Free LLM model On Qualitative & Quatitative Data

This model offers a free open source Large Language model applied on Qualitative & Quatitative Data

• Qualitative Data: Pre-trained google/flan-t5-large model from huggingface along with Meta/FAISS for similarity search. The prompt is applied on files in pdf format
• Quatitative Data: LLM Starcoder Model for code generation from huggingface that is pretained on license data, along with PandasAI to generate plot. The prompt is applied on files in csv format

Installation

Follow the installation guide

• Download and Unzip the files in one folder on laptop
• Install Python: https://www.python.org/downloads/
• Open Command Prompt
• Create virual environment: python -m venv env
• Used CD command to move to folder where all files are unzipped
• Install all packages: pip install -r requirements.txt
• Register and request a free API token from Huggingface: https://huggingface.co/settings/tokens
• Using notepad app, type `HUGGINGFACEHUB_API_TOKEN = "your_API_Token", then save as .env file in your unzip folder with all the source files

Running the Application

To run the application, use the following command:

streamlit run main.py

Reference

• PandasAI Starcoder
• google/flan-t5-large Model
• Streamlit File Uploader