LLM chat implementation
A chat web component interface made using vanilla js, bootstrap and font-awesome css. The repo also contains an example usage with ollama.
Steps to execute
- Install ollama from https://ollama.com/
- Pull llama3 model with
ollama pull llama3
- Install chromadb from https://www.trychroma.com/
- Place .txt, .docx and .pdf contents in docs/ folder
- On root folder of this repo, run
npm install
npm start
- Launch http://localhost:3000
Now, all questions asked will be replied based on the contents in the docs folder. If no context is available from file contents, the model weights will be used.
Reference used : https://github.com/vteam27/RagXOllama
Flaticons have been used in this project
User icons created by Freepik - Flaticon
Virtual assistant
icons created by Job Moon - Flaticon