# Copyright 2021 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
'''RNN operators module, include RNN, GRU'''
import math
import numpy as np
import mindspore.nn as nn
import mindspore.ops as P
import mindspore.context as context
import mindspore.common.dtype as mstype
from mindspore.ops.primitive import constexpr
from mindspore.common.tensor import Tensor
from mindspore.common.parameter import ParameterTuple, Parameter
from mindspore.nn.cell import Cell
from mindspore import log as logger
from mindspore._checkparam import Validator as validator
from mindspore.ops.operations._rl_inner_ops import CudnnGRU
from .rnn_cells import _rnn_relu_cell, _rnn_tanh_cell, _gru_cell, _lstm_cell
from .rnn_utils import _Reverse, _ReverseSequence
__all__ = ['LSTM', 'GRU', 'RNN']
@constexpr
def arange(start, stop, step, dtype):
return Tensor(np.arange(start, stop, step), dtype)
@constexpr
def _init_state(shape, dtype, is_lstm):
hx = Tensor(np.zeros(shape), dtype)
cx = Tensor(np.zeros(shape), dtype)
if is_lstm:
return (hx, cx)
return hx
@constexpr
def _check_input_dtype(input_dtype, param_name, allow_dtypes, cls_name):
validator.check_type_name(param_name, input_dtype, allow_dtypes, cls_name)
@constexpr
def _check_input_dtype_same_and_valid(args_name, args_value, valid_values, cls_name):
args = {args_name[i]: args_value[i] for i in range(len(args_value))}
validator.check_types_same_and_valid(args, valid_values, cls_name)
@constexpr
def _check_is_tensor(param_name, input_data, cls_name):
"""Internal function, used to check whether the input data is Tensor."""
if input_data is not None and not isinstance(P.typeof(input_data), mstype.tensor_type):
raise TypeError(f"For '{cls_name}', the '{param_name}' should be '{mstype.tensor_type}', "
f"but got '{P.typeof(input_data)}'")
@constexpr
def _check_is_tuple(param_name, input_data, cls_name):
"""Internal function, used to check whether the input data is Tensor."""
if input_data is not None and not isinstance(P.typeof(input_data), mstype.Tuple):
raise TypeError(f"For '{cls_name}', the '{param_name}' should be '{mstype.Tuple}', "
f"but got '{P.typeof(input_data)}'")
@constexpr
def _check_tuple_length(param_name, input_data, length, cls_name):
"""Internal function, used to check whether the input data is Tensor."""
if input_data is not None and len(input_data) != length:
raise TypeError(f"For '{cls_name}', the length of '{param_name}' should be '{length}', "
f"but got '{len(input_data)}'")
@constexpr
def _check_seq_length_size(batch_size_x, seq_length_size, cls_name):
if batch_size_x != seq_length_size:
raise ValueError(f"For '{cls_name}' batch size of x and seq_length should be equal, "
f"but got {batch_size_x} of x and {seq_length_size} of seq_length.")
def sequence_mask(lengths, maxlen):
"""generate mask matrix by seq_length"""
range_vector = arange(0, maxlen, 1, lengths.dtype)
result = range_vector < lengths.view(lengths.shape + (1,))
return result.astype(mstype.int32)
def select_by_mask(inputs, mask):
"""mask hiddens by mask matrix"""
return mask.view(mask.shape + (1,)).swapaxes(0, 1) \
.expand_as(inputs).astype(mstype.bool_) * inputs
def get_hidden(output, seq_length):
"""get hidden state by seq_length"""
batch_index = arange(0, seq_length.shape[0], 1, seq_length.dtype)
indices = P.Concat(1)((seq_length.view(-1, 1) - 1, batch_index.view(-1, 1)))
return P.GatherNd()(output, indices)
class _DynamicRNNBase(Cell):
'''Dynamic RNN module to compute RNN cell by timesteps'''
def __init__(self, mode):
super().__init__()
if mode == "RNN_RELU":
cell = _rnn_relu_cell
elif mode == "RNN_TANH":
cell = _rnn_tanh_cell
elif mode == "LSTM":
