LLM Chat Service

This is an LLM Chat for corporate users. The purpose is to act as a proxy for different LMM model APIs like OpenAI's GPT, Llama2, Claude etc. Providing a unified experience for the organization users, and also home grown features like monitoring, load-balancing, authorization and more.

The core of this project is a web service using FastAPI and later we will provide a chat UI for the user.

Note that we are service the APIs over: http://<hostname>:<port>/api/v1/....

Getting started:

• Enable virtual environment using poetry: poetry shell
• Install packages: poetry install
• If you want to install 3rd party libraries, run: poetry add <pacakge> (e.g. poetry add openai)
• To use openai api, add api key enviromental veriable: OPENAI_API_KEY
• Create a .env file in the root folder and set OPENAI_API_KEY env var with the key value.

Running Tests

• Run ut: pytest tests/unit
• Run e2e: pytest tests/e2e

Running tools:

• Run pylint: ./run_pylint.sh

Running the service

In your console - export the `OPENAI_API_KEY`` env var with the key value.

Running the demo

• Run the service in one terminal: uvicorn app.main:app --reload
• Run the demo in another terminal: streamlit run demo/demo.py

Dev mode:

• Run this: uvicorn app.main:app --reload

• Access the /simple api: http://127.0.0.1:8000/api/v1/simple

"Prod" mode (no reload):

uvicorn app.main:app

Debugging the service

Run main.py and add breakpoints

Adding logs to the files

Add this import code line at the beginning of the file: from app.internal import infra_logger

Methods to use the loggers:

Info logger: infra_logger.log_info("Insert your text here")

Error logger: infra_logger.log_error("Insert your text here")

Warning logger: infra_logger.log_warning("Insert your text here")

Debug logger: infra_logger.log_debug("Insert your text here")

Project structure

To understand the diff when structuring a "big" application, read the following: https://fastapi.tiangolo.com/tutorial/bigger-applications/