tf.Graph

tf.Graph.control_dependencies(control_inputs)

用在上下文管理器中,参数control_inputs是op或tensor的list,其运行先于上下文管理器中的op或tensor。也可以为None,此时取消了控制依赖关系。

with tf.Session() as sess:
    a = tf.Variable(0)
    b = tf.Variable(0)
    c = tf.add(a, b)
    tf.global_variables_initializer().run()
    with tf.control_dependencies([c]):
        a = tf.assign(a, 5)
        b = tf.assign(b, 3)
        print(sess.run([a, b, c]))
        # =>[5, 3, 0]
        with tf.control_dependencies([c]):
            a = tf.assign(a, 0)
            print(sess.run([a, b, c]))
            # =>[0, 3, 8]

你可能感兴趣的:(tensorflow)