GraniteFlowstateHyperparams

granite_flowstate_model.GraniteFlowstateHyperparams · inherits PydanticBaseModel

(No description provided in source.)

GraniteFlowstateModel

granite_flowstate_model.GraniteFlowstateModel · inherits BaseModel

IBM Granite FlowState foundation model wrapper.

Zero-shot stochastic forecasting via the granite-tsfm package. FlowState outputs 9 quantile forecasts (shape: batch, quantiles, horizon, channels) which are converted to pseudo-samples for stochastic metrics.

__init__(self, params, settings)

Initializes the Granite FlowState model wrapper.

Parameter	Type	Default	Description
params	Dict[str, Any]	-	(undocumented)
settings	Dict[str, Any]	-	(undocumented)

Raises: ImportError if granite-tsfm>=0.3.0 is not installed.

train(self, y_context, y_target, timestamps_context, timestamps_target, x_context=None, x_target=None, **kwargs) -> GraniteFlowstateModel

Loads the pretrained FlowState model.

Parameter	Type	Default	Description
y_context	np.ndarray	-	(undocumented)
y_target	np.ndarray	-	(undocumented)
timestamps_context	np.ndarray	-	(undocumented)
timestamps_target	np.ndarray	-	(undocumented)
x_context	Optional[np.ndarray]	None	(undocumented)
x_target	Optional[np.ndarray]	None	(undocumented)
**kwargs	-	-	(undocumented)

Returns: Self (the fitted forecaster; sets self.is_fitted = True).

predict(self, y_context, timestamps_context, timestamps_target, x_context=None, x_target=None, **kwargs) -> np.ndarray

Generates stochastic forecasts using the loaded FlowState model.

Parameter	Type	Default	Description
y_context	np.ndarray	-	(undocumented)
timestamps_context	np.ndarray	-	(undocumented)
timestamps_target	np.ndarray	-	(undocumented)
x_context	Optional[np.ndarray]	None	(undocumented)
x_target	Optional[np.ndarray]	None	(undocumented)
**kwargs	-	-	(undocumented)

Returns: np.ndarray (Forecast samples of shape (num_samples, forecast_horizon, num_targets)). Raises: ValueError if train() has not been called.