Question

我正在尝试在Tensorflow中训练网络，该网络基于3个功能将网络流量分为2类。

这是我的数据集：

[[[80, 0.0, 1],
  [80, 10.00087022781372, 1],
  [80, 20.00172996520996, 1],
  [80, 30.002599954605103, 1],
  [80, 40.003480195999146, 1],
  [80, 50.00434994697571, 1]],
 [[80, 0.0, 1],
  [80, 10.00091004371643, 1],
  [80, 20.00171995162964, 1],
  [80, 30.00259017944336, 1],
  [80, 40.00345993041992, 1],
  [80, 50.00433015823364, 1]],
 [[372, 0.0, 1],
  [372, 10.000890016555786, 1],
  [372, 24.0006902217865, 1],
  [372, 45.0004301071167, 1]],...]

这是我的网络：

# Parameters
units_in_cell = 128
n_epochs = 3
learning_rate =0.001

batch_size = 1

n_features = 3
n_output = 1

#placeholder

x = tf.placeholder(tf.float32,[None , None, n_features], name = 'inputs')
y = tf.placeholder(tf.float32, name = 'labels')

#cells and initial state

cell = tf.contrib.rnn.OutputProjectionWrapper(tf.contrib.rnn.BasicLSTMCell(num_units=units_in_cell, activation=tf.nn.relu), output_size=n_output)

outputs, state = tf.nn.dynamic_rnn(cell, x, dtype=tf.float32)

# An additional dense layer before the final predictions
y_pred = tf.layers.dense(inputs=outputs, units=1, activation=None)
# Predictions (one dimensional)
y_pred = tf.squeeze(y_pred)

#loss
loss = tf.reduce_mean(tf.square(outputs-y))

optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)

train = optimizer.minimize(loss)

init = tf.global_variables_initializer()

with tf.Session() as sess:
    output_pred = []
    tf.global_variables_initializer().run()
    for epoch in range(n_epochs):
        print("Epoch: ", epoch+1,end=' \n ')
        for step in range(steps_per_epoch):
            for (a, b) in zip(train_x, train_y):
                X_values = np.array(a).reshape(1, np.array(a).shape[0], np.array(a).shape[1])
                y_values = b
                sess.run(train, feed_dict={x: X_values, y: y_values})

            if step % 10 == 0:
                c = sess.run(loss, feed_dict={x: X_values, y: y_values})
                print("Step: ", step+1, "loss = ", c,)

    print ("Opimization Finished")
    training_loss = sess.run(loss, feed_dict={x: X_values, y: y_values})
    print ("Training loss= ", training_loss)
    save_path = saver.save(sess, "./models/tensorflow1.ckpt")

运行网络后，出现此错误：

Epoch:  1 


---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-28-95172c547171> in <module>
     15             #y_values = next(iter_train_y)
     16 
---> 17                 sess.run(train, feed_dict={x: X_values, y: y_values})
     18 
     19             if step % 10 == 0:

C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\client\session.py in run(self, fetches, feed_dict, options, run_metadata)
    898     try:
    899       result = self._run(None, fetches, feed_dict, options_ptr,
--> 900                          run_metadata_ptr)
    901       if run_metadata:
    902         proto_data = tf_session.TF_GetBuffer(run_metadata_ptr)

C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\client\session.py in _run(self, handle, fetches, feed_dict, options, run_metadata)
   1102             feed_handles[subfeed_t] = subfeed_val
   1103           else:
-> 1104             np_val = np.asarray(subfeed_val, dtype=subfeed_dtype)
   1105 
   1106           if (not is_tensor_handle_feed and

C:\ProgramData\Anaconda3\lib\site-packages\numpy\core\numeric.py in asarray(a, dtype, order)
    499 
    500     """
--> 501     return array(a, dtype, copy=False, order=order)
    502 
    503 

TypeError: float() argument must be a string or a number, not 'set'

如果我使用与数据集相似的虚拟数据来运行网络，那么它将正常工作。

有人可以告诉我可能是什么问题吗？

提前感谢：）

Tensorflow：float（）参数必须是字符串或数字，而不是'set'

0 个答案: