Python tensorflow.flatten方法代码示例

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def bw_dynamic_rnn(cell, inputs, sequence_length=None, initial_state=None,
                   dtype=None, parallel_iterations=None, swap_memory=False,
                   time_major=False, scope=None):
    assert not time_major  # TODO : to be implemented later!

    flat_inputs = flatten(inputs, 2)  # [-1, J, d]
    flat_len = None if sequence_length is None else tf.cast(flatten(sequence_length, 0), 'int64')

    flat_inputs = tf.reverse(flat_inputs, 1) if sequence_length is None \
        else tf.reverse_sequence(flat_inputs, sequence_length, 1)
    flat_outputs, final_state = _dynamic_rnn(cell, flat_inputs, sequence_length=flat_len,
                                             initial_state=initial_state, dtype=dtype,
                                             parallel_iterations=parallel_iterations, swap_memory=swap_memory,
                                             time_major=time_major, scope=scope)
    flat_outputs = tf.reverse(flat_outputs, 1) if sequence_length is None \
        else tf.reverse_sequence(flat_outputs, sequence_length, 1)

    outputs = reconstruct(flat_outputs, inputs, 2)
    return outputs, final_state 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def bidirectional_dynamic_rnn(cell_fw, cell_bw, inputs, sequence_length=None,
                              initial_state_fw=None, initial_state_bw=None,
                              dtype=None, parallel_iterations=None,
                              swap_memory=False, time_major=False, scope=None):
    assert not time_major

    flat_inputs = flatten(inputs, 2)  # [-1, J, d]
    flat_len = None if sequence_length is None else tf.cast(flatten(sequence_length, 0), 'int64')

    (flat_fw_outputs, flat_bw_outputs), final_state = \
        _bidirectional_dynamic_rnn(cell_fw, cell_bw, flat_inputs, sequence_length=flat_len,
                                   initial_state_fw=initial_state_fw, initial_state_bw=initial_state_bw,
                                   dtype=dtype, parallel_iterations=parallel_iterations, swap_memory=swap_memory,
                                   time_major=time_major, scope=scope)

    fw_outputs = reconstruct(flat_fw_outputs, inputs, 2)
    bw_outputs = reconstruct(flat_bw_outputs, inputs, 2)
    # FIXME : final state is not reshaped!
    return (fw_outputs, bw_outputs), final_state 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def bidirectional_rnn(cell_fw, cell_bw, inputs,
                      initial_state_fw=None, initial_state_bw=None,
                      dtype=None, sequence_length=None, scope=None):

    flat_inputs = flatten(inputs, 2)  # [-1, J, d]
    flat_len = None if sequence_length is None else tf.cast(flatten(sequence_length, 0), 'int64')

    (flat_fw_outputs, flat_bw_outputs), final_state = \
        tf.nn.bidirectional_dynamic_rnn(cell_fw, cell_bw, flat_inputs, sequence_length=flat_len,
                                        initial_state_fw=initial_state_fw, initial_state_bw=initial_state_bw,
                                        dtype=dtype, scope=scope)

    fw_outputs = reconstruct(flat_fw_outputs, inputs, 2)
    bw_outputs = reconstruct(flat_bw_outputs, inputs, 2)
    # FIXME : final state is not reshaped!
    return (fw_outputs, bw_outputs), final_state 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def linear(args, output_size, bias, bias_start=0.0, scope=None, squeeze=False, wd=0.0, input_keep_prob=1.0,
           is_train=None):
    if args is None or (nest.is_sequence(args) and not args):
        raise ValueError("`args` must be specified")
    if not nest.is_sequence(args):
        args = [args]

    flat_args = [flatten(arg, 1) for arg in args]
    if input_keep_prob < 1.0:
        assert is_train is not None
        flat_args = [tf.cond(is_train, lambda: tf.nn.dropout(arg, input_keep_prob), lambda: arg)
                     for arg in flat_args]
    with tf.variable_scope(scope or 'Linear'):
        flat_out = _linear(flat_args, output_size, bias, bias_initializer=tf.constant_initializer(bias_start))
    out = reconstruct(flat_out, args[0], 1)
    if squeeze:
        out = tf.squeeze(out, [len(args[0].get_shape().as_list())-1])
    if wd:
        add_wd(wd)

    return out 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def __init__(self, cell, memory, mask=None, controller=None, mapper=None, input_keep_prob=1.0, is_train=None):
        """
        Early fusion attention cell: uses the (inputs, state) to control the current attention.

        :param cell:
        :param memory: [N, M, m]
        :param mask:
        :param controller: (inputs, prev_state, memory) -> memory_logits
        """
        self._cell = cell
        self._memory = memory
        self._mask = mask
        self._flat_memory = flatten(memory, 2)
        self._flat_mask = flatten(mask, 1)
        if controller is None:
            controller = AttentionCell.get_linear_controller(True, is_train=is_train)
        self._controller = controller
        if mapper is None:
            mapper = AttentionCell.get_concat_mapper()
        elif mapper == 'sim':
            mapper = AttentionCell.get_sim_mapper()
        self._mapper = mapper 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def linear(args, output_size, bias, bias_start=0.0, scope=None, squeeze=False, wd=0.0, input_keep_prob=1.0,
           is_train=None):
    if args is None or (nest.is_sequence(args) and not args):
        raise ValueError("`args` must be specified")
    if not nest.is_sequence(args):
        args = [args]

