def seq2seq_inputs(x, y, input_length, output_length, sentinel=None, name=None):
  """Processes inputs for Sequence to Sequence models.

  Args:
    x: Input Tensor [batch_size, input_length, embed_dim].
    y: Output Tensor [batch_size, output_length, embed_dim].
    input_length: length of input x.
    output_length: length of output y.
    sentinel: optional first input to decoder and final output expected.
      If sentinel is not provided, zeros are used. Due to fact that y is not
      available in sampling time, shape of sentinel will be inferred from x.
    name: Operation name.

  Returns:
    Encoder input from x, and decoder inputs and outputs from y.
  """
  with ops.name_scope(name, "seq2seq_inputs", [x, y]):
    in_x = array_ops_.unpack(x, axis=1)
    y = array_ops_.unpack(y, axis=1)
    if not sentinel:
      # Set to zeros of shape of y[0], using x for batch size.
      sentinel_shape = array_ops_.pack(
          [array_ops_.shape(x)[0], y[0].get_shape()[1]])
      sentinel = array_ops_.zeros(sentinel_shape)
      sentinel.set_shape(y[0].get_shape())
    in_y = [sentinel] + y
    out_y = y + [sentinel]
    return in_x, in_y, out_y

  def __call__(self,
               inputs,
               initial_state=None,
               dtype=None,
               sequence_length=None,
               scope=None):
    is_list = isinstance(inputs, list)
    if self._use_dynamic_rnn:
      if is_list:
        inputs = array_ops.pack(inputs)
      outputs, state = rnn.dynamic_rnn(
          self._cell,
          inputs,
          sequence_length=sequence_length,
          initial_state=initial_state,
          dtype=dtype,
          time_major=True,
          scope=scope)
      if is_list:
        # Convert outputs back to list
        outputs = array_ops.unpack(outputs)
    else:  # non-dynamic rnn
      if not is_list:
        inputs = array_ops.unpack(inputs)
      outputs, state = rnn.rnn(self._cell,
                               inputs,
                               initial_state=initial_state,
                               dtype=dtype,
                               sequence_length=sequence_length,
                               scope=scope)
      if not is_list:
        # Convert outputs back to tensor
        outputs = array_ops.pack(outputs)

    return outputs, state

 def testCannotInferNumFromUnknownShape(self):
   x = array_ops.placeholder(np.float32)
   with self.assertRaisesRegexp(ValueError,
                                r'Cannot infer num from shape <unknown>'):
     array_ops.unpack(x)
   with self.assertRaisesRegexp(ValueError,
                                r'Cannot infer num from shape <unknown>'):
     array_ops.unstack(x)

 def testCannotInferNumFromNoneShape(self):
   x = array_ops.placeholder(np.float32, shape=(None,))
   with self.assertRaisesRegexp(ValueError,
                                r'Cannot infer num from shape \(\?,\)'):
     array_ops.unpack(x)
   with self.assertRaisesRegexp(ValueError,
                                r'Cannot infer num from shape \(\?,\)'):
     array_ops.unstack(x)

 def testSimple(self):
   np.random.seed(7)
   with self.test_session(use_gpu=True):
     for shape in (2,), (3,), (2, 3), (3, 2), (4, 3, 2):
       data = np.random.randn(*shape)
       # Convert data to a single tensorflow tensor
       x = constant_op.constant(data)
       # Unpack into a list of tensors
       cs_unpacked = array_ops.unpack(x, num=shape[0])
       cs_unstacked = array_ops.unpack(x, num=shape[0])
       for cs in (cs_unpacked, cs_unstacked):
         self.assertEqual(type(cs), list)
         self.assertEqual(len(cs), shape[0])
         cs = [c.eval() for c in cs]
         self.assertAllEqual(cs, data)

 def _cat_probs(self, log_probs):
   """Get a list of num_components batchwise probabilities."""
   which_softmax = nn_ops.log_softmax if log_probs else nn_ops.softmax
   cat_probs = which_softmax(self.cat.logits)
   cat_probs = array_ops.unpack(
       cat_probs, num=self.num_components, axis=-1)
   return cat_probs

def _reverse_seq(input_seq, lengths):
  """Reverse a list of Tensors up to specified lengths.

