vspagent npm package 🤖 AI-Powered Agent

Description

This npm package provides an AI-powered agent with information about Vishnu Suresh Perumbavoor. Powered by Qwen2.5-0.5B by Alibaba Cloud for intelligent conversations.

🆕 What's New in v2.0

• ✅ AI-Powered Conversations - Chat with an AI that knows all about VSP
• ✅ Qwen2.5-0.5B - Fast and efficient model by Alibaba Cloud (500M parameters)
• ✅ Lightweight - Only ~500MB model download
• ✅ Streaming Support - Real-time response streaming
• ✅ Backward Compatible - Original static data API still works
• ✅ Runs Locally - No API keys needed, runs on your machine via Transformers.js

Prerequisites

• Node.js 18+ (for ESM support in Transformers.js)
• 1-2GB RAM
• Internet connection (first run only, to download ~500MB model)

Installation

npm install vspagent

Quick Start

🎯 Interactive Chat via CLI (Recommended!)

After installing, chat directly with the AI agent from command line:

# Install globally
npm install -g vspagent

# Start chatting with the agent!
vspagent

Or install locally and use:

# Install in your project
npm install vspagent

# Run chat
npx vspagent
# or
npm run chat

This starts an interactive conversation with the AI agent where you can ask multiple questions!

Option 1: Static Data (Original Feature - Backward Compatible)

const vspagent = require('vspagent');

// Access static biodata
console.log(vspagent.name);        // "VSP Agent"
console.log(vspagent.creator);     // "Vishnu Suresh Perumbavoor"
console.log(vspagent.biodata);     // Full biodata object
console.log(vspagent.socials);     // Social media links

Option 2: AI-Powered Chat (New Feature 🚀)

const vspagent = require('vspagent');

async function chat() {
  // Initialize AI (first time downloads model ~500MB)
  await vspagent.initAI();
  
  // Ask questions about VSP
  const response = await vspagent.chat("Who is Vishnu Suresh Perumbavoor?");
  console.log(response);
  
  // Ask about accomplishments
  const response2 = await vspagent.chat("Tell me about VSP's achievements");
  console.log(response2);
  
  // Get social media info
  const response3 = await vspagent.chat("How can I connect with VSP?");
  console.log(response3);
}

chat();

Option 3: Interactive Continuous Chat (Best! 🎯)

# Run the interactive chat interface
node chat.js

This gives you a real continuous conversation:

🤖 VSP Agent - Interactive Chat Mode
============================================================
✅ Model loaded successfully!
============================================================

💬 You: Who is VSP?
🤖 VSP Agent: Vishnu Suresh Perumbavoor is an engineer, entrepreneur...

💬 You: What hackathons did he win?
🤖 VSP Agent: He won 3rd prize in Vaiga Agrihack 2023...

💬 You: Tell me more about his work
🤖 VSP Agent: He works at Trenser and has experience with...

💬 You: exit
👋 Thanks for chatting! Goodbye!

Option 4: Streaming Chat

const vspagent = require('vspagent');

async function streamChat() {
  await vspagent.initAI();
  
  // Streaming response (real-time output)
  await vspagent.chatStream("Tell me about VSP's hackathon achievements");
}

streamChat();

API Reference

Static Data (Original API)

All original properties are available:

vspagent.name              // "VSP Agent"
vspagent.creator           // "Vishnu Suresh Perumbavoor"
vspagent.founderOf         // "VSP dot AI"
vspagent.createdOn         // "28 April 2023"
vspagent.whoIsHe           // Array of roles
vspagent.interests         // Array of interests
vspagent.entertainments    // Array of entertainment preferences
vspagent.internships       // Array of internships
vspagent.placement         // Current placement
vspagent.accomplishments   // Array of accomplishments
vspagent.participations    // Array of event participations
vspagent.socials           // Object with social media links
vspagent.featured          // Featured media links
vspagent.biodata           // Complete biodata object

AI-Powered Methods (New)

initAI(options?)

Initialize the Qwen2.5-0.5B AI model.

Parameters:

• options (object, optional): Model options
  • dtype (string): Data type/quantization. Default: "q4"
  • device (string): Device to use. Default: "auto"

Returns: Promise

Example:

// Initialize with default settings
await vspagent.initAI();

// Custom options
await vspagent.initAI({
  dtype: "q4",
  device: "cpu"
});

chat(userMessage, options?)

Chat with the AI bot about VSP.

Parameters:

• userMessage (string): Your question or message
• options (object, optional): Generation options
  • max_new_tokens (number): Maximum tokens to generate. Default: 512
  • temperature (number): Sampling temperature. Default: 0.7
  • do_sample (boolean): Enable sampling. Default: false

Returns: Promise - AI response

Example:

const response = await vspagent.chat(
  "What are VSP's interests?",
  { max_new_tokens: 256, temperature: 0.8 }
);

chatStream(userMessage, options?)

Stream AI responses in real-time (outputs to console).

