🤖 LLM Chat Interface

A beautiful web interface for interacting with Large Language Models via Hugging Face's router API.

✨ Features

• 🎨 Beautiful UI: Responsive web interface with gradient design
• 🔄 Multiple Models: Support for various LLMs including Qwen, Llama, Mistral
• ⚙️ Parameter Control: Adjust temperature, max tokens, top-p, and penalties
• 📊 Detailed Responses: View metadata including token usage and model info
• 🚀 Real-time: Live loading states and error handling
• 🔒 Secure: Environment variables for API keys

🛠️ Models Supported

• Qwen/Qwen2.5-1.5B-Instruct:featherless-ai
• meta-llama/Meta-Llama-3-8B-Instruct
• microsoft/DialoGPT-medium
• mistralai/Mistral-7B-Instruct-v0.1
• HuggingFaceH4/zephyr-7b-beta

🚀 Quick Start

Prerequisites

• Node.js (v18 or higher)
• Hugging Face API token

Installation

1. Clone the repository

   git clone https://github.com/herbowicz/huggingface-llm-chat.git
   cd huggingface-llm-chat

2. Install dependencies

   npm install

3. Set up environment variables Create a .env file in the root directory:

   HF_TOKEN=your_hugging_face_token_here

4. Start the server

   npm start

5. Open your browser Navigate to http://localhost:3000

📁 Project Structure

├── index.html          # Web interface
├── server.mjs          # Express server
├── my-llm.mjs         # Original CLI script
├── test-llm.mjs       # Test script
├── package.json       # Dependencies
├── .env              # Environment variables (create this)
└── .gitignore        # Git ignore rules

🎛️ Parameters

• Temperature (0-2): Controls randomness in responses
  _{Lower values (0.1-0.3) make responses more focused and predictable, higher values (0.7-1.0) make them more creative and varied}

• Max Tokens (1-2048): Maximum length of response
  _{Tokens are pieces of words - roughly 1 token ≈ 0.75 words, so 100 tokens ≈ 75 words}

• Top P (0-1): Nucleus sampling parameter
  _{Controls diversity by considering only the most likely words that make up the top P% probability mass}

• Frequency Penalty (-1 to 1): Reduces repetition
  _{Positive values discourage the model from repeating the same words/phrases it already used}

• Presence Penalty (-1 to 1): Encourages topic diversity
  _{Positive values encourage the model to talk about new topics rather than staying on the same subject}

💡 Recommended Settings for Beginners

• General questions: Temperature: 0.7, Max Tokens: 150, Top P: 0.9, Penalties: 0.0
• Creative writing: Temperature: 0.9, Max Tokens: 300, Top P: 0.9, Penalties: 0.0
• Factual answers: Temperature: 0.3, Max Tokens: 100, Top P: 0.9, Penalties: 0.0
• Detailed explanations: Temperature: 0.5, Max Tokens: 500, Top P: 0.9, Penalties: 0.0

🔧 Configuration

Getting a Hugging Face Token

1. Go to Hugging Face Settings
2. Create a new token with "Read" and "Inference" permissions
3. Copy the token to your .env file

Environment Variables

HF_TOKEN=hf_your_token_here    # Required: Your Hugging Face API token
PORT=3000                      # Optional: Server port (default: 3000)

🖥️ Usage

Web Interface

1. Enter your question in the text area
2. Select a model from the dropdown
3. Adjust parameters if needed (defaults work well)
4. Click "Ask Question" and wait for the response

CLI Usage

For direct API testing:

node my-llm.mjs

🛡️ Security

• API tokens are stored in environment variables
• .env file is excluded from git
• No sensitive data in client-side code

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly
5. Submit a pull request

📄 License

MIT License - feel free to use this project however you'd like!

🔗 Links

• Hugging Face
• OpenAI API Documentation
• Express.js

---

Made with ❤️ by Grzegorz Herbowicz