# MIT License
#
# Copyright (c) 2021 Soohwan Kim and Sangchun Ha and Soyoung Cho
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
import torch.nn as nn
import torch.nn.functional as F
from torch import Tensor
[docs]class BNReluRNN(nn.Module):
r"""
Recurrent neural network with batch normalization layer & ReLU activation function.
Args:
input_size (int): size of input
hidden_state_dim (int): the number of features in the hidden state `h`
rnn_type (str, optional): type of RNN cell (default: gru)
bidirectional (bool, optional): if True, becomes a bidirectional encoders (defulat: True)
dropout_p (float, optional): dropout probability (default: 0.1)
Inputs: inputs, input_lengths
- **inputs** (batch, time, dim): Tensor containing input vectors
- **input_lengths**: Tensor containing containing sequence lengths
Returns: outputs
- **outputs**: Tensor produced by the BNReluRNN module
"""
supported_rnns = {
'lstm': nn.LSTM,
'gru': nn.GRU,
'rnn': nn.RNN,
}
def __init__(
self,
input_size: int,
hidden_state_dim: int = 512,
rnn_type: str = 'gru',
bidirectional: bool = True,
dropout_p: float = 0.1,
):
super(BNReluRNN, self).__init__()
self.hidden_state_dim = hidden_state_dim
self.batch_norm = nn.BatchNorm1d(input_size)
rnn_cell = self.supported_rnns[rnn_type]
self.rnn = rnn_cell(
input_size=input_size,
hidden_size=hidden_state_dim,
num_layers=1,
bias=True,
batch_first=True,
dropout=dropout_p,
bidirectional=bidirectional,
)
def forward(self, inputs: Tensor, input_lengths: Tensor):
total_length = inputs.size(0)
inputs = F.relu(self.batch_norm(inputs.transpose(1, 2)))
inputs = inputs.transpose(1, 2)
outputs = nn.utils.rnn.pack_padded_sequence(inputs, input_lengths.cpu())
outputs, hidden_states = self.rnn(outputs)
outputs, _ = nn.utils.rnn.pad_packed_sequence(outputs, total_length=total_length)
return outputs