vykhovanets / RAG-playground

Local-first web app for experimentation with LLMs

RAG-playground

Iteration #1

The goal of the first iteration is to have local application which will make API calls to LLM providers

Currently supports Ollama's Mistral and OpenAI, uses Chroma as a vector storage. Implemented basic RAG capabilities.

Instructions

Setup

1. clone repository

https://github.com/vykhovanets/RAG-playground.git && cd RAG-playground

2. add .env file into RAG-playground folder with the following content:

# API keys
OPENAI_API_KEY=...
# COHERE_API_KEY=...
# ANTHROPIC_API_KEY=...
# HF_API_KEY=...

# Persistence
PROJECTS_DIR='./data/projects'
HISTORIES_DIR='./data/histories'
DB='./data/db'

3. install dependencies

python3.12 -m venv .envs/py-12 && source .envs/py-12/bin/activate
pip install uv && uv pip install -r requirements.txt

4. run app from the virtual environment

source .envs/py-12/bin/activate
streamlit run playground/main.py

Documents

• Stages of RAG
• Web app with Streamlit
• Chroma retrieval
• Python stuff