Project's Goal

The Eliza project is an academic project aimed at developing a chatbot to address a problem of our choice.

To do this, we have two options:

• Build an LLM from scratch ourselves
• Use an existing LLM, fine-tune it ourselves, and run it locally.

We chose the second option for our “MédiRDV” project: a chatbot that acts as a medical appointment assistant, helping patients find the right healthcare provider.

Installation

Prerequisites

• Node.js v18+
• npm
• Ollama installed and running

Project

First, you must clone the repo.

npm install || bun install

Running Ollama

ollama pull phi4-mini
ollama pull nomic-embed-text

The project's model can be changed in llm/model.ts. phi4-mini is the default model.
If you want to use a model other than the default one, you must run Ollama with the corresponding model at the same time.

Ingesting PDFs

npm run ingest

The file llm/ragVectors.json will be automatically generated.

Then

npm run dev || bun dev

The app will then be available on http://localhost:8080