为什么这个TensorFlow图形与代码不匹配？

Question

以下是一些hello world level tf代码：

import tensorflow as tf
import numpy as np

def model():
    layer1 = layer('layer1', None)
    output_layer = layer('output_layer', layer1)
    print(tf.get_variable_scope().name) # print 1
    return output_layer

def layer(scope_name, x):
    with tf.variable_scope(scope_name):
        print(tf.get_variable_scope().name) # print 2
        w = tf.get_variable('w', shape=(1,), dtype=tf.float32,
                            initializer=tf.truncated_normal_initializer(mean=0, stddev=1))
        if x is not None:
            return w + x
        else:
            return w

with tf.Graph().as_default():
    with tf.variable_scope('model') as scope:
        prediction_1 = model()
        scope.reuse_variables()
        prediction_2 = model()
    with tf.Session() as sess:
        summary_writer = tf.train.SummaryWriter('log', graph=sess.graph)
        init_op = tf.initialize_all_variables()
        sess.run(init_op)
        [p1, p2] = sess.run([prediction_1, prediction_2])
        print('p1={} p2={}'.format(p1, p2))
        summary_writer.close()

上述代码中的print 1和print 2会产生：

model/layer1
model/output_layer
model
model/layer1
model/output_layer
model

这是我的预期，但它产生的图表：

model's graph

中有两个output_layer，为什么会有output_layer_1？