CX ChatGPT

This is the custom package to run ChatGPT as endpoints with transfer learning techniques.

• Train own organization data to match business needs and feed it to knowledge base in the form of GPTSimpleVectorIndex.
• Retrieve data from one's organizations' data combined with ChatGPT-3.5-turbo and response back to user queries.

Installation

This project is using Python version 3.11.4 with PIP requirements as in "requirements.txt" file.

• Check current working directory command:
  • pwd
• Check python version command:
  • python --version
• Get virtual environment command:
  • python -m venv .cx
• Activate environment command:
  • source .cx/bin/activate
• Recheck python version and libraries command:
  • python --version && pip freeze
• Install requirements command:
  • python -m pip install -r requirements.txt

Now we need to create a knowledge base for the transferred learning. This Python main script is to read Excel file and write to firestore. Then, a knowledge base created under a folder name "storage" to contain all store files.

• command: python main.py

After created a knowledge base, we can run the script to listen for request from the user with uvicorn library.

• command: uvicorn main:app

Docker Setup