Tensorflow TypeError：Fetch argument None 的类型无效 NoneType？

如何解决Tensorflow TypeError：Fetch argument None 的类型无效 NoneType？

代码如下：

for epoch in range(nr_epochs):

    #================== Training Dataset ======================

    for i in range(nr_iterations):
        batch_x,batch_y = next_batch(batch_size=size_of_batch,data=x_train,labels=y_train)
        feed_dictionary = {X:batch_x,Y:batch_y}
        sess.run(train_step,feed_dict={X:batch_x,Y:batch_y})
    
    s,batch_accuracy = sess.run(fetches=[merged_summary,accuracy],feed_dict=feed_dictionary)

    train_writer.add_summary(s,epoch)    

    print(f'Epoch {epoch} \t | Training Accuracy = {batch_accuracy}')

    #=================== Validation ============================

    summary = sess.run(fetches = mergerd_summary,feed_dict = {X: x_val,Y: y_val})
    validation_writer.add_summary(summary,epoch)

print('Done Training!')

在 fetch 参数的第 10 行中获取 TypeError：

TypeError                                 Traceback (most recent call last)
<ipython-input-174-67a551b94b09> in <module>
      8         sess.run(train_step,Y:batch_y})
      9 
---> 10     s,feed_dict=feed_dictionary)
     11 
     12     train_writer.add_summary(s,epoch)

TypeError: Fetch argument None has invalid type <class 'NoneType'>

代码在第 8 行的 sess.run 的第一次迭代中完美运行，但在第二次迭代中抛出错误。