def from_config(cls, config, custom_objects=None):
"""Creates a RNNModel from its config.
Args:
config: A Python dictionary, typically the output of `get_config`.
custom_objects: Optional dictionary mapping names (strings) to custom
classes or functions to be considered during deserialization.
Returns:
A RNNModel.
"""
rnn_layer = keras_layers.deserialize(
config.pop('rnn_layer'), custom_objects=custom_objects)
sequence_feature_columns = fc.deserialize_feature_columns(
config.pop('sequence_feature_columns'), custom_objects=custom_objects)
context_feature_columns = config.pop('context_feature_columns', None)
if context_feature_columns:
context_feature_columns = fc.deserialize_feature_columns(
context_feature_columns, custom_objects=custom_objects)
activation = activations.deserialize(
config.pop('activation', None), custom_objects=custom_objects)
return cls(
rnn_layer=rnn_layer, sequence_feature_columns=sequence_feature_columns,
context_feature_columns=context_feature_columns, activation=activation,
**config)
def from_config(cls, config, custom_objects=None):
from tensorflow.python.feature_column.feature_column_lib import deserialize_feature_columns
config_cp = config.copy()
config_cp["columns"] = deserialize_feature_columns(config["columns"])
config_cp["cross_columns"] = deserialize_feature_columns(
config["cross_columns"])
del config["columns"]
del config["cross_columns"]
return cls(config_cp, custom_objects=custom_objects)
def deserialize_feature_columns(feature_column_configs, custom_objects=None):
"""Deserializes dict of feature column configs.
Args:
feature_column_configs: (dict) A dict mapping feature names to Keras feature
column config, could be generated using `serialize_feature_columns`.
custom_objects: (dict) Optional dictionary mapping names to custom classes
or functions to be considered during deserialization.
Returns:
A dict mapping feature names to feature columns.
"""
if not feature_column_configs:
return {}
feature_columns = {}
sorted_fc_configs = sorted(six.iteritems(feature_column_configs))
sorted_names, sorted_configs = zip(*sorted_fc_configs)
sorted_feature_columns = fc.deserialize_feature_columns(
sorted_configs, custom_objects=custom_objects)
feature_columns = dict(zip(sorted_names, sorted_feature_columns))
return feature_columns