Skip to content

Chat demo

The chat UI is a Gradio app driven by the same router configs as llmrouter infer.

Prereqs

  • Install gradio (the CLI prints an error if it is missing).
  • For full inference, set API_KEYS and make sure your llm_data has api_endpoint and model.

Run

llmrouter chat --router knnrouter --config configs/model_config_test/knnrouter.yaml --host 0.0.0.0 --port 7860

Key flags

  • --mode: full_context, current_only, or retrieval
  • --top_k: only used in retrieval mode
  • --temp: generation temperature

See the full flag list in CLI reference.