python - 索引错误: list index out of range when save model in tensorflow

Question

有人可以帮助我吗？我使用tensorflow来训练LSTM网络。训练运行良好，但当我想保存模型时，出现以下错误。

Step 1, Minibatch Loss= 0.0146, Training Accuracy= 1.000
Step 1, Minibatch Loss= 0.0129, Training Accuracy= 1.000
Optimization Finished!
Traceback (most recent call last):
  File ".\lstm.py", line 169, in <module>
    save_path = saver.save(sess, "modelslstm/" + str(time.strftime("%d-%m-%Y-%H-%M-%S")) + ".ckpt")
  File "C:\Python35\lib\site-packages\tensorflow\python\client\session.py", line 1314, in __exit__
    self._default_graph_context_manager.__exit__(exec_type, exec_value, exec_tb)
  File "C:\Python35\lib\contextlib.py", line 66, in __exit__
    next(self.gen)
  File "C:\Python35\lib\site-packages\tensorflow\python\framework\ops.py", line 3815, in get_controller
    if self.stack[-1] is not default:
IndexError: list index out of range

我的代码:

with tf.Session() as sess:
    os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
    # from tensorflow.examples.tutorials.mnist import input_data
    # mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)
    # a,b = mnist.train.next_batch(5)
    # print(b)
    # Run the initializer
    sess.run(init)
    saver = tf.train.Saver()
    merged_summary_op = tf.summary.merge_all()
    writer = tf.summary.FileWriter("trainlstm", sess.graph)
    #print(str(data.train.num_examples))
    for step in range(1, training_steps+1):
        for batch_i in range(data.train.num_examples // batch_size):
            batch_x, batch_y,name = data.train.next_batch(batch_size)
            #hasil,cost = encode(batch_x[0][0],"models/25-09-2017-15-25-54.ckpt")
            temp = []
            for batchi in range(batch_size):
                temp2 = []
                for ti in range(timesteps):
                    hasil,cost = encode(batch_x[batchi][ti],"models/25-09-2017-15-25-54.ckpt")
                    hasil = np.reshape(hasil,[num_input])
                    temp2.append(hasil.copy())
                temp.append(temp2.copy())
            batch_x = temp
            # Reshape data to get 28 seq of 28 elements
            #batch_x = batch_x.reshape((batch_size, timesteps, num_input))
            #dlib.hit_enter_to_continue()
            # Run optimization op (backprop)
            sess.run(train_op, feed_dict={X: batch_x, Y: batch_y})
            # Calculate batch loss and accuracy
            loss, acc = sess.run([loss_op, accuracy], feed_dict={X: batch_x,
                                                                 Y: batch_y})
            print("Step " + str(step) + ", Minibatch Loss= " + \
                  "{:.4f}".format(loss) + ", Training Accuracy= " + \
                  "{:.3f}".format(acc))
            f.write("Step " + str(step) + ", Minibatch Loss= " + \
                  "{:.4f}".format(loss) + ", Training Accuracy= " + \
                  "{:.3f}".format(acc)+"\n")

    print("Optimization Finished!")
    save_path = saver.save(sess, "modelslstm/" + str(time.strftime("%d-%m-%Y-%H-%M-%S")) + ".ckpt")
f.close()

我添加了 tf.reset_default_graph() 但它不起作用。请帮助我解决我的问题。谢谢!

Answer 1

是否必须使用上下文管理器(第 1 行的 with 语句？)。上下文管理器似乎很难销毁您的对象。这可能是 __exit__ 内置的问题。建议您向开发人员提交错误报告。