小编Ian*_*rci的帖子

我正在调整其中一个MNIST tensorflow教程,我收到了这个TypeError.根据这个问题,你必须在字典键中使用占位符,因为numpy数组是可变的.我相信我这样做,但我仍然收到这个错误.

# Network Parameters
n_input = 44100 # length of FFT
n_classes = 6 # 6 instrument classes
dropout = 0.75 # Dropout, probability to keep units

# TF Graph input
x = tf.placeholder(tf.float32, [None, n_input])
y = tf.placeholder(tf.float32, [None, n_classes])
keep_prob = tf.placeholder(tf.float32)

我填写我的批次,然后将它们传递给会话.

for file_name in os.listdir('./Input_FFTs'):
    if file_name.endswith('.txt'):
        path = './Input_FFTs/' + file_name
        y, x = getData(path)
        batch_ys[count] = y
        batch_xs[count] = x
        count += 1
sess.run(optimizer, feed_dict={x: batch_xs, y: batch_ys,
                                   keep_prob: dropout})

当我打印并检查batch_xs和batch_ys的大小时,它们是[batch_size,44100]和[batch_size,6],并带有正确的数据.这些匹配x和y占位符的预期大小. …