¡Reserva! Google I / O regresa del 18 al 20 de mayo Regístrese ahora

tf.estimator.export.build_raw_serving_input_receiver_fn

Build a serving_input_receiver_fn expecting feature Tensors.

Creates an serving_input_receiver_fn that expects all features to be fed directly.

features a dict of string to Tensor.
default_batch_size the number of query examples expected per batch. Leave unset for variable batch size (recommended).

A serving_input_receiver_fn.