Python构建对象实例：object（）不带参数

Question

I'm writing a tensorflow program similar to their MNIST LSTM example code. I'm building my data file, and I am unable to build an object instance.

Essentially the process goes: define an empyty instance data_sets = DataSet() and then build the object data_sets.train = DataSet(arg1, arg2...) and data_sets.test = DataSet(arg1, arg2...) and so on

I'm getting the error (exact error at the bottom) when I try to build data_sets.train = DataSet(arg1, arg2...)

The MNIST code looks like this:

class DataSet(object):

  def __init__(self, images, labels, fake_data=False, one_hot=False,
           dtype=tf.float32):
    """Construct a DataSet.

    one_hot arg is used only if fake_data is true.  `dtype` can be either
    `uint8` to leave the input as `[0, 255]`, or `float32` to rescale into
    `[0, 1]`.
    """
    dtype = tf.as_dtype(dtype).base_dtype
    #pdb.set_trace()
    if dtype not in (tf.uint8, tf.float32):
      raise TypeError('Invalid image dtype %r, expected uint8 or float32' %
                      dtype)
    if fake_data:
      self._num_examples = 10000
      self.one_hot = one_hot
    else:
      pdb.set_trace()
      assert images.shape[0] == labels.shape[0], (
          'images.shape: %s labels.shape: %s' % (images.shape,
                                                 labels.shape))
      self._num_examples = images.shape[0]

      # Convert shape from [num examples, rows, columns, depth]
      # to [num examples, rows*columns] (assuming depth == 1)
      assert images.shape[3] == 1
      images = images.reshape(images.shape[0],
                              images.shape[1] * images.shape[2])
      if dtype == tf.float32:
        # Convert from [0, 255] -> [0.0, 1.0].
        images = images.astype(numpy.float32)
        images = numpy.multiply(images, 1.0 / 255.0)
    self._images = images
    self._labels = labels
    self._epochs_completed = 0
    self._index_in_epoch = 0

  @property
  def images(self):
    return self._images

  @property
  def labels(self):
    return self._labels

  @property
  def num_examples(self):
    return self._num_examples

  @property
  def epochs_completed(self):
    return self._epochs_completed

So then in the same file they have a function that defines an instance without arguments (after a pass ), builds the dataset (I have left that part out), and then builds the object data_set with data_set.train, data_set.validation, and data_set.test -- each time they call the class constructor again, but this time they include arguments. As shown below

def read_data_sets(train_dir, fake_data=False, one_hot=False, dtype=tf.float32):
  class DataSets(object):
    pass
  pdb.set_trace()
  data_sets = DataSets()
  ...(build dataset)...
  data_sets.train = DataSet(train_images, train_labels, dtype=dtype)
  data_sets.validation = DataSet(validation_images, validation_labels,
                             dtype=dtype)
  data_sets.test = DataSet(test_images, test_labels, dtype=dtype)
  pdb.set_trace()
  return data_sets

I have essentially built the exact same thing, but with a different dataset

Here is my class definition (ignore tabs, copy&paste messed up identation -- I don't think indentation is the problem)

class ScrollData(object):

  def __init__(self, images, labels, dtype=tf.float32):

    dtype = tf.as_dtype(dtype).base_dtype
    if dtype not in (tf.float64, tf.float32):
        raise TypeError('Invalid image dtype %r, expected float64 or float32' %
                  dtype)
    assert images.shape[0] == labels.shape[0], (
      'images.shape: %s labels.shape: %s' % (images.shape,
                                             labels.shape))
    self._num_examples = images.shape[0]
    pdb.set_trace()
    assert images.shape[3] == 1
    images = images.reshape(images.shape[0],
                          images.shape[1] * images.shape[2])
    if dtype == tf.float32:
        # Convert from [0, 255] -> [0.0, 1.0].
        images = images.astype(numpy.float32)
        images = numpy.multiply(images, 1.0 / 255.0)

    self._images = images
    self._labels = labels
    self._epochs_completed = 0
    self._index_in_epoch = 0

@property
def images(self):
    return self._images

@property
def labels(self):
    return self._labels

@property
def num_examples(self):
    return self._num_examples

@property
def epochs_completed(self):
    return self._epochs_completed

And then I build the object in the following method:

def read_data(data_dir):
    dtype=tf.float32
    VALIDATION_SIZE = 1
    TEST_SIZE = 1

    class ScrollData(object):
        pass

    data_sets = ScrollData()

    ...(build dataset)...

    data_sets.train = ScrollData(train_images, train_labels,  dtype=tf.float32)
    data_sets.validation = ScrollData(validation_images, validation_labels,  dtype=tf.float32)
    data_sets.test = ScrollData(testtest_images, test_labels,  dtype=tf.float32)

    return data_sets

I'm getting the following error:

data_sets.train = ScrollData(train_images, train_labels,  dtype=tf.float32)
TypeError: object() takes no parameters

Answer 1

You overwrite ScrollData within you read_data function with a class which doesn't take constructor arguments.

leave this redefinition out and either add arguments to the first call or define standard values in the constructor