cell = _lstm_cell
elif mode == "GRU":
cell = _gru_cell
else:
raise ValueError("Unrecognized RNN mode: " + mode)
self.cell = cell
self.is_lstm = mode == "LSTM"
def recurrent(self, x, h_0, w_ih, w_hh, b_ih, b_hh):
'''recurrent steps without sequence length'''
time_step = x.shape[0]
outputs = []
t = 0
h = h_0
while t < time_step:
x_t = x[t:t + 1:1]
x_t = P.Squeeze(0)(x_t)
h = self.cell(x_t, h, w_ih, w_hh, b_ih, b_hh)
if self.is_lstm:
outputs.append(h[0])
else:
outputs.append(h)
t += 1
outputs = P.Stack()(outputs)
return outputs, h
def variable_recurrent(self, x, h, seq_length, w_ih, w_hh, b_ih, b_hh):
'''recurrent steps with sequence length'''
time_step = x.shape[0]
h_t = h
if self.is_lstm:
hidden_size = h[0].shape[-1]
zero_output = P.ZerosLike()(h_t[0])
else:
hidden_size = h.shape[-1]
zero_output = P.ZerosLike()(h_t)
seq_length = P.Cast()(seq_length, mstype.float32)
seq_length = P.BroadcastTo((hidden_size, -1))(seq_length)
seq_length = P.Cast()(seq_length, mstype.int32)
seq_length = P.Transpose()(seq_length, (1, 0))
outputs = []
state_t = h_t
t = 0
while t < time_step:
x_t = x[t:t + 1:1]
x_t = P.Squeeze(0)(x_t)
h_t = self.cell(x_t, state_t, w_ih, w_hh, b_ih, b_hh)
seq_cond = seq_length > t
if self.is_lstm:
state_t_0 = P.Select()(seq_cond, h_t[0], state_t[0])
state_t_1 = P.Select()(seq_cond, h_t[1], state_t[1])
output = P.Select()(seq_cond, h_t[0], zero_output)
state_t = (state_t_0, state_t_1)
else:
state_t = P.Select()(seq_cond, h_t, state_t)
output = P.Select()(seq_cond, h_t, zero_output)
outputs.append(output)
t += 1
outputs = P.Stack()(outputs)
return outputs, state_t
def construct(self, x, h, seq_length, w_ih, w_hh, b_ih, b_hh):
x_dtype = x.dtype
w_ih = w_ih.astype(x_dtype)
w_hh = w_hh.astype(x_dtype)
if b_ih is not None:
b_ih = b_ih.astype(x_dtype)
b_hh = b_hh.astype(x_dtype)
if seq_length is None:
return self.recurrent(x, h, w_ih, w_hh, b_ih, b_hh)
return self.variable_recurrent(x, h, seq_length, w_ih, w_hh, b_ih, b_hh)
class _DynamicRNNRelu(_DynamicRNNBase):
'''Dynamic RNN module with Relu activation'''
def __init__(self):
mode = 'RNN_RELU'
super().__init__(mode)
class _DynamicRNNTanh(_DynamicRNNBase):
'''Dynamic RNN module with Tanh activation'''
def __init__(self):
mode = 'RNN_TANH'
super().__init__(mode)
class _DynamicGRUCPUGPU(Cell):
'''Dynamic GRU module on CPU and GPU'''
def __init__(self):
super().__init__()
self.concat = P.Concat()
self.is_gpu = context.get_context("device_target") == "GPU"
def construct(self, x, h_0, seq_length, w_ih, w_hh, b_ih, b_hh):
gate_size, input_size = w_ih.shape
hidden_size = gate_size // 3
if self.is_gpu and seq_length is None:
if b_ih is None:
weights = self.concat((
w_ih.view(-1, 1, 1),
w_hh.view(-1, 1, 1)
))
has_bias = False
else:
has_bias = True
weights = self.concat((
w_ih.view(-1, 1, 1),
w_hh.view(-1, 1, 1),
b_ih.view(-1, 1, 1),
b_hh.view(-1, 1, 1)
))
output, h_n, _, _ = CudnnGRU(input_size, hidden_size, 1, has_bias, False, 0.0)(
x,
h_0.view(1, *h_0.shape),
weights.astype(x.dtype)
)
else:
output, h_n = _DynamicRNNBase('GRU')(x, h_0, seq_length, w_ih, w_hh, b_ih, b_hh)
return output, h_n
class _DynamicGRUAscend(Cell):
'''Dynamic GRU module on Ascend'''
def __init__(self):
super().__init__()
self.gru = P.DynamicGRUV2(gate_order='rzh')
self.transpose = P.Transpose()
self.dtype = mstype.float16
def construct(self, x, h_0, seq_length, w_ih, w_hh, b_ih, b_hh):
if b_ih is None:
b_ih = P.Zeros()(w_ih.shape[0], w_ih.dtype)
b_hh = P.Zeros()(w_ih.shape[0], w_ih.dtype)
outputs, _, _, _, _, _ = self.gru(self.cast(x, self.dtype), \
self.cast(self.transpose(w_ih, (1, 0)), self.dtype), \
self.cast(self.transpose(w_hh, (1, 0)), self.dtype), \
