EmpowerMen is a Question-Answering (QA) application designed to address queries related to false accusations of rape against men, enriched through a knowledge base sourced from podcasts from {Journalist | Filmmaker | Social Activist} Deepika Bhardwaj Twitter. This project aims to provide accurate and informed responses using RAG based on 5 hours of content from three podcasts.

• URLs to podcast videos:
  • Video 1
  • Video 2
  • Video 3

To create the knowledge base:

Provide a list of URLs to podcast videos. Utilize the DatasetCreation.ipynb(Use google colab) notebook to generate .txt transcripts from the podcasts. These transcripts will serve as the foundational text data for the QA app.

AskQuery.py Use FastEmbed Embeddings from ChromaDB to convert the transcript text into embeddings. Store these embeddings in ChromaDB's vector store for efficient retrieval and similarity scoring.

app.py Develop a Streamlit-based user interface (UI) that allows users to input questions. Display responses generated by the QA model based on the knowledge base.

AskQuery.py Implement the Ollama framework integrated with langchain to facilitate the llama3 model. Utilize this model to analyze and generate responses based on the enriched knowledge base of podcast transcripts.

Note: Ollama should be installed on your system. Then run this command Ollama pull llama3.

1. Create a virtual environment python3 -m venv empowermen_venv in your preferred project directory using terminal.
2. Activate environment empowermen_venv\Scripts\Activate for windows, source empowermen_venv\bin\activate for mac.
3. Run pip3 install -r requirements.txt.
4. Run streamlit run app.py.