tensorflow 中dataset模块迭代器的用法

本文主要讲述tensorflow中迭代器的用法

import tensorflow as tf
dateset1 = tf.data.Dataset.from_tensor_slices(tf.random_uniform([4, 10], 0, 8))
print(dateset1.output_types)  #"tf.float32"
print(dateset1.output_shapes)  # (10,)
dataset = tf.data.Dataset.range(100)
iterator = dataset.make_one_shot_iterator()
next_element = iterator.get_next()

with tf.Session() as sess:
    for i in range(100):
        value = sess.run(next_element)
        assert i == value, "caca"
    else:
        print("guoyi")


max_value = tf.placeholder(tf.int64, shape=[])
# 定义的dataset有参数，只能使用参数化迭代器
dateset = tf.data.Dataset.range(max_value)
# 定义参数化迭代器
iterator = dateset.make_initializable_iterator()
next_element = iterator.get_next()

with tf.Session() as sess:
    # 需要用参数初始化迭代器
    sess.run(iterator.initializer, feed_dict={max_value: 10})
    for i in range(10):
        value = sess.run(next_element)
        assert i == value

with tf.Session() as sess:
    # 可以再次使用迭代器
    sess.run(iterator.initializer, feed_dict={max_value: 100})
    for i in range(100):
        value = sess.run(next_element)
        assert i == value