emanuelepsilon/rag-agent-evaluation-workbench

1---2title: RAG Agent Evaluation Workbench3emoji: "🔎"4colorFrom: blue5colorTo: green6sdk: gradio7sdk_version: 6.20.08python_version: "3.13"9app_file: app.py10pinned: false11short_description: Grounded RAG answers with citations and retrieval traces.12---13 14# RAG Agent Evaluation Workbench15 16A compact Retrieval-Augmented Generation application backed by NVIDIA Nemotron through OpenRouter. It retrieves evidence from a local knowledge base, produces cited answers and exposes the source ranking behind each response.17 18The included knowledge base covers RAG reliability, evaluation metrics, agent observability and edge AI optimization.19 20## Configuration21 22The Space expects these settings:23 24- Secret: `OPENROUTER_API_KEY`25- Variable: `OPENROUTER_MODEL=nvidia/nemotron-3-super-120b-a12b:free`26 27Only public demonstration material is included in this Space.28