server: env_name: ${APP_ENV:mock} # This configuration allows you to use GPU for creating embeddings while avoiding loading LLM into vRAM llm: mode: mock embedding: mode: huggingface