reduce ignores

ZeyuChen · Dec 25, 2017 · e73eb1d · e73eb1d
1 parent 1e53828
commit e73eb1d
Show file tree

Hide file tree

Showing 6 changed files with 47 additions and 53 deletions.
diff --git a/setup.cfg b/setup.cfg
@@ -4,7 +4,7 @@ testpaths = test
 python_paths = src/nets src/utils src/bin
 
 [flake8]
-ignore = H102,H301,H306,H404,H405
+ignore = H102,H306
 # 120 is a workaround, 79 is good
 max-line-length = 120
 exclude = src/utils
diff --git a/src/bin/asr_train.py b/src/bin/asr_train.py
@@ -218,9 +218,7 @@ def delete_feat(batch):
 
 
 def adadelta_eps_decay(eps_decay):
-    '''
-    Extension to perform adadelta eps decay
-    '''
+    '''Extension to perform adadelta eps decay'''
     @training.make_extension(trigger=(1, 'epoch'))
     def adadelta_eps_decay(trainer):
         _adadelta_eps_decay(trainer, eps_decay)
@@ -236,9 +234,7 @@ def _adadelta_eps_decay(trainer, eps_decay):
 
 
 def restore_snapshot(model, snapshot, load_fn=chainer.serializers.load_npz):
-    '''
-    Extension to restore snapshot
-    '''
+    '''Extension to restore snapshot'''
     @training.make_extension(trigger=(1, 'epoch'))
     def restore_snapshot(trainer):
         _restore_snapshot(model, snapshot, load_fn)

diff --git a/src/bin/asr_train_th.py b/src/bin/asr_train_th.py
@@ -121,14 +121,14 @@ def update_core(self):
 
 # Custom trigger
 class CompareValueTrigger(object):
-    '''Trigger invoked when key value getting bigger or lower than before
+    '''\Trigger invoked when key value getting bigger or lower than before
 
     Args:
         key (str): Key of value.
         compare_fn: Function to compare the values.
         trigger: Trigger that decide the comparison interval
 
-    '''
+    '''\
 
     def __init__(self, key, compare_fn, trigger=(1, 'epoch')):
         self._key = key
@@ -222,9 +222,7 @@ def delete_feat(batch):
 
 
 def adadelta_eps_decay(eps_decay):
-    '''
-    Extension to perform adadelta eps decay
-    '''
+    '''Extension to perform adadelta eps decay'''
     @training.make_extension(trigger=(1, 'epoch'))
     def adadelta_eps_decay(trainer):
         _adadelta_eps_decay(trainer, eps_decay)
@@ -240,9 +238,7 @@ def _adadelta_eps_decay(trainer, eps_decay):
 
 
 def restore_snapshot(model, snapshot, load_fn=chainer.serializers.load_npz):
-    '''
-    Extension to restore snapshot
-    '''
+    '''Extension to restore snapshot'''
     @training.make_extension(trigger=(1, 'epoch'))
     def restore_snapshot(trainer):
         _restore_snapshot(model, snapshot, load_fn)

diff --git a/src/nets/e2e_asr_attctc.py b/src/nets/e2e_asr_attctc.py
@@ -42,8 +42,7 @@ def _get_vgg2l_odim(idim, in_channel=3, out_channel=128):
 
 
 def linear_tensor(linear, x):
-    '''
-    Apply linear matrix operation only for the last dimension of a tensor
+    '''Apply linear matrix operation only for the last dimension of a tensor
 
     :param Link linear: Linear link (M x N matrix)
     :param Variable x: Tensor (D_1 x D_2 x ... x M matrix)
@@ -72,7 +71,7 @@ def __init__(self, predictor, mtlalpha):
             self.predictor = predictor
 
     def __call__(self, x):
-        '''
+        '''Loss forward
 
         :param x:
         :return:
@@ -143,7 +142,7 @@ def __init__(self, idim, odim, args):
 
     # x[i]: ('utt_id', {'ilen':'xxx',...}})
     def __call__(self, data):
-        '''
+        '''E2E forward
 
         :param data:
         :return:
@@ -184,7 +183,7 @@ def __call__(self, data):
         return loss_ctc, loss_att, acc
 
     def recognize(self, x, recog_args, char_list):
-        '''
+        '''E2E greedy/beam search
 
