tensorflow中tf.train.init_from_checkpoint的使用

比如用下面代码保存一个模型：

import tensorflow as tf
import numpy as np

input = np.random.random([10, 2])
print(input)

output = tf.layers.dense(input, units=2)

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    print(sess.run(output))

    for n in tf.global_variables():
        print(n)
        print(sess.run(n))

    saver = tf.train.Saver()
    saver.save(sess, "models/test")

使用tf.train.init_from_checkpoint将模型恢复出来：

import tensorflow as tf
import numpy as np

input = tf.placeholder(dtype=tf.float64, shape=[None, 2])

with tf.variable_scope("inference"):
    output = tf.layers.dense(input, units=2)
    output = tf.layers.dense(output, 1, use_bias=False)

assignment_map = {"dense/kernel": "inference/dense/kernel"}
assignment_map["dense/bias"] = "inference/dense/bias"

with tf.Session() as sess:
    tf.train.init_from_checkpoint("models", assignment_map)
    sess.run(tf.global_variables_initializer())
    for n in tf.global_variables():
        print(n)
        print(sess.run(n))