self.cast(b_ih, self.dtype), \
self.cast(b_hh, self.dtype), \
None, self.cast(h_0, self.dtype))
if seq_length is not None:
h = get_hidden(outputs, seq_length)
mask = sequence_mask(seq_length, x.shape[0])
outputs = select_by_mask(outputs, mask)
else:
h = outputs[-1]
return outputs, h
class _DynamicLSTMCPUGPU(Cell):
'''Dynamic LSTM module on CPU and GPU'''
def __init__(self):
super().__init__()
self.concat = P.Concat()
self.is_gpu = context.get_context("device_target") == "GPU"
def construct(self, x, h_0, seq_length, w_ih, w_hh, b_ih, b_hh):
gate_size, input_size = w_ih.shape
hidden_size = gate_size // 4
if seq_length is not None:
output, (h_n, c_n) = _DynamicRNNBase('LSTM')(x, h_0, seq_length, w_ih, w_hh, b_ih, b_hh)
else:
if b_ih is None:
weights = self.concat((
w_ih.view(-1, 1, 1),
w_hh.view(-1, 1, 1)
))
has_bias = False
else:
has_bias = True
if self.is_gpu:
weights = self.concat((
w_ih.view(-1, 1, 1),
w_hh.view(-1, 1, 1),
b_ih.view(-1, 1, 1),
b_hh.view(-1, 1, 1)
))
else:
bias = b_ih + b_hh
weights = self.concat((
w_ih.view(-1, 1, 1),
w_hh.view(-1, 1, 1),
bias.view(-1, 1, 1)
))
output, h_n, c_n, _, _ = P.LSTM(input_size, hidden_size, 1, has_bias, False, 0.0)(
x,
P.ExpandDims()(h_0[0], 0),
P.ExpandDims()(h_0[1], 0),
weights.astype(x.dtype)
)
return output, (h_n, c_n)
class _DynamicLSTMAscend(Cell):
'''Dynamic LSTM module on Ascend'''
def __init__(self):
super().__init__()
self.lstm = P.DynamicRNN()
self.concat_dim1 = P.Concat(axis=1)
self.concat_dim0 = P.Concat(axis=0)
self.transpose = P.Transpose()
self.cast = P.Cast()
self.split = P.Split(axis=0, output_num=4)
self.dtype = mstype.float16
def construct(self, x, h_0, seq_length, w_ih, w_hh, b_ih, b_hh):
w_ih_i, w_ih_f, w_ih_g, w_ih_o = self.split(w_ih)
w_hh_i, w_hh_f, w_hh_g, w_hh_o = self.split(w_hh)
w_ih = self.concat_dim0((w_ih_i, w_ih_g, w_ih_f, w_ih_o))
w_hh = self.concat_dim0((w_hh_i, w_hh_g, w_hh_f, w_hh_o))
weight = self.concat_dim1((w_ih, w_hh))
if b_ih is None:
bias = P.Zeros()(w_ih.shape[0], w_ih.dtype)
else:
b_ih_i, b_ih_f, b_ih_g, b_ih_o = self.split(b_ih)
b_hh_i, b_hh_f, b_hh_g, b_hh_o = self.split(b_hh)
bias = self.concat_dim0((b_ih_i + b_hh_i, \
b_ih_g + b_hh_g, \
b_ih_f + b_hh_f, \
b_ih_o + b_hh_o))
outputs, h, c, _, _, _, _, _ = self.lstm(self.cast(x, self.dtype), \
self.cast(self.transpose(weight, (1, 0)), self.dtype), \
self.cast(bias, self.dtype), None, \
self.cast(P.ExpandDims()(h_0[0], 0), self.dtype), \
self.cast(P.ExpandDims()(h_0[1], 0), self.dtype))
if seq_length is not None:
h = get_hidden(h, seq_length)
c = get_hidden(c, seq_length)
mask = sequence_mask(seq_length, x.shape[0])
outputs = select_by_mask(outputs, mask)
else:
h = h[-1]
c = c[-1]
return outputs, (h, c)
class _RNNBase(Cell):
'''Basic class for RNN operators'''
def __init__(self, mode, input_size, hidden_size, num_layers=1, has_bias=True,
batch_first=False, dropout=0., bidirectional=False):
super().__init__()
validator.check_positive_int(hidden_size, "hidden_size", self.cls_name)
validator.check_positive_int(input_size, "input_size", self.cls_name)
validator.check_positive_int(num_layers, "num_layers", self.cls_name)
validator.check_is_float(dropout, "dropout", self.cls_name)
validator.check_value_type("has_bias", has_bias, [bool], self.cls_name)
validator.check_value_type("batch_first", batch_first, [bool], self.cls_name)
validator.check_value_type("bidirectional", bidirectional, [bool], self.cls_name)
if not 0 <= dropout < 1:
raise ValueError(f"For '{self.cls_name}', the 'dropout' should be a number in range [0, 1) "
f"representing the probability of an element being zeroed, but got {dropout}.")
