|View source on GitHub|
A time series head which exports a single stateless serving signature.
The serving default signature exported by this head expects
and any exogenous features, but no state.
values has shape
filter_length, num_features] and
times has shape
total_length > filter_length. Any exogenous features
must have their shapes prefixed by the shape of the
When serving, first performs filtering on the series up to
starting from the default start state for the model, then computes predictions
on the remainder of the series, returning them.
Model state is neither accepted nor returned, so filtering must be performed each time predictions are requested when using this head.
__init__( model, state_manager, optimizer, input_statistics_generator=None, name=None )
_Head for time series regression.
model: A model for time series regression.
state_manager: A state manager.
optimizer: An optimizer.
input_statistics_generator: A input statistics generator.
name: An optional name for the model.
create_estimator_spec( features, mode, labels=None )
Performs basic error checking and returns an EstimatorSpec.
create_loss( features, mode, logits=None, labels=None )