def _export_model(self, estimator: tf.estimator.Estimator,
save_location: str) -> None:
"""
Used to export your model in a format that can be used with
Tf.Serving
:param estimator: your estimator function
"""
# this should match the input shape of your model
# TODO: update this to your input used in prediction/serving
x1 = tf.feature_column.numeric_column("input", shape=[480, 640, 1])
# create a list in case you have more than one input
feature_columns = [x1]
feature_spec = tf.feature_column.make_parse_example_spec(
feature_columns)
export_input_fn = tf.estimator.export.build_parsing_serving_input_receiver_fn(
feature_spec)
# export the saved model
estimator.export_savedmodel(save_location, export_input_fn)
def _export_model(self, estimator: tf.estimator.Estimator,
save_location: str) -> None:
"""
Used to export your model in a format that can be used with
Tf.Serving
:param estimator: your estimator function
"""
# this should match the input shape of your model
x1 = tf.feature_column.numeric_column(
"input",
shape=[
self.config["train_batch_size"], conf.c_maxnum, conf.c_maxlen
])
# create a list in case you have more than one input
## feature_columns = [x1]
feature_columns = x1
feature_spec = tf.feature_column.make_parse_example_spec(
feature_columns)
export_input_fn = tf.estimator.export.build_parsing_serving_input_receiver_fn(
feature_spec)
# export the saved model
estimator.export_savedmodel(save_location, export_input_fn)