if dropout > 0 and num_layers == 1:
logger.warning("dropout option adds dropout after all but last "
"recurrent layer, so non-zero dropout expects "
"num_layers greater than 1, but got dropout={} and "
"num_layers={}".format(dropout, num_layers))
is_ascend = context.get_context("device_target") == "Ascend"
if mode == "LSTM":
gate_size = 4 * hidden_size
self.rnn = _DynamicLSTMAscend() if is_ascend else _DynamicLSTMCPUGPU()
elif mode == "GRU":
if is_ascend and hidden_size % 16 != 0:
raise ValueError(f"GRU on ascend do not support hidden size that is not divisible by 16, "
f"but get hidden size {hidden_size}, please reset the argument.")
gate_size = 3 * hidden_size
self.rnn = _DynamicGRUAscend() if is_ascend else _DynamicGRUCPUGPU()
elif mode == "RNN_TANH":
gate_size = hidden_size
self.rnn = _DynamicRNNTanh()
elif mode == "RNN_RELU":
gate_size = hidden_size
self.rnn = _DynamicRNNRelu()
else:
raise ValueError(f"For '{self.cls_name}', the 'mode' should be in ['RNN_RELU', 'RNN_TANH', 'LSTM', 'GRU'], "
f"but got {mode}.")
if context.get_context("device_target") == "CPU":
self.reverse = _Reverse(0)
self.reverse_sequence = _ReverseSequence(0, 1)
else:
self.reverse = P.ReverseV2([0])
self.reverse_sequence = P.ReverseSequence(0, 1)
self.hidden_size = hidden_size
self.batch_first = batch_first
self.num_layers = num_layers
self.dropout = dropout
self.dropout_op = nn.Dropout(float(1 - dropout))
self.bidirectional = bidirectional
self.has_bias = has_bias
num_directions = 2 if bidirectional else 1
self.is_lstm = mode == "LSTM"
self.w_ih_list = []
self.w_hh_list = []
self.b_ih_list = []
self.b_hh_list = []
stdv = 1 / math.sqrt(self.hidden_size)
for layer in range(num_layers):
for direction in range(num_directions):
layer_input_size = input_size if layer == 0 else hidden_size * num_directions
suffix = '_reverse' if direction == 1 else ''
self.w_ih_list.append(Parameter(
Tensor(np.random.uniform(-stdv, stdv, (gate_size, layer_input_size)).astype(np.float32)),
name='weight_ih_l{}{}'.format(layer, suffix)))
self.w_hh_list.append(Parameter(
Tensor(np.random.uniform(-stdv, stdv, (gate_size, hidden_size)).astype(np.float32)),
name='weight_hh_l{}{}'.format(layer, suffix)))
if has_bias:
self.b_ih_list.append(Parameter(
Tensor(np.random.uniform(-stdv, stdv, (gate_size)).astype(np.float32)),
name='bias_ih_l{}{}'.format(layer, suffix)))
self.b_hh_list.append(Parameter(
Tensor(np.random.uniform(-stdv, stdv, (gate_size)).astype(np.float32)),
name='bias_hh_l{}{}'.format(layer, suffix)))
self.w_ih_list = ParameterTuple(self.w_ih_list)
self.w_hh_list = ParameterTuple(self.w_hh_list)
self.b_ih_list = ParameterTuple(self.b_ih_list)
self.b_hh_list = ParameterTuple(self.b_hh_list)
def _stacked_bi_dynamic_rnn(self, x, h, seq_length):
"""stacked bidirectional dynamic_rnn"""
pre_layer = x
h_n = ()
c_n = ()
output = 0
for i in range(self.num_layers):
offset = i * 2
if self.has_bias:
w_f_ih, w_f_hh, b_f_ih, b_f_hh = \
self.w_ih_list[offset], self.w_hh_list[offset], \
self.b_ih_list[offset], self.b_hh_list[offset]
w_b_ih, w_b_hh, b_b_ih, b_b_hh = \
self.w_ih_list[offset + 1], self.w_hh_list[offset + 1], \
self.b_ih_list[offset + 1], self.b_hh_list[offset + 1]
else:
w_f_ih, w_f_hh = self.w_ih_list[offset], self.w_hh_list[offset]
w_b_ih, w_b_hh = self.w_ih_list[offset + 1], self.w_hh_list[offset + 1]
b_f_ih, b_f_hh, b_b_ih, b_b_hh = None, None, None, None
if self.is_lstm:
h_f_i = (h[0][offset], h[1][offset])
