In-Browser LLM Chat

This application allows you to run large language models (LLMs) directly in your browser using Web-LLM.

Features

• Run LLMs completely in the browser - no server required
• Choose from multiple models
• Simple chat interface
• Conversation history (memory) within a session
• Progress display during model loading
• No API keys needed

Getting Started

1. Clone this repository

2. Serve the files with a local web server (due to CORS restrictions)

   # Using Python
   python -m http.server
   
   # Using Node.js
   npx serve

3. Open your browser and navigate to http://localhost:8000 (or the URL provided by your server)

4. Select a model and click "Load Model"

5. Wait for the model to download and initialize (this may take some time depending on the model size)

6. Start chatting!

Model Information

The application uses models from Web-LLM. The first time you use a model, it will be downloaded to your browser's cache. Subsequent uses will be faster as the model is loaded from the cache.

System Requirements

• A modern web browser (Chrome, Firefox, Safari, Edge)
• Sufficient RAM (8GB minimum recommended, more for larger models)
• For mobile devices, the smaller models are recommended

Notes

• The first model load may take some time as the model needs to be downloaded. Progress will be shown.
• All computation happens locally in your browser - your conversations are private
• Models are compressed to be efficient for browser usage
• Conversation history is maintained for the current session but is reset when changing models or reloading the page.