Tensorflow: SavedModelBuilder，如何以最佳验证精度保存模型

import tensorflow as tf
from tensorflow.python.saved_model import builder
from tensorflow.python.saved_model import tag_constants

export_dir = 'path/to/export/directory'
builder = builder.SavedModelBuilder(export_dir)

# 假设输入是一个形状为[batch_size, input_size]的张量
input_tensor = tf.placeholder(tf.float32, shape=[None, input_size], name='input_tensor')

# 假设输出是一个形状为[batch_size, num_classes]的张量
output_tensor = tf.placeholder(tf.float32, shape=[None, num_classes], name='output_tensor')

# 假设你的模型是一个简单的全连接神经网络
hidden_layer = tf.layers.dense(input_tensor, hidden_units, activation=tf.nn.relu)
output_layer = tf.layers.dense(hidden_layer, num_classes, activation=None)

loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=output_tensor, logits=output_layer))
optimizer = tf.train.AdamOptimizer(learning_rate)
train_op = optimizer.minimize(loss)

sess = tf.Session()
sess.run(tf.global_variables_initializer())

saver = tf.train.Saver()
saver.restore(sess, 'path/to/trained/weights')

inputs = {'input_tensor': tf.saved_model.utils.build_tensor_info(input_tensor)}
outputs = {'output_tensor': tf.saved_model.utils.build_tensor_info(output_tensor)}
signature_def = tf.saved_model.signature_def_utils.build_signature_def(
    inputs=inputs,
    outputs=outputs,
    method_name=tf.saved_model.signature_constants.PREDICT_METHOD_NAME
)
builder.add_meta_graph_and_variables(
    sess,
    [tf.saved_model.tag_constants.SERVING],
    signature_def_map={tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY: signature_def}
)

builder.save()