h_b_i = (h[0][offset + 1], h[1][offset + 1])
else:
h_f_i = h[offset]
h_b_i = h[offset + 1]
if seq_length is None:
x_b = self.reverse(pre_layer)
else:
x_b = self.reverse_sequence(pre_layer, seq_length)
output_f, h_t_f = self.rnn(pre_layer, h_f_i, seq_length, w_f_ih, w_f_hh, b_f_ih, b_f_hh)
output_b, h_t_b = self.rnn(x_b, h_b_i, seq_length, w_b_ih, w_b_hh, b_b_ih, b_b_hh)
if seq_length is None:
output_b = self.reverse(output_b)
else:
output_b = self.reverse_sequence(output_b, seq_length)
output = P.Concat(2)((output_f, output_b))
pre_layer = self.dropout_op(output) if (self.dropout != 0 and i < self.num_layers - 1) else output
if self.is_lstm:
h_n += (h_t_f[0], h_t_b[0],)
c_n += (h_t_f[1], h_t_b[1],)
else:
h_n += (h_t_f, h_t_b,)
if self.is_lstm:
h_n = P.Concat(0)(h_n)
c_n = P.Concat(0)(c_n)
h_n = h_n.view(h[0].shape)
c_n = c_n.view(h[1].shape)
return output, (h_n.view(h[0].shape), c_n.view(h[1].shape))
h_n = P.Concat(0)(h_n)
return output, h_n.view(h.shape)
def _stacked_dynamic_rnn(self, x, h, seq_length):
"""stacked mutil_layer dynamic_rnn"""
pre_layer = x
h_n = ()
c_n = ()
output = 0
for i in range(self.num_layers):
if self.has_bias:
w_ih, w_hh, b_ih, b_hh = self.w_ih_list[i], self.w_hh_list[i], self.b_ih_list[i], self.b_hh_list[i]
else:
w_ih, w_hh = self.w_ih_list[i], self.w_hh_list[i]
b_ih, b_hh = None, None
if self.is_lstm:
h_i = (h[0][i], h[1][i])
else:
h_i = h[i]
output, h_t = self.rnn(pre_layer, h_i, seq_length, w_ih, w_hh, b_ih, b_hh)
pre_layer = self.dropout_op(output) if (self.dropout != 0 and i < self.num_layers - 1) else output
if self.is_lstm:
h_n += (h_t[0],)
c_n += (h_t[1],)
else:
h_n += (h_t,)
if self.is_lstm:
h_n = P.Concat(0)(h_n)
c_n = P.Concat(0)(c_n)
h_n = h_n.view(h[0].shape)
c_n = c_n.view(h[1].shape)
return output, (h_n.view(h[0].shape), c_n.view(h[1].shape))
h_n = P.Concat(0)(h_n)
return output, h_n.view(h.shape)
def construct(self, x, hx=None, seq_length=None):
'''Defines the RNN like operators performed'''
max_batch_size = x.shape[0] if self.batch_first else x.shape[1]
num_directions = 2 if self.bidirectional else 1
_check_is_tensor("x", x, self.cls_name)
x_dtype = x.dtype
if hx is not None:
if not self.is_lstm:
_check_is_tensor("h", hx, self.cls_name)
_check_input_dtype_same_and_valid(['x', 'hx'], [x_dtype, hx.dtype], \
[mstype.float32, mstype.float16], self.cls_name)
else:
_check_is_tuple('hx', hx, self.cls_name)
_check_tuple_length('hx', hx, 2, self.cls_name)
_check_is_tensor('hx[0]', hx[0], self.cls_name)
_check_is_tensor('hx[1]', hx[1], self.cls_name)
_check_input_dtype_same_and_valid(['x', 'hx[0]', 'hx[1]'], [x_dtype, hx[0].dtype, hx[1].dtype], \
[mstype.float32, mstype.float16], self.cls_name)
else:
hx = _init_state((self.num_layers * num_directions, max_batch_size, self.hidden_size), \
x_dtype, self.is_lstm)
if seq_length is not None:
_check_input_dtype(seq_length.dtype, "seq_length", [mstype.int32, mstype.int64], self.cls_name)
_check_seq_length_size(max_batch_size, seq_length.shape[0], self.cls_name)
if self.batch_first:
x = P.Transpose()(x, (1, 0, 2))
if self.bidirectional:
x_n, hx_n = self._stacked_bi_dynamic_rnn(x, hx, seq_length)
else:
x_n, hx_n = self._stacked_dynamic_rnn(x, hx, seq_length)
if self.batch_first:
x_n = P.Transpose()(x_n, (1, 0, 2))
if not self.is_lstm:
return x_n.astype(x_dtype), hx_n.astype(x_dtype)
return x_n.astype(x_dtype), (hx_n[0].astype(x_dtype), hx_n[1].astype(x_dtype))
[docs]class RNN(_RNNBase):
r"""
Stacked Elman RNN layers.
