This repository contains one component of the various python repositories for NLIP Proof of Concept implementation. The various repositories are:
- nlip_sdk: this provides a software development kit in python that implements the abstractions of NLIP message, submessages along with a factory to ease the creation of NLIP messages and submessages
- nlip_client: this provides a package that simplifies the task of sending NLIP messages using the base underlying protocol. The current implementation uses httpx as the base transfer package.
- nlip_server: this provides a paclage that simplifies the task of writing a server-side NLIP application. This provides the abstractions of a NLIPApplication and a NLIP Session. An application consists of multiple sessions.
The above three are the components needed to write a client or a server. To write a client, you need to use nlip_sdk and nlip_client. To write a server side application, you need to use nlip_sdk and nlip_server.
The following repositories contain a few simple clients and server side applications:
- nlip_soln: this provides a few simple prototype server side solutions using the nlip_server package
- text_client: this provides a simple text based chatbot to interface with a NLIP server
- kivy_client: this provides a python kivy based visual client to interact with an NLIP server
Welcome to the NLIP Solutions! This project provides a few simple server side solutions that support the NLIP protocol.
The solutions included in the repository are:
- A simple stateless chatbot that responds to NLIP messages
- An integrator bot which polls the answer across several backend servers.
This project uses Poetry for dependency management. First, please install Poetry.
To set up the Python project, create a virtual environment using the following commands.
-
Create the virtual environment:
poetry env use python
-
Install the application dependencies
poetry install
Once the Python environment is set up, you can run the server.
Note: This solution assumes that you have an Ollama Server running and properly configured. You can see the directions for running the Ollama server at ollama.com .The default configuration is an ollama server running a the local machine using granite3-moe model.
The README in individual solutions describe how to configure the solution.
Start the chat server with:
poetry run start-chatTo start the chat server with a specific model (e.g., llama2):
CHAT_MODEL=llama2 poetry run start-chatStart the integration server with:
poetry run start-integrationStart the MCP server with:
poetry run start-mcp