Nawa GPT-Codebase

1. Create conda env :

conda create -n gpt-orch python=3.11

2. Activate env :

conda activate gpt-orch

3. Installing dependencies.

There are two options, using pip or poetry:

• For the long term project, it is preferable to use poetry, it is lock your dependencies to lower the chances you have dependencies conflict
• But, if project needs to be done short in time use pip instead.

Click in the link, to install Poetry: https://python-poetry.org/docs/

3.a. Install All Dependencies using PIP:

pip install -r requirements.txt

3.b. Install All Dependencies using Poetry :

poetry install

4. How to run :

1. Set up Additional resource.
   • PostgresDB with vector extensions for vectorstore
   • Redis for History Chat Engine
2. Set up your FAQ list in PG DB.
   • Checkout Databank_Ingestion/Databank-create.ipynb.
   • Store the data that want to be ingested in FAQ_Data.
3. Set up Open AI key and additional resource in .env files.
   • change the env file to .env.
   • Fill out all the necessary path and openai key.
4. Run FastAPI locally.

   make run

List of available endpoints

API Service	Method	Full Path
GPT Completions	POST	{Endpoint}:8080/conversation
Delete History	POST	{Endpoint}:8080/clear-history