Apply RNN layer with :math:`\tanh` or :math:`\text{ReLU}` non-linearity to the input.
For each element in the input sequence, each layer computes the following function:
.. math::
h_t = activation(W_{ih} x_t + b_{ih} + W_{hh} h_{(t-1)} + b_{hh})
Here :math:`h_t` is the hidden state at time `t`, :math:`x_t` is
the input at time `t`, and :math:`h_{(t-1)}` is the hidden state of the
previous layer at time `t-1` or the initial hidden state at time `0`.
If :attr:`nonlinearity` is ``'relu'``, then :math:`\text{ReLU}` is used instead of :math:`\tanh`.
Args:
input_size (int): Number of features of input.
hidden_size (int): Number of features of hidden layer.
num_layers (int): Number of layers of stacked RNN. Default: 1.
nonlinearity (str): The non-linearity to use. Can be either ``'tanh'`` or ``'relu'``. Default: ``'tanh'``
has_bias (bool): Whether the cell has bias `b_ih` and `b_hh`. Default: True.
batch_first (bool): Specifies whether the first dimension of input `x` is batch_size. Default: False.
dropout (float): If not 0.0, append `Dropout` layer on the outputs of each
RNN layer except the last layer. Default 0.0. The range of dropout is [0.0, 1.0).
bidirectional (bool): Specifies whether it is a bidirectional RNN,
num_directions=2 if bidirectional=True otherwise 1. Default: False.
Inputs:
- **x** (Tensor) - Tensor of data type mindspore.float32 or mindspore.float16 and
shape :math:`(seq\_len, batch\_size, input\_size)` or :math:`(batch\_size, seq\_len, input\_size)` .
- **hx** (Tensor) - Tensor of data type mindspore.float32 or mindspore.float16 and
shape :math:`(num\_directions * num\_layers, batch\_size, hidden\_size)` .
The data type of `hx` must be the same as `x`.
- **seq_length** (Tensor) - The length of each sequence in an input batch.
Tensor of shape :math:`(batch\_size)` . Default: None.
This input indicates the real sequence length before padding to avoid padded elements
have been used to compute hidden state and affect the final output. It is recommended to
use this input when `x` has padding elements.
Outputs:
Tuple, a tuple contains (`output`, `hx_n`).
- **output** (Tensor) - Tensor of shape :math:`(seq\_len, batch\_size, num\_directions * hidden\_size)` or
:math:`(batch\_size, seq\_len, num\_directions * hidden\_size)` .
- **hx_n** (Tensor) - Tensor of shape :math:`(num\_directions * num\_layers, batch\_size, hidden\_size)` .
Raises:
TypeError: If `input_size`, `hidden_size` or `num_layers` is not an int.
TypeError: If `has_bias`, `batch_first` or `bidirectional` is not a bool.
TypeError: If `dropout` is not a float.
ValueError: If `dropout` is not in range [0.0, 1.0).
ValueError: If `nonlinearity` is not in ['tanh', 'relu'].
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> net = nn.RNN(10, 16, 2, has_bias=True, batch_first=True, bidirectional=False)
>>> x = Tensor(np.ones([3, 5, 10]).astype(np.float32))
>>> h0 = Tensor(np.ones([1 * 2, 3, 16]).astype(np.float32))
>>> output, hn = net(x, h0)
>>> print(output.shape)
(3, 5, 16)
"""
def __init__(self, *args, **kwargs):
if 'nonlinearity' in kwargs:
if kwargs['nonlinearity'] == 'tanh':
mode = 'RNN_TANH'
elif kwargs['nonlinearity'] == 'relu':
mode = 'RNN_RELU'
else:
raise ValueError(f"For '{self.cls_name}', the 'nonlinearity' should be in ['tanh', 'relu'], "
f"but got {kwargs['nonlinearity']}.")
del kwargs['nonlinearity']
else:
mode = 'RNN_TANH'
super(RNN, self).__init__(mode, *args, **kwargs)
[docs]class GRU(_RNNBase):
r"""
Stacked GRU (Gated Recurrent Unit) layers.
Apply GRU layer to the input.
There are two gates in a GRU model; one is update gate and the other is reset gate.
Denote two consecutive time nodes as :math:`t-1` and :math:`t`.
Given an input :math:`x_t` at time :math:`t`, a hidden state :math:`h_{t-1}`, the update and reset gate at
time :math:`t` is computed using a gating mechanism. Update gate :math:`z_t` is designed to protect the cell
from perturbation by irrelevant inputs and past hidden state. Reset gate :math:`r_t` determines how much
information should be reset from old hidden state. New memory state :math:`{n}_t` is
calculated with the current input, on which the reset gate will be applied. Finally, current hidden state
:math:`h_{t}` is computed with the calculated update grate and new memory state. The complete
formulation is as follows.