    flat_args = [flatten(arg, 1) for arg in args]
    if input_keep_prob < 1.0:
        assert is_train is not None
        flat_args = [tf.cond(is_train, lambda: tf.nn.dropout(arg, input_keep_prob), lambda: arg)
                     for arg in flat_args]
    with tf.variable_scope(scope or 'Linear'):
        flat_out = _linear(flat_args, output_size, bias, bias_start=bias_start)
    out = reconstruct(flat_out, args[0], 1)
    if squeeze:
        out = tf.squeeze(out, [len(args[0].get_shape().as_list())-1])
    if wd:
        add_wd(wd)

    return out 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def dynamic_rnn(cell, inputs, sequence_length=None, initial_state=None,
                dtype=None, parallel_iterations=None, swap_memory=False,
                time_major=False, scope=None):
    assert not time_major  # TODO : to be implemented later!
    print("dynamic rnn input")
    print(inputs.get_shape())
    flat_inputs = flatten(inputs, 2)  # [-1, J, d]
    print("dynamic rnn flatten shape")
    print(flat_inputs.get_shape())
    flat_len = None if sequence_length is None else tf.cast(flatten(sequence_length, 0), 'int64')

    flat_outputs, final_state = _dynamic_rnn(cell, flat_inputs, sequence_length=flat_len,
                                             initial_state=initial_state, dtype=dtype,
                                             parallel_iterations=parallel_iterations, swap_memory=swap_memory,
                                             time_major=time_major, scope=scope)
    print("flat_outputs shape")
    print(flat_outputs.get_shape())
    outputs = reconstruct(flat_outputs, inputs, 2)
    return outputs, final_state 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def linear(args, output_size, bias, bias_start=0.0, scope=None, squeeze=False, wd=0.0, input_keep_prob=1.0,
           is_train=None):
    with tf.variable_scope(scope or "linear"):
        if args is None or (nest.is_sequence(args) and not args):
            raise ValueError("`args` must be specified")
        if not nest.is_sequence(args):
            args = [args]

        flat_args = [flatten(arg, 1) for arg in args]
        # if input_keep_prob < 1.0:
        assert is_train is not None
        flat_args = [tf.cond(is_train, lambda: tf.nn.dropout(arg, input_keep_prob), lambda: arg)
                         for arg in flat_args]
        flat_out = _linear(flat_args, output_size, bias)
        out = reconstruct(flat_out, args[0], 1)
        if squeeze:
            out = tf.squeeze(out, [len(args[0].get_shape().as_list())-1])
        if wd:
            add_wd(wd)

    return out 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def bidirectional_rnn(cell_fw, cell_bw, inputs,
                      initial_state_fw=None, initial_state_bw=None,
                      dtype=None, sequence_length=None, scope=None):

    flat_inputs = flatten(inputs, 2)  # [-1, J, d]
    flat_len = None if sequence_length is None else tf.cast(flatten(sequence_length, 0), 'int64')

    (flat_fw_outputs, flat_bw_outputs), final_state = \
        _bidirectional_rnn(cell_fw, cell_bw, flat_inputs, sequence_length=flat_len,
                           initial_state_fw=initial_state_fw, initial_state_bw=initial_state_bw,
                           dtype=dtype, scope=scope)

    fw_outputs = reconstruct(flat_fw_outputs, inputs, 2)
    bw_outputs = reconstruct(flat_bw_outputs, inputs, 2)
    # FIXME : final state is not reshaped!
    return (fw_outputs, bw_outputs), final_state 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def linear(args, output_size, bias, bias_start=0.0, scope=None, squeeze=False, wd=0.0, input_keep_prob=1.0,
           is_train=None):
    if args is None or (nest.is_sequence(args) and not args):
        raise ValueError("`args` must be specified")
    if not nest.is_sequence(args):
        args = [args]

    flat_args = [flatten(arg, 1) for arg in args]
    if input_keep_prob < 1.0:
        assert is_train is not None
        flat_args = [tf.cond(is_train, lambda: tf.nn.dropout(arg, input_keep_prob), lambda: arg)
                     for arg in flat_args]
    flat_out = _linear(flat_args, output_size, bias, bias_start=bias_start, scope=scope)
    out = reconstruct(flat_out, args[0], 1)
    if squeeze:
        out = tf.squeeze(out, [len(args[0].get_shape().as_list())-1])
    if wd:
        add_wd(wd)

    return out 

# 需要导入模块: from my import tensorflow [as 别名]
# 或者: from my.tensorflow import flatten [as 别名]
def dynamic_rnn(cell, inputs, sequence_length=None, initial_state=None,
                dtype=None, parallel_iterations=None, swap_memory=False,
                time_major=False, scope=None):
    assert not time_major  # TODO : to be implemented later!
    flat_inputs = flatten(inputs, 2)  # [-1, J, d]
    flat_len = None if sequence_length is None else tf.cast(flatten(sequence_length, 0), 'int64')

    flat_outputs, final_state = _dynamic_rnn(cell, flat_inputs, sequence_length=flat_len,
                                             initial_state=initial_state, dtype=dtype,
                                             parallel_iterations=parallel_iterations, swap_memory=swap_memory,
                                             time_major=time_major, scope=scope)

    outputs = reconstruct(flat_outputs, inputs, 2)
    return outputs, final_state