QUICKSTART.md

🚀 Quick Start Guide

Get the Financial QA System up and running in minutes!

⚡ Quick Setup (5 minutes)

1. Install Dependencies

pip install -r requirements.txt

2. Test the System

python test_system.py

3. Start the Web Interface

python main.py interface

4. Open your browser and start asking questions!

🔧 What Each Command Does

Command	Purpose	Time
python test_system.py	Verify all components work	2-3 min
python main.py interface	Launch web app	30 sec
python main.py data	Process documents only	1-2 min
python main.py rag	Test RAG system	2-3 min
python main.py fine-tune	Test fine-tuning	5-10 min
python main.py evaluate	Run full evaluation	10-15 min
python main.py all	Complete pipeline	15-20 min

🎯 First Questions to Try

1. "What type of company is this?"
2. "What was the company's revenue in 2024?"
3. "What are the total assets?"
4. "What are the main business segments?"

🆘 Troubleshooting

Common Issues

"Module not found" errors:

# Make sure you're in the project root directory
cd financial-qa-system
# Install requirements again
pip install -r requirements.txt

CUDA/GPU issues:

# The system works on CPU, just slower
# For GPU acceleration, install PyTorch with CUDA support
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

Memory issues:

# Reduce batch size in fine-tuning
# Edit src/fine_tune_system.py, change batch_size to 1

PDF processing errors:

# Install system dependencies
# Ubuntu/Debian: sudo apt-get install tesseract-ocr
# macOS: brew install tesseract
# Windows: Download from https://github.com/UB-Mannheim/tesseract/wiki

📊 Expected Results

After running the test, you should see:

• ✅ Documents processed: 3
• ✅ Q&A pairs generated: 10-15
• ✅ Text chunks created: 20-30
• ✅ All systems ready

🎉 Success!

If you see all green checkmarks, you're ready to:

1. Ask questions in the web interface
2. Compare RAG vs Fine-tuning performance
3. Run evaluations to see detailed metrics
4. Explore the code to understand how it works

🔍 Next Steps

• Read the full README.md for detailed documentation
• Try different question types to see system behavior
• Run comprehensive evaluation to see performance metrics
• Modify parameters to experiment with different settings

💡 Pro Tips

• Start with simple questions to verify the system works
• Use the web interface for interactive exploration
• Check the logs if something goes wrong
• GPU acceleration makes fine-tuning much faster
• Smaller models work faster but may be less accurate