.. math::
\begin{array}{ll}
r_t = \sigma(W_{ir} x_t + b_{ir} + W_{hr} h_{(t-1)} + b_{hr}) \\
z_t = \sigma(W_{iz} x_t + b_{iz} + W_{hz} h_{(t-1)} + b_{hz}) \\
n_t = \tanh(W_{in} x_t + b_{in} + r_t * (W_{hn} h_{(t-1)}+ b_{hn})) \\
h_t = (1 - z_t) * n_t + z_t * h_{(t-1)}
\end{array}
Here :math:`\sigma` is the sigmoid function, and :math:`*` is the Hadamard product. :math:`W, b`
are learnable weights between the output and the input in the formula. For instance,
:math:`W_{ir}, b_{ir}` are the weight and bias used to transform from input :math:`x` to :math:`r`.
Details can be found in paper
`Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation
<https://aclanthology.org/D14-1179.pdf>`_.
Note:
When using GRU on Ascend, the hidden size only supports multiples of 16.
Args:
input_size (int): Number of features of input.
hidden_size (int): Number of features of hidden layer.
num_layers (int): Number of layers of stacked GRU. Default: 1.
has_bias (bool): Whether the cell has bias `b_in` and `b_hn`. Default: True.
batch_first (bool): Specifies whether the first dimension of input `x` is batch_size. Default: False.
dropout (float): If not 0.0, append `Dropout` layer on the outputs of each
GRU layer except the last layer. Default 0.0. The range of dropout is [0.0, 1.0).
bidirectional (bool): Specifies whether it is a bidirectional GRU,
num_directions=2 if bidirectional=True otherwise 1. Default: False.
Inputs:
- **x** (Tensor) - Tensor of data type mindspore.float32 or mindspore.float16 and
shape (seq_len, batch_size, `input_size`) or (batch_size, seq_len, `input_size`).
- **hx** (Tensor) - Tensor of data type mindspore.float32 or mindspore.float16 and
shape (num_directions * `num_layers`, batch_size, `hidden_size`). The data type of `hx` must be the same as
`x`.
- **seq_length** (Tensor) - The length of each sequence in an input batch.
Tensor of shape :math:`(\text{batch_size})`. Default: None.
This input indicates the real sequence length before padding to avoid padded elements
have been used to compute hidden state and affect the final output. It is recommended to
use this input when **x** has padding elements.
Outputs:
Tuple, a tuple contains (`output`, `h_n`).
- **output** (Tensor) - Tensor of shape (seq_len, batch_size, num_directions * `hidden_size`) or
(batch_size, seq_len, num_directions * `hidden_size`).
- **hx_n** (Tensor) - Tensor of shape (num_directions * `num_layers`, batch_size, `hidden_size`).
Raises:
TypeError: If `input_size`, `hidden_size` or `num_layers` is not an int.
TypeError: If `has_bias`, `batch_first` or `bidirectional` is not a bool.
TypeError: If `dropout` is not a float.
ValueError: If `dropout` is not in range [0.0, 1.0).
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> net = nn.GRU(10, 16, 2, has_bias=True, batch_first=True, bidirectional=False)
>>> x = Tensor(np.ones([3, 5, 10]).astype(np.float32))
>>> h0 = Tensor(np.ones([1 * 2, 3, 16]).astype(np.float32))
>>> output, hn = net(x, h0)
>>> print(output.shape)
(3, 5, 16)
"""
def __init__(self, *args, **kwargs):
mode = 'GRU'
super(GRU, self).__init__(mode, *args, **kwargs)
[docs]class LSTM(_RNNBase):
r"""
Stacked LSTM (Long Short-Term Memory) layers.
Apply LSTM layer to the input.
There are two pipelines connecting two consecutive cells in a LSTM model; one is cell state pipeline
and the other is hidden state pipeline. Denote two consecutive time nodes as :math:`t-1` and :math:`t`.
Given an input :math:`x_t` at time :math:`t`, an hidden state :math:`h_{t-1}` and an cell
state :math:`c_{t-1}` of the layer at time :math:`{t-1}`, the cell state and hidden state at
time :math:`t` is computed using an gating mechanism. Input gate :math:`i_t` is designed to protect the cell
from perturbation by irrelevant inputs. Forget gate :math:`f_t` affords protection of the cell by forgetting
some information in the past, which is stored in :math:`h_{t-1}`. Output gate :math:`o_t` protects other
units from perturbation by currently irrelevant memory contents. Candidate cell state :math:`\tilde{c}_t` is
calculated with the current input, on which the input gate will be applied. Finally, current cell state
:math:`c_{t}` and hidden state :math:`h_{t}` are computed with the calculated gates and cell states. The complete
formulation is as follows.
