semantic_router.decision_models.systemone

decision models served over HTTP at /v1/systemone (TypeSafe's Jev, and servers that copy it).

SystemOneDecisionModel Objects

class SystemOneDecisionModel(DecisionModel)

any server that speaks /v1/systemone: Jev, Laya, Von, or one of your own.

Arguments:

• name: The model id sent with every request.
• base_url: The server; requests go to {base_url}/v1/systemone.
• api_key: Sent as a Bearer token. Read from api_key_env when not given; a server that needs no key gets no Authorization header.
• timeout: Seconds per request.
• retries: Retries on rate limits (429, 529), server errors and timeouts.
• concurrency: Requests in flight when answering many states at once. With one request per state, this caps a batch's throughput.
• session: An aiohttp session to use instead of the model's own (tests).

aask

async def aask(state: object, questions: dict[str,
                                              Question]) -> dict[str, Answer]

answer several typed questions about one state in a single request.

TypeSafeDecisionModel Objects

class TypeSafeDecisionModel(SystemOneDecisionModel)

TypeSafe's Jev over its API. Reads JEV_API_KEY when no key is passed.

Arguments:

• name: The Jev model to call.