📊 Enhanced RAG CSV Chatbot

📁 Step 1 - Configure and Initialize RAG Pipeline

🤖 Step 2 - Configure and Load LLM

💬 Step 3 - Chat with Your CSV Data

💡 Tips for Better Results

• Start simple: Ask basic questions first (e.g., "What columns are in this data?")
• Be specific: Ask about particular columns or values (e.g., "What are the highest sales values?")
• Request summaries: Ask for overviews (e.g., "Summarize the key patterns in this dataset")
• Compare data: Ask comparative questions (e.g., "Compare category A vs category B")
• Be patient: Local models may take 30-60 seconds to load and respond
• Check sources: Review the source references to understand what data informed the answer

🔧 Model Information

Available Models:

• Qwen2.5-7B-Instruct: Advanced Chinese-English bilingual model, excellent for analysis
• Llama-3.1-8B-Instruct: Meta's powerful instruction-following model

Note: Models are loaded locally with 4-bit quantization for memory efficiency. First load may take several minutes.

🔧 Troubleshooting

If you get errors:

• Wait for model loading to complete (check progress messages)
• Ensure sufficient GPU/RAM memory (models use ~4-6GB)
• Try simpler questions if responses are incomplete
• Make sure your CSV files have clear column headers
• Recreate the database if issues persist