.. math::
\begin{array}{ll} \\
i_t = \sigma(W_{ix} x_t + b_{ix} + W_{ih} h_{(t-1)} + b_{ih}) \\
f_t = \sigma(W_{fx} x_t + b_{fx} + W_{fh} h_{(t-1)} + b_{fh}) \\
\tilde{c}_t = \tanh(W_{cx} x_t + b_{cx} + W_{ch} h_{(t-1)} + b_{ch}) \\
o_t = \sigma(W_{ox} x_t + b_{ox} + W_{oh} h_{(t-1)} + b_{oh}) \\
c_t = f_t * c_{(t-1)} + i_t * \tilde{c}_t \\
h_t = o_t * \tanh(c_t) \\
\end{array}
Here :math:`\sigma` is the sigmoid function, and :math:`*` is the Hadamard product. :math:`W, b`
are learnable weights between the output and the input in the formula. For instance,
:math:`W_{ix}, b_{ix}` are the weight and bias used to transform from input :math:`x` to :math:`i`.
Details can be found in paper `LONG SHORT-TERM MEMORY
<https://www.bioinf.jku.at/publications/older/2604.pdf>`_ and
`Long Short-Term Memory Recurrent Neural Network Architectures for Large Scale Acoustic Modeling
<https://static.googleusercontent.com/media/research.google.com/zh-CN//pubs/archive/43905.pdf>`_.
LSTM hides the cycle of the whole cyclic neural network on the time step of the sequence,
and input the sequence and initial state to obtain the matrix spliced by
the hidden state of each time step and the hidden state of the last time step.
We use the hidden state of the last time step as the coding feature of the input sentence and
output it to the next layer.
.. math::
h_{0:n},(h_{n}, c_{n}) = LSTM(x_{0:n},(h_{0},c_{0}))
Args:
input_size (int): Number of features of input.
hidden_size (int): Number of features of hidden layer.
num_layers (int): Number of layers of stacked LSTM . Default: 1.
has_bias (bool): Whether the cell has bias `b_ih` and `b_hh`. Default: True.
batch_first (bool): Specifies whether the first dimension of input `x` is batch_size. Default: False.
dropout (float, int): If not 0, append `Dropout` layer on the outputs of each
LSTM layer except the last layer. Default 0. The range of dropout is [0.0, 1.0).
bidirectional (bool): Specifies whether it is a bidirectional LSTM,
num_directions=2 if bidirectional=True otherwise 1. Default: False.
Inputs:
- **x** (Tensor) - Tensor of data type mindspore.float32 or mindspore.float16 and
shape (seq_len, batch_size, `input_size`) or (batch_size, seq_len, `input_size`).
- **hx** (tuple) - A tuple of two Tensors (h_0, c_0) both of data type mindspore.float32
or mindspore.float16 and shape (num_directions * `num_layers`, batch_size, `hidden_size`).
The data type of `hx` must be the same as `x`.
- **seq_length** (Tensor) - The length of each sequence in an input batch.
Tensor of shape :math:`(\text{batch_size})`. Default: None.
This input indicates the real sequence length before padding to avoid padded elements
have been used to compute hidden state and affect the final output. It is recommended to
use this input when **x** has padding elements.
Outputs:
Tuple, a tuple contains (`output`, (`h_n`, `c_n`)).
- **output** (Tensor) - Tensor of shape (seq_len, batch_size, num_directions * `hidden_size`).
- **hx_n** (tuple) - A tuple of two Tensor (h_n, c_n) both of shape
(num_directions * `num_layers`, batch_size, `hidden_size`).
Raises:
TypeError: If `input_size`, `hidden_size` or `num_layers` is not an int.
TypeError: If `has_bias`, `batch_first` or `bidirectional` is not a bool.
TypeError: If `dropout` is not a float.
ValueError: If `dropout` is not in range [0.0, 1.0).
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> net = nn.LSTM(10, 16, 2, has_bias=True, batch_first=True, bidirectional=False)
>>> x = Tensor(np.ones([3, 5, 10]).astype(np.float32))
>>> h0 = Tensor(np.ones([1 * 2, 3, 16]).astype(np.float32))
>>> c0 = Tensor(np.ones([1 * 2, 3, 16]).astype(np.float32))
>>> output, (hn, cn) = net(x, (h0, c0))
>>> print(output.shape)
(3, 5, 16)
"""
def __init__(self, *args, **kwargs):
mode = 'LSTM'
super(LSTM, self).__init__(mode, *args, **kwargs)