  Args:
    input_seq: Sequence of seq_len tensors of dimension (batch_size, depth)
    lengths:   A tensor of dimension batch_size, containing lengths for each
               sequence in the batch. If "None" is specified, simply
               reverses the list.

  Returns:
    time-reversed sequence
  """
  if lengths is None:
    return list(reversed(input_seq))

  for input_ in input_seq:
    input_.set_shape(input_.get_shape().with_rank(2))

  # Join into (time, batch_size, depth)
  s_joined = array_ops_.pack(input_seq)

  # Reverse along dimension 0
  s_reversed = array_ops_.reverse_sequence(s_joined, lengths, 0, 1)
  # Split again into list
  result = array_ops_.unpack(s_reversed)
  return result

def _reverse_seq(input_seq, lengths):
    """Reverse a list of Tensors up to specified lengths.

  Args:
    input_seq: Sequence of seq_len tensors of dimension (batch_size, n_features)
    lengths:   A tensor of dimension batch_size, containing lengths for each
               sequence in the batch. If "None" is specified, simply reverses
               the list.

  Returns:
    time-reversed sequence
  """
    if lengths is None:
        return list(reversed(input_seq))

    input_shape = tensor_shape.unknown_shape(ndims=input_seq[0].get_shape().ndims)
    for input_ in input_seq:
        input_shape.merge_with(input_.get_shape())
        input_.set_shape(input_shape)

    # Join into (time, batch_size, depth)
    s_joined = array_ops.pack(input_seq)

    # TODO(schuster, ebrevdo): Remove cast when reverse_sequence takes int32
    if lengths is not None:
        lengths = math_ops.to_int64(lengths)

    # Reverse along dimension 0
    s_reversed = array_ops.reverse_sequence(s_joined, lengths, 0, 1)
    # Split again into list
    result = array_ops.unpack(s_reversed)
    for r in result:
        r.set_shape(input_shape)
    return result

 def __call__(self, inputs, state, scope=None):
   """Run this multi-layer cell on inputs, starting from state."""
   with vs.variable_scope(scope or type(self).__name__):  # "MultiRNNCell"
     cur_state_pos = 0
     cur_inp = inputs
     new_states = []
     for i, cell in enumerate(self._cells):
       with vs.variable_scope("Cell%d" % i):
         if self._state_is_tuple:
           if not nest.is_sequence(state):
             raise ValueError(
                 "Expected state to be a tuple of length %d, but received: %s"
                 % (len(self.state_size), state))
           cur_state = state[i]
         else:
           # print("STATE",state)
           """
           cur_state = array_ops.slice(
               state, [0, cur_state_pos], [-1, cell.state_size])
           """
           cur_state = array_ops.unpack(state)[i]
           # cur_state_pos += cell.state_size
         cur_inp, new_state = cell(cur_inp, cur_state)
         new_states.append(new_state)
   """
   new_states = (tuple(new_states) if self._state_is_tuple
                 else array_ops.concat(1, new_states))
   """
   new_states = array_ops.pack(new_states)
   return cur_inp, new_states

  def testZeroLengthDim(self):
    with self.test_session():
      x = array_ops.zeros(shape=(0, 1, 2))
      y = array_ops.unpack(x, axis=1)[0].eval()
      self.assertEqual(y.shape, (0, 2))

      y = array_ops.unstack(x, axis=1)[0].eval()
      self.assertEqual(y.shape, (0, 2))

  def testInferNum(self):
    with self.test_session():
      for shape in (2,), (3,), (2, 3), (3, 2), (4, 3, 2):
        x = array_ops.placeholder(np.float32, shape=shape)
        cs = array_ops.unpack(x)
        self.assertEqual(type(cs), list)
        self.assertEqual(len(cs), shape[0])

        cs = array_ops.unstack(x)
        self.assertEqual(type(cs), list)
        self.assertEqual(len(cs), shape[0])

  def testAxis0Default(self):
    with self.test_session() as sess:
      a = constant_op.constant([[1, 2, 3], [4, 5, 6]], name='a')

      unpacked = sess.run(array_ops.unpack(a))
      unstacked = sess.run(array_ops.unstack(a))

    self.assertEqual(len(unpacked), 2)
    self.assertAllEqual(unpacked[0], [1, 2, 3])
    self.assertAllEqual(unpacked[1], [4, 5, 6])
    self.assertEqual(len(unstacked), 2)
    self.assertAllEqual(unstacked[0], [1, 2, 3])
    self.assertAllEqual(unstacked[1], [4, 5, 6])