Parameters:

• userMessage (string): Your question
• options (object, optional): Same as chat()

Returns: Promise

Example:

await vspagent.chatStream("Tell me about VSP's achievements");
// Output streams to console in real-time

getModelInfo()

Get information about the AI model being used.

Returns: Object with model details

Example:

const modelInfo = vspagent.getModelInfo();
console.log(modelInfo.name);        // "Qwen2.5-0.5B-Instruct"
console.log(modelInfo.size);        // "0.5B parameters"
console.log(modelInfo.provider);    // "Alibaba Cloud - Qwen Team"

AI Model

Qwen2.5-0.5B-Instruct by Alibaba Cloud

| Feature | Details | |---------|---------| | Size | 0.5B parameters (~500MB download) | | Speed | ⚡⚡⚡⚡ Very Fast | | Quality | ⭐⭐⭐ Good | | RAM | 1-2GB | | Provider | Alibaba Cloud - Qwen Team | | Model ID | onnx-community/Qwen2.5-0.5B-Instruct |

Why Qwen2.5-0.5B?

• ✅ Lightweight and fast
• ✅ Great for conversational AI
• ✅ Excellent multilingual support (English & Chinese)
• ✅ Low resource requirements
• ✅ Optimized with ONNX and quantization

Examples

Basic Usage

const vspagent = require('vspagent');

// Check if AI is enabled
console.log('AI Enabled:', vspagent.aiEnabled);    // true
console.log('Version:', vspagent.version);         // "2.0.0"

// Use static data
console.log('Creator:', vspagent.creator);
console.log('LinkedIn:', vspagent.socials.linkedin);

AI Chat Examples

const vspagent = require('vspagent');

async function examples() {
  // Initialize once
  await vspagent.initAI();
  
  // Example 1: General info
  const resp1 = await vspagent.chat("Who is VSP?");
  
  // Example 2: Specific questions
  const resp2 = await vspagent.chat("What hackathons did VSP participate in?");
  
  // Example 3: Social media
  const resp3 = await vspagent.chat("Give me VSP's social media links");
  
  // Example 4: Interests
  const resp4 = await vspagent.chat("What does VSP like to do?");
}

examples();

Interactive Chat via Terminal

# If installed globally
vspagent

# If installed locally
npx vspagent

# Or using npm script
npm run chat

This opens an interactive chat interface where you can have continuous conversations with the AI agent!

Performance Tips

1. First Run: Model downloads on first use (~500MB). Subsequent runs are instant.
2. Memory: Uses ~1-2GB RAM during inference
3. Caching: Model is initialized once and cached for subsequent calls
4. Storage: Model cached in ~/.cache/huggingface/ (~500MB disk space needed)

How It Works

1. Transformers.js: Uses HuggingFace Transformers.js for Node.js
2. ONNX Runtime: Model runs via ONNX Runtime (optimized inference)
3. Local Execution: Everything runs locally, no API calls needed
4. Qwen2.5: Latest model from Alibaba Cloud's Qwen team
5. Multilingual: Supports both English and Chinese natively
6. Quantization: Q4 quantization for faster inference and smaller size

Troubleshooting

Error: Module not found

# Make sure transformers is installed
npm install @huggingface/transformers

Model download fails

• Check internet connection
• Model downloads to ~/.cache/huggingface/ by default
• Need ~500MB free disk space

Out of memory

• Close other memory-intensive applications
• Reduce max_new_tokens in chat options (try 128 or 256)
• Restart Node.js process

Backward Compatibility

✅ All original v1.x features work unchanged:

const vspagent = require('vspagent');

// v1.x code still works perfectly
console.log(vspagent);              // Shows all properties
console.log(vspagent.name);         // "VSP Agent"
console.log(vspagent.creator);      // "Vishnu Suresh Perumbavoor"
console.log(vspagent.socials);      // Social links object

Version History

v2.0.0 (Latest) - AI-Powered Edition

• Added AI chat capabilities powered by Qwen2.5-0.5B
• Fast and lightweight model (500M parameters)
• Streaming responses support
• Transformers.js integration
• Local execution (no API keys)

v1.0.x

• Original static biodata API
• Basic information export

Technical Details

• Framework: Transformers.js v3.3.0+
• Model: Qwen2.5-0.5B-Instruct (ONNX format)
• Quantization: Q4 (4-bit quantization for efficiency)
• Runtime: ONNX Runtime
• Languages: English, Chinese (multilingual support)
• Model Size: ~500MB download
• Memory Usage: 1-2GB RAM during inference

Contact

For any inquiries or support, you can reach out to Vishnu Suresh Perumbavoor

License

ISC

Credits

• Created by: Vishnu Suresh Perumbavoor
• AI Model: Qwen2.5-0.5B by Alibaba Cloud
• Framework: HuggingFace Transformers.js
• ONNX Models: onnx-community

🚀 Powered by Qwen2.5-0.5B - Alibaba Cloud's Latest AI Model