Spaces:
Running
Running
| title: Technical RAG Assistant | |
| emoji: 🔍 | |
| colorFrom: blue | |
| colorTo: purple | |
| sdk: docker | |
| app_port: 8501 | |
| pinned: false | |
| # Technical Documentation RAG Assistant | |
| Advanced RAG system with local Ollama LLM integration running in HuggingFace Spaces. | |
| ## Features | |
| - 🦙 **Local Ollama LLM** - No API dependencies | |
| - 🔍 **Hybrid Search** - Semantic + keyword retrieval | |
| - 📚 **Citation Support** - Automatic source attribution | |
| - 🎯 **Confidence Scoring** - Answer quality assessment | |
| - ⚡ **Fast Performance** - Local inference | |
| ## Models Used | |
| - **LLM**: Llama 3.2 3B (via Ollama) | |
| - **Embeddings**: all-MiniLM-L6-v2 | |
| - **Fallback**: HuggingFace API if Ollama fails | |
| ## How it Works | |
| 1. Upload PDF documents | |
| 2. Documents are chunked and indexed | |
| 3. Ask questions in natural language | |
| 4. Get answers with citations and confidence scores | |
| Built for technical documentation Q&A with Swiss engineering standards. |