 def _sample_n(self, n, seed=None):
     # We use 2 uniform random floats to generate polar random variates.
     # http://dl.acm.org/citation.cfm?id=179631
     # Theorem 2. Let G, H be iid variates, uniformly distributed on [0,1].
     # Let theta = 2*pi*H, let R = sqrt(df*(G^(-2/df) - 1)) for df > 0.
     # Let X = R*cos(theta), and let Y = R*sin(theta).
     # Then X ~ t_df and Y ~ t_df.
     # The variates X and Y are not independent.
     shape = array_ops.concat(0, ([2, n], self.batch_shape()))
     uniform = random_ops.random_uniform(shape=shape, dtype=self.dtype, seed=seed)
     samples_g, samples_h = array_ops.unpack(uniform, num=2)
     theta = (2.0 * math.pi) * samples_h
     r = math_ops.sqrt(self.df * (math_ops.pow(samples_g, -2 / self.df) - 1))
     samples = r * math_ops.cos(theta)
     return samples * self.sigma + self.mu

  def testAgainstNumpy(self):
    # For 1 to 5 dimensions.
    for i in range(1, 6):
      a = np.random.random(np.random.permutation(i) + 1)

      # For all the possible axis to split it, including negative indices.
      for j in range(-i, i):
        expected = np_split_squeeze(a, j)

        with self.test_session() as sess:
          actual_unpack = sess.run(array_ops.unpack(a, axis=j))
          actual_unstack = sess.run(array_ops.unstack(a, axis=j))

        self.assertAllEqual(expected, actual_unpack)
        self.assertAllEqual(expected, actual_unstack)

  def testGradientsAxis0(self):
    for shape in (2,), (3,), (2, 3), (3, 2), (4, 3, 2):
      data = np.random.randn(*shape)
      shapes = [shape[1:]] * shape[0]
      for i in xrange(shape[0]):
        with self.test_session(use_gpu=True):
          x = constant_op.constant(data)
          cs = array_ops.unpack(x, num=shape[0])
          err = gradient_checker.compute_gradient_error(x, shape, cs[i],
                                                        shapes[i])
          self.assertLess(err, 1e-6)

          cs = array_ops.unstack(x, num=shape[0])
          err = gradient_checker.compute_gradient_error(x, shape, cs[i],
                                                        shapes[i])
          self.assertLess(err, 1e-6)

def _reverse_seq(input_seq, lengths):
  """Reverse a list of Tensors up to specified lengths.

  Args:
    input_seq: Sequence of seq_len tensors of dimension (batch_size, n_features)
               or nested tuples of tensors.
    lengths:   A tensor of dimension batch_size, containing lengths for each
               sequence in the batch. If "None" is specified, simply reverses
               the list.

  Returns:
    time-reversed sequence
  """
  if lengths is None:
    return list(reversed(input_seq))

  input_is_tuple = nest.is_sequence(input_seq[0])
  flat_input_seq = (nest.flatten(input_) if input_is_tuple else [input_]
                    for input_ in input_seq)

  flat_results = [[] for _ in range(len(input_seq))]
  for sequence in zip(*flat_input_seq):
    input_shape = tensor_shape.unknown_shape(
        ndims=sequence[0].get_shape().ndims)
    for input_ in sequence:
      input_shape.merge_with(input_.get_shape())
      input_.set_shape(input_shape)

    # Join into (time, batch_size, depth)
    s_joined = array_ops.pack(sequence)

    # TODO(schuster, ebrevdo): Remove cast when reverse_sequence takes int32
    if lengths is not None:
      lengths = math_ops.to_int64(lengths)