         :param x:
         :param recog_args:
@@ -222,7 +221,7 @@ def __init__(self, odim, eprojs, dropout_rate):
             self.ctc_lo = L.Linear(eprojs, odim)
 
     def __call__(self, hs, ys):
-        '''
+        '''CTC forward
 
         :param hs:
         :param ys:
@@ -273,7 +272,7 @@ def __init__(self, eprojs, dunits, att_dim):
         self.pre_compute_enc_h = None
 
     def reset(self):
-        '''
+        '''reset states
 
         :return:
         '''
@@ -282,7 +281,7 @@ def reset(self):
         self.pre_compute_enc_h = None
 
     def __call__(self, enc_hs, dec_z, att_prev, scaling=2.0):
-        '''
+        '''AttDot forward
 
         :param enc_hs:
         :param dec_z:
@@ -337,7 +336,7 @@ def __init__(self, eprojs, dunits, att_dim, aconv_chans, aconv_filts):
         self.aconv_chans = aconv_chans
 
     def reset(self):
-        '''
+        '''reset states
 
         :return:
         '''
@@ -346,7 +345,7 @@ def reset(self):
         self.pre_compute_enc_h = None
 
     def __call__(self, enc_hs, dec_z, att_prev, scaling=2.0):
-        '''
+        '''AttLoc forward
 
         :param enc_hs:
         :param dec_z:
@@ -424,7 +423,7 @@ def __init__(self, eprojs, odim, dlayers, dunits, sos, eos, att, verbose=0, char
         self.char_list = char_list
 
     def __call__(self, hs, ys):
-        '''
+        '''Decoder forward
 
         :param hs:
         :param ys:
@@ -499,7 +498,7 @@ def __call__(self, hs, ys):
         return self.loss, acc, att_weight_all
 
     def recognize(self, h, recog_args):
-        '''
+        '''greedy search implementation
 
         :param h:
         :param recog_args:
@@ -534,7 +533,7 @@ def recognize(self, h, recog_args):
         return y_seq
 
     def recognize_beam(self, h, recog_args, char_list):
-        '''
+        '''beam search implementation
 
         :param h:
         :param recog_args:
@@ -693,7 +692,7 @@ def __init__(self, etype, idim, elayers, eunits, eprojs, subsample, dropout, in_
         self.etype = etype
 
     def __call__(self, xs, ilens):
-        '''
+        '''Encoder forward
 
         :param xs:
         :param ilens:
@@ -737,7 +736,7 @@ def __init__(self, idim, elayers, cdim, hdim, subsample, dropout):
         self.subsample = subsample
 
     def __call__(self, xs, ilens):
-        '''
+        '''BLSTMP forward
 
         :param xs:
         :param ilens:
@@ -773,7 +772,7 @@ def __init__(self, idim, elayers, cdim, hdim, dropout):
             self.l_last = L.Linear(cdim * 2, hdim)
 
     def __call__(self, xs, ilens):
-        '''
+        '''BLSTM forward
 
         :param xs:
         :param ilens:
@@ -809,7 +808,7 @@ def __init__(self, in_channel=1):
         self.in_channel = in_channel
 
     def __call__(self, xs, ilens):
-        '''
+        '''VGG2L forward
 
         :param xs:
         :param ilens:

diff --git a/src/nets/e2e_asr_attctc_th.py b/src/nets/e2e_asr_attctc_th.py
@@ -7,7 +7,8 @@
 import torch
 from torch.nn import functional
 from torch.autograd import Variable
-from torch.nn.utils.rnn import pad_packed_sequence, pack_padded_sequence
+from torch.nn.utils.rnn import pad_packed_sequence
+from torch.nn.utils.rnn import pack_padded_sequence
 from warpctc_pytorch import _CTC
 import sys
 import logging
@@ -76,8 +77,7 @@ def _get_max_pooled_size(idim, out_channel=128, n_layers=2, ksize=2, stride=2):
 
 
 def linear_tensor(linear, x):
-    '''
-    Apply linear matrix operation only for the last dimension of a tensor
+    '''Apply linear matrix operation only for the last dimension of a tensor
 
     :param Link linear: Linear link (M x N matrix)
     :param Variable x: Tensor (D_1 x D_2 x ... x M matrix)
@@ -113,7 +113,7 @@ def __init__(self, predictor, mtlalpha):
         self.reporter = Reporter()
 
     def forward(self, x):
-        '''
+        '''Loss forward
 
         :param x:
         :return:
@@ -204,7 +204,8 @@ def __init__(self, idim, odim, args):
         #                 set_forget_bias_to_one(p)
 
     def init_like_chainer(self):
-        """
+        """Initialize weight like chainer
+
         chainer basically uses LeCun way: W ~ Normal(0, fan_in ** -0.5), b = 0
         pytorch basically uses W, b ~ Uniform(-fan_in**-0.5, fan_in**-0.5)
 
