This project involves fine-tuning a GPT-2 model on a mental health FAQ dataset to generate text related to mental health. The fine-tuning process adapts the model to better understand and respond to queries in the domain of mental health.
Data/: This directory contains the dataset used to train and fine-tune the GPT-2 model.MentalHealthFAQ_FineTuner.ipynb: A Jupyter notebook that details the process of fine-tuning the GPT-2 model on the mental health FAQ dataset.data_exploratory_analysis.ipynb: A Jupyter notebook that provides an exploratory analysis of the mental health FAQ dataset, offering insights into the data used for training the chatbot model.streamlit_app.py: The Streamlit application script that can be used to run the chatbot interface, allowing users to ask questions and receive responses generated by the fine-tuned model.
-
Clone the Repository: First, clone this repository to your local machine using Git. Open a terminal or command prompt and run the following command:
git clone https://github.com/akbarpourmaryam/FineTunedGPT-2MentalHealthStreamlitApp)
-
Download the fine-tuned GPT-2 model: From the Releases section of this repository, download the fine-tuned GPT-2 model. Unzip the file and place it in the root directory of the cloned repository or update the streamlit_app.py file to point to the location where you have saved the model files.
-
Install Streamlit and Transformers:
pip install streamlit transformers
-
Run the Streamlit App:
streamlit run app.py
After running the Streamlit app, your web browser should automatically open the app, or you can navigate to the local URL provided in the terminal.
The app.py script expects the fine-tuned GPT-2 model to be located in a folder named my_model in the same directory as the script.
The data used in this project comes from the "Mental Health FAQ for Chatbot" dataset at the following URL: https://www.kaggle.com/datasets/narendrageek/mental-health-faq-for-chatbot/data