    # Reverse along dimension 0
    s_reversed = array_ops.reverse_sequence(s_joined, lengths, 0, 1)
    # Split again into list
    result = array_ops.unpack(s_reversed)
    for r, flat_result in zip(result, flat_results):
      r.set_shape(input_shape)
      flat_result.append(r)

  results = [nest.pack_sequence_as(structure=input_, flat_sequence=flat_result)
             if input_is_tuple else flat_result[0]
             for input_, flat_result in zip(input_seq, flat_results)]
  return results

  def testGradientsAxis1(self):
    for shape in (2, 3), (3, 2), (4, 3, 2):
      data = np.random.randn(*shape)
      out_shape = list(shape)
      del out_shape[1]
      for i in xrange(shape[1]):
        with self.test_session(use_gpu=True):
          x = constant_op.constant(data)
          cs = array_ops.unpack(x, num=shape[1], axis=1)
          err = gradient_checker.compute_gradient_error(x, shape, cs[i],
                                                        out_shape)
          self.assertLess(err, 1e-6)

          cs = array_ops.unstack(x, num=shape[1], axis=1)
          err = gradient_checker.compute_gradient_error(x, shape, cs[i],
                                                        out_shape)
          self.assertLess(err, 1e-6)

def _ImageDimensions(image):
    """Returns the dimensions of an image tensor.

  Args:
    image: A 3-D Tensor of shape `[height, width, channels]`.

  Returns:
    A list of `[height, width, channels]` corresponding to the dimensions of the
    input image.  Dimensions that are statically known are python integers,
    otherwise they are integer scalar tensors.
  """
    if image.get_shape().is_fully_defined():
        return image.get_shape().as_list()
    else:
        static_shape = image.get_shape().with_rank(3).as_list()
        dynamic_shape = array_ops.unpack(array_ops.shape(image), 3)
        return [s if s is not None else d for s, d in zip(static_shape, dynamic_shape)]

def dense_to_sparse_tensor(dense_tensor, ignore_value=None):
  """Converts a dense Tensor to a SparseTensor, dropping ignore_value cells.

  Args:
    dense_tensor: A `Tensor`.
    ignore_value: Entries in `dense_tensor` equal to this value will be
      absent from the return `SparseTensor`. If `None`, default value of
      dense_tensor's dtype will be used (e.g. '' for `str`, 0 for `int`).

  Returns:
    A `SparseTensor` with the same shape as `dense_tensor`.

  Raises:
    ValueError: when `dense_tensor`'s rank is `None`.
  """
  with ops.name_scope("DenseToSparseTensor"):
    dense_t = ops.convert_to_tensor(dense_tensor)
    if dense_t.get_shape().ndims is None:
      # TODO(b/32318825): Implement dense_to_sparse_tensor for undefined rank.
      raise ValueError("dense_tensor.get_shape() should be defined, got None.")
    if ignore_value is None:
      if dense_t.dtype == dtypes.string:
        # Exception due to TF strings are converted to numpy objects by default.
        ignore_value = ""
      else:
        ignore_value = dense_t.dtype.as_numpy_dtype()
    dense_shape = math_ops.cast(array_ops.shape(dense_t), dtypes.int64)
    indices = array_ops.where(
        math_ops.not_equal(dense_t, math_ops.cast(ignore_value, dense_t.dtype)))
    index_dims = len(dense_t.get_shape())
    # Flattens the tensor and indices for use with gather.
    flat_tensor = array_ops.reshape(dense_t, [-1])
    flat_indices = indices[:, index_dims - 1]
    # Computes the correct flattened indices for 2d (or higher) tensors.
    if index_dims > 1:
      higher_dims = indices[:, :index_dims - 1]
      shape_multipliers = array_ops.pack(
          _multiplier_helper(array_ops.unpack(dense_shape)[1:]))
      offsets = math_ops.reduce_sum(
          math_ops.mul(higher_dims, shape_multipliers), reduction_indices=[1])
      flat_indices = math_ops.add(flat_indices, offsets)
    values = array_ops.gather(flat_tensor, flat_indices)
    return sparse_tensor.SparseTensor(indices, values, dense_shape)

def _ImageDimensions(images, static_only=True):
  """Returns the dimensions of an image tensor.

  Args:
    images: 4-D Tensor of shape `[batch, height, width, channels]`
    static_only: Boolean, whether to return only static shape.

  Returns:
    list of integers `[batch, height, width, channels]`, when static shape is
    fully defined or `static_only` is `True`.
    list of integer scalar tensors `[batch, height, width, channels]`, when
    static shape is not fully defined.
  """
  # A simple abstraction to provide names for each dimension. This abstraction
  # should make it simpler to switch dimensions in the future (e.g. if we ever
  # want to switch height and width.)
  if static_only or images.get_shape().is_fully_defined():
    return images.get_shape().as_list()
  else:
    return array_ops.unpack(array_ops.shape(images))