@@ -223,7 +224,7 @@ def init_like_chainer(self):
 
     # x[i]: ('utt_id', {'ilen':'xxx',...}})
     def forward(self, data):
-        '''
+        '''E2E forward
 
         :param data:
         :return:
@@ -271,7 +272,7 @@ def forward(self, data):
         return loss_ctc, loss_att, acc
 
     def recognize(self, x, recog_args, char_list):
-        '''
+        '''E2E greedy/beam search
 
         :param x:
         :param recog_args:
@@ -312,7 +313,8 @@ def backward(self, grad_output):
 
 
 def chainer_like_ctc_loss(acts, labels, act_lens, label_lens):
-    """
+    """Chainer like CTC Loss
+
     acts: Tensor of (seqLength x batch x outputDim) containing output from network
     labels: 1 dimensional Tensor containing all the targets of the batch in one sequence
     act_lens: Tensor of size (batch) containing size of each output sequence from the network
@@ -335,7 +337,7 @@ def __init__(self, odim, eprojs, dropout_rate):
         self.loss_fn = chainer_like_ctc_loss  # CTCLoss()
 
     def forward(self, hpad, ilens, ys):
-        '''
+        '''CTC forward
 
         :param hs:
         :param ys:
@@ -392,7 +394,7 @@ def __init__(self, eprojs, dunits, att_dim):
         self.pre_compute_enc_h = None
 
     def reset(self):
-        '''
+        '''reset states
 
         :return:
         '''
@@ -401,7 +403,7 @@ def reset(self):
         self.pre_compute_enc_h = None
 
     def forward(self, enc_hs_pad, enc_hs_len, dec_z, att_prev, scaling=2.0):
-        '''
+        '''AttDot
 
         :param enc_hs:
         :param dec_z:
@@ -455,7 +457,7 @@ def __init__(self, eprojs, dunits, att_dim, aconv_chans, aconv_filts):
         self.aconv_chans = aconv_chans
 
     def reset(self):
-        '''
+        '''reset states
 
         :return:
         '''
@@ -464,7 +466,7 @@ def reset(self):
         self.pre_compute_enc_h = None
 
     def forward(self, enc_hs_pad, enc_hs_len, dec_z, att_prev, scaling=2.0):
-        '''
+        '''AttLoc forward
 
         :param enc_hs:
         :param dec_z:
@@ -551,7 +553,7 @@ def zero_state(self, hpad):
         return Variable(hpad.data.new(hpad.size(0), self.dunits).zero_())
 
     def forward(self, hpad, hlen, ys):
-        '''
+        '''Decoder forward
 
         :param hs:
         :param ys:
@@ -627,7 +629,7 @@ def forward(self, hpad, hlen, ys):
         return self.loss, acc, att_weight_all
 
     def recognize(self, h, recog_args):
-        '''
+        '''greedy search implementation
 
         :param h:
         :param recog_args:
@@ -666,7 +668,7 @@ def recognize(self, h, recog_args):
         return y_seq
 
     def recognize_beam(self, h, recog_args, char_list):
-        '''
+        '''beam search implementation
 
         :param h:
         :param recog_args:
@@ -829,7 +831,7 @@ def __init__(self, etype, idim, elayers, eunits, eprojs, subsample, dropout, in_
         self.etype = etype
 
     def forward(self, xs, ilens):
-        '''
+        '''Encoder forward
 
         :param xs:
         :param ilens:
@@ -871,7 +873,7 @@ def __init__(self, idim, elayers, cdim, hdim, subsample, dropout):
         self.subsample = subsample
 
     def forward(self, xpad, ilens):
-        '''
+        '''BLSTMP forward
 
         :param xs:
         :param ilens:
@@ -904,7 +906,7 @@ def __init__(self, idim, elayers, cdim, hdim, dropout):
         self.l_last = torch.nn.Linear(cdim * 2, hdim)
 
     def forward(self, xpad, ilens):
-        '''
+        '''BLSTM forward
 
         :param xs:
         :param ilens:
@@ -935,7 +937,7 @@ def __init__(self, in_channel=1):
         self.in_channel = in_channel
 
     def forward(self, xs, ilens):
-        '''
+        '''VGG2L forward
 
         :param xs:
         :param ilens: