Talk With Me

This project is python-based mini AI chat project for testing LLM thai language performance with chat interface built with streamlit. This project supports two methods for connecting to an LLM:

1. OpenRouter (using free tier with 50 requests per day limit)
2. Ollama (Local deployment only, cannot run on Streamlit Cloud)

Features

• Chat Interface: Streamlit UI for easy interaction with the LLM.
• Multiple Connection Methods:
  • OpenRouter: Access via OpenRouter API with a limit of 50 requests per day.
  • Ollama: Local connection for running the model.
• Conversation Memory: Talk with LLM base on your chat history until end message session
  • 2 ways to end conversation session
    1. press Reset session button on sidebar
    2. send message /quit
• Thai language assistant: Model will perform as Thai assitant and will answer with Thai language.

Demo

Requirements

• Open router API key
• Installed ollama on your device (for local test)

To run this app

• create secrets.toml under /.streamlit
• enter your key on secrets.toml

APP_TOKEN=YOUR_APP_TOKEN
OPENROUTER_API_KEY=YOUR_OPENROUTER_API_KEY
GOOGLE_API_KEY=YOUR_GOOGLE_API_KEY
TYPHOON_AI_API_KEY=YOUR_TYPHOON_AI_API_KEY
OPENROUTER_BASE_URL="https://openrouter.ai/api/v1"
OLLAMA_HOST="http://localhost:11434"

• run app.py

streamlit run app.py