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.
First, you must clone the repo.
npm install || bun installollama pull phi4-mini
ollama pull nomic-embed-textThe project's model can be changed in
llm/model.ts.phi4-miniis 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.
npm run ingestThe file llm/ragVectors.json will be automatically generated.
npm run dev || bun devThe app will then be available on http://localhost:8080