深度学习之tensorflow框架（下）

def tensor_demo():
    """
    张量的演示
    :return:
    """
    tensor1 = tf.constant(4.0)
    tensor2 = tf.constant([1, 2, 3, 4])
    linear_squares = tf.constant([[4], [9], [16], [25]], dtype=tf.int32)
    print("tensor1:
", tensor1)
    print("tensor2:
", tensor2)
    print("linear_squares:
", linear_squares)

    # 生成常用张量
    tensor3 = tf.zeros(shape=(3, 4))
    print("tensor3:
", tensor3)
    tensor4 = tf.ones(shape=(2, 3, 4))
    print("tensor4:
", tensor4)
    tensor5 = tf.random_normal(shape=(2, 3), mean=1.75, stddev=0.2)
    print("tensor5:
", tensor5)

    with tf.compat.v1.Session() as sess:
        print("tensor3_value:
", tensor3.eval())
        print("tensor4_value:
", tensor4.eval())
        print("tensor4_value:
", tensor5.eval())

    return None


def tensoredit_demo():
    """
    张量类型的修改
    :return:
    """
    linear_squares = tf.constant([[4], [9], [16], [25]], dtype=tf.int32)
    print("linear_squares_before:
", linear_squares)

    l_cast = tf.cast(linear_squares, dtype=tf.float32)
    print("linear_squares_after:
", linear_squares)
    print("l_cast:
", l_cast)
    return None


def editstaticshape_demo():
    """
    更新/改变静态形状
    :return:
    """
    a = tf.compat.v1.placeholder(dtype=tf.float32, shape=[None, None])
    b = tf.compat.v1.placeholder(dtype=tf.float32, shape=[None, 10])
    c = tf.compat.v1.placeholder(dtype=tf.float32, shape=[3, 2])
    print("a:
", a)
    print("b:
", b)
    print("c:
", c)

    # 更新形状未确定的部分
    a.set_shape([2, 3])
    b.set_shape([2, 10])
    print("a:
", a)
    print("b:
", b)

    return None;

def editshape_demo():
    """
    更新/改变动态形状
    不会改变原始的tensor
    返回新的改变类型后的tensor
    :return:
    """
    a = tf.compat.v1.placeholder(dtype=tf.float32, shape=[None, None])
    print("a:
", a)
    a.set_shape([2, 3])
    print("a_setShape:
", a)
    # 元素个数没有变，还是2*3*1=6个
    a_reshape = tf.reshape(a,shape=[2,3,1])
    print("a_reshape:
", a_reshape)
    print("a:
", a)

    return None;

def variable_demo():
    """
    变量的演示
    变量需要显式初始化，才能运行值
    :return:
    """
    # 创建变量
    # 使用命名空间可以使图的结构更加清晰
    with tf.variable_scope("myscope"):
        a = tf.Variable(initial_value=50)
        b = tf.Variable(initial_value=40)
    with tf.variable_scope("yourscope"):
        c= tf.add(a,b)
    print("a:
",a)
    print("b:
",b)
    print("c:
",c)

    # 初始化变量
    init = tf.global_variables_initializer()

    # 开启会话
    with tf.Session() as sess:
        sess.run(init)
        a_value,b_value,c_value=sess.run([a,b,c])
        print("a_value:
",a_value)
        print("b_value:
",b_value)
        print("c_value:
",c_value)

    return None