statements list in the diversity_measurement.py script. Replace the existing statements with your own, ensuring each statement reflects a unique opinion or perspective.
The output "Diversity of opinions" is a numerical value that represents the diversity of the provided statements based on their semantic content. Higher values indicate greater diversity.
Contributions to this project are welcome! Feel free to fork the repository, make your changes, and submit a pull request.
This project is licensed under the MIT License - see the LICENSE file for details.