grums18/mySpace

🇮🇳 Tamil Study Buddy - Fine-tuned Model & Streamlit App

AI Tutor for Colloquial Tamil Learning - A fine-tuned language model trained on colloquial Tamil data with a beautiful Streamlit interface.

  

🎯 Project Overview

Problem: Tamil students and professionals cannot access world-class AI help in their native language.

Solution: Tamil Study Buddy - An AI tutor that explains any concept in Tamil with examples relevant to Tamil Nadu context.

Technology: Fine-tuned GPT-2 model trained on colloquial Tamil data.

✨ Features

✅ Fine-tuned Tamil Model

• —Base model: GPT-2 (124M parameters)
• —Trained on: Colloquial Tamil data (85+ examples)
• —Specialized for: Conversational Tamil

✅ Beautiful Streamlit Interface

• —Chat interface with history
• —Settings sidebar (temperature, response length, diversity)
• —Statistics dashboard
• —Example queries
• —Error handling

✅ Easy Deployment

• —Deploy to Hugging Face Spaces in 5 minutes
• —Free hosting (no credit card needed)
• —Auto-deploy from GitHub
• —Public sharing link

✅ Production Ready

• —Model caching for performance
• —Error handling
• —Responsive design
• —Mobile-friendly

📁 Repository Structure

tamil-study-buddy-repo/
├── streamlit_app.py                      # Main Streamlit application
├── requirements.txt                      # Python dependencies
├── README.md                             # This file
├── .streamlit/
│   └── config.toml                       # Streamlit configuration
├── tamil-study-buddy-finetuned/          # Fine-tuned model files
│   ├── config.json                       # Model config
│   ├── model.safetensors                 # Model weights
│   ├── tokenizer.json                    # Tokenizer
│   └── ...
├── HUGGINGFACE_SPACES_DEPLOYMENT.md      # Deployment guide
├── STREAMLIT_EXPLANATION.md              # Streamlit tutorial
└── MURUGA_FINETUNING_STRATEGY.md         # Fine-tuning details

🚀 Quick Start

Option 1: Deploy to Hugging Face Spaces (Recommended)

5-minute deployment:

1. 1.Go to: https://huggingface.co/spaces
2. 2.Click "Create new Space"
3. 3.Select "Streamlit" SDK
4. 4.Connect this GitHub repo
5. 5.Done! Auto-deployed

Your app will be live at:

https://huggingface.co/spaces/YOUR_USERNAME/tamil-study-buddy

Option 2: Run Locally

Requirements:

• —Python 3.8+
• —4GB RAM minimum

Installation:

# Clone repository
git clone https://github.com/grums18/Tamil_data_Models1.git
cd Tamil_data_Models1

# Install dependencies
pip install -r requirements.txt

# Run app
streamlit run streamlit_app.py

Access at: http://localhost:8501

📊 Model Details

Base Model

• —Architecture: GPT-2
• —Parameters: 124M
• —Framework: PyTorch
• —Tokenizer: GPT-2 Tokenizer

Fine-tuning Data

• —Source: Colloquial Tamil conversations
• —Size: 85 examples (55 train, 6 val, 8 test)
• —Categories: Greetings, Questions, Statements, Responses
• —Language: Colloquial Tamil (பேச்சு வழக்கு)

Performance

• —Inference Time: ~2 seconds per query
• —Model Size: 500MB
• —Memory: 2GB RAM minimum

💻 Usage Examples

Example 1: Greeting

Input: வணக்கம்
Output: வணக்கம், நான் தமிழ் கிக். நிமவ் ...

Example 2: Question

Input: நல்ல நாள்
Output: நல்ல நாள்ல வாள்க். ன்க். பற் நி...

Example 3: Learning

Input: தமிழ்நாடு
Output: தமிழ்நாடுக் நயக்க் வாற்க் வள் ...

🔧 Customization

Modify Model Parameters

Edit streamlit_app.py:

temperature = st.slider(
    "Temperature (Creativity)",
    min_value=0.0,
    max_value=1.0,
    value=0.7,  # Adjust this
    step=0.1
)

max_length = st.slider(
    "Max Response Length",
    min_value=50,
    max_value=500,
    value=150,  # Adjust this
    step=50
)

Change UI Theme

Edit .streamlit/config.toml:

[theme]
primaryColor = "#ff6b35"        # Change primary color
backgroundColor = "#ffffff"     # Change background
secondaryBackgroundColor = "#f0f2f6"
textColor = "#262730"

📚 Documentation

• —[HUGGINGFACE_SPACES_DEPLOYMENT.md](HUGGINGFACE_SPACES_DEPLOYMENT.md) - Complete deployment guide
• —[STREAMLIT_EXPLANATION.md](STREAMLIT_EXPLANATION.md) - Streamlit tutorial
• —[MURUGA_FINETUNING_STRATEGY.md](MURUGA_FINETUNING_STRATEGY.md) - Fine-tuning details

🎨 Building Modern UI/UX

This repository includes the fine-tuned model. You can build a modern frontend on top:

Frontend Options

1. 1.React.js
2. 2.Modern, responsive UI
3. 3.Component-based architecture
4. 4.Easy state management

1. 1.Next.js
2. 2.Full-stack framework
3. 3.API routes
4. 4.Deployment-ready

1. 1.Vue.js
2. 2.Progressive framework
3. 3.Lightweight
4. 4.Easy to learn

1. 1.Flutter
2. 2.Mobile-first
3. 3.Cross-platform
4. 4.Native performance

API Integration

Create a simple API endpoint to use the model:

from fastapi import FastAPI
from transformers import AutoModelForCausalLM, AutoTokenizer

app = FastAPI()

@app.post("/generate")
async def generate(query: str):
    # Load model
    model = AutoModelForCausalLM.from_pretrained("./tamil-study-buddy-finetuned")
    tokenizer = AutoTokenizer.from_pretrained("./tamil-study-buddy-finetuned")
    
    # Generate response
    inputs = tokenizer.encode(query, return_tensors="pt")
    outputs = model.generate(inputs, max_length=150)
    response = tokenizer.decode(outputs[0])
    
    return {"response": response}

🤝 Contributing

Contributions welcome! Please:

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

📝 License

MIT License - see LICENSE file for details

🙏 Acknowledgments

• —Base model: OpenAI GPT-2
• —Fine-tuning framework: Hugging Face Transformers
• —UI framework: Streamlit
• —Hosting: Hugging Face Spaces

📞 Support

• —Issues: GitHub Issues
• —Discussions: GitHub Discussions
• —Email: grums18@github.com

🎯 Roadmap

• —[ ] Add voice interface (Tamil speech-to-text)
• —[ ] Improve model with more training data
• —[ ] Create mobile app
• —[ ] Add multi-turn conversation memory
• —[ ] Implement user feedback loop
• —[ ] Create admin dashboard
• —[ ] Add analytics
• —[ ] Deploy to multiple platforms

📊 Statistics

Built with ❤️ for Tamil learners worldwide

🇮🇳 Tamil Study Buddy - Learn Tamil with AI! 🤖