InvalidArgumentError（请参阅上面的回溯）：必须输入占位符张量的值

Question

我的错误： InvalidArgumentError (see above for traceback): You must feed a value for placeholder tensor 'label1_' with dtype complex64 and shape [100,129] [[Node: label1_ = Placeholder[dtype=DT_COMPLEX64, shape=[100,129], _device="/job:localhost/replica:0/task:0/device:GPU:0"]()]]

我的代码：

def model_inputs():
"""
Create the model inputs
"""
print('Input data to network..')
inputs_ = tf.placeholder(tf.complex64, [batch_size, sequence_length, freq_resolution], name='inputs')
label1_ = tf.placeholder(tf.complex64, [batch_size, freq_resolution], name='label1_')
label2_ = tf.placeholder(tf.complex64, [batch_size, freq_resolution], name='label2_')
keep_prob_ = tf.placeholder(tf.float32, name='keep_prob')
return inputs_, label1_, label2_, keep_prob_


##################### ###########################################
    with tf.Session() as sess:

    sess.run(tf.global_variables_initializer())
    n_batches = (len(mix) // batch_size) -2
    for e in range(epochs):
        for ii, (x, y, z) in enumerate(utl.get_batches(mix, src1, src2,  batch_size), 1):
            feed = {inputs_: x,  # shape = (batch_size, sequence_length, 129)
                    label1_: y,  # shape = (batch_size, 129)
                    label2_: z,  # shape = (batch_size, 129)
                    keep_prob_: keep_prob
                    }

sess.run（tf.global_variables_initializer（）） 中发生错误！

来自(utl.get_batches(mix, src1, src2, batch_size), 1):的输入与占位符的形状相同。

为什么会发生这种错误？