INNER CODE UNIT · Python

chat_completions

kaito-project/kaito · presets/ragengine/main.py:387

async def chat_completions(request: dict):
    start_time = time.perf_counter()
    status = STATUS_FAILURE  # Default status
    try:
        # Check if InferenceService is configured via environment variable
        if not os.getenv("LLM_INFERENCE_URL"):
            raise HTTPException(
                status_code=503,
                detail="InferenceService not configured. This RAGEngine instance only supports document retrieve via /retrieve API. To use chat completions, configure an InferenceService in the RAGEngine spec.",
            )

        guardrails = guardrails_reloader.get_current()
        if request.get("tools") or request.get("functions"):
            raise HTTPException(
                status_code=400,
                detail="tools and functions are not supported.",
            )
        if request.get("n", 1) > 1:

View source record →