From e0c8c53c251cc8a950df91bb8e373d66a7224ccd Mon Sep 17 00:00:00 2001
From: glynpu <839019390@qq.com>
Date: Wed, 25 Nov 2020 13:13:38 +0800
Subject: [PATCH] [doc]: fix typo and remove unused variable

---
 wenet/transformer/asr_model.py     |  2 --
 wenet/transformer/attention.py     | 12 +++++-------
 wenet/transformer/encoder_layer.py |  8 +++-----
 3 files changed, 8 insertions(+), 14 deletions(-)

diff --git a/wenet/transformer/asr_model.py b/wenet/transformer/asr_model.py
index db659139e..8a756d5ef 100644
--- a/wenet/transformer/asr_model.py
+++ b/wenet/transformer/asr_model.py
@@ -229,7 +229,6 @@ def ctc_greedy_search(self,
         '''
         assert speech.shape[0] == speech_lengths.shape[0]
         assert decoding_chunk_size != 0
-        device = speech.device
         batch_size = speech.shape[0]
         # Let's assume B = batch_size
         encoder_out, encoder_mask = self.encoder(
@@ -266,7 +265,6 @@ def _ctc_prefix_beam_search(
         '''
         assert speech.shape[0] == speech_lengths.shape[0]
         assert decoding_chunk_size != 0
-        device = speech.device
         batch_size = speech.shape[0]
         # For CTC prefix beam search, we only support batch_size=1
         assert batch_size == 1
diff --git a/wenet/transformer/attention.py b/wenet/transformer/attention.py
index 68ae55fb1..0df2f2336 100644
--- a/wenet/transformer/attention.py
+++ b/wenet/transformer/attention.py
@@ -22,7 +22,6 @@ class MultiHeadedAttention(nn.Module):
         dropout_rate (float): Dropout rate.
 
     """
-
     def __init__(self, n_head: int, n_feat: int, dropout_rate: float):
         """Construct an MultiHeadedAttention object."""
         super().__init__()
@@ -37,7 +36,7 @@ def __init__(self, n_head: int, n_feat: int, dropout_rate: float):
         self.dropout = nn.Dropout(p=dropout_rate)
 
     def forward_qkv(
-            self, query: torch.Tensor, key: torch.Tensor, value: torch.Tensor
+        self, query: torch.Tensor, key: torch.Tensor, value: torch.Tensor
     ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
         """Transform query, key and value.
 
@@ -87,8 +86,9 @@ def forward_attention(self, value: torch.Tensor, scores: torch.Tensor,
 
         p_attn = self.dropout(attn)
         x = torch.matmul(p_attn, value)  # (batch, head, time1, d_k)
-        x = (x.transpose(1, 2).contiguous().view(n_batch, -1, self.h * self.d_k)
-            )  # (batch, time1, d_model)
+        x = (x.transpose(1, 2).contiguous().view(n_batch, -1,
+                                                 self.h * self.d_k)
+             )  # (batch, time1, d_model)
 
         return self.linear_out(x)  # (batch, time1, d_model)
 
@@ -121,11 +121,10 @@ class RelPositionMultiHeadedAttention(MultiHeadedAttention):
         n_feat (int): The number of features.
         dropout_rate (float): Dropout rate.
     """
-
     def __init__(self, n_head, n_feat, dropout_rate):
         """Construct an RelPositionMultiHeadedAttention object."""
         super().__init__(n_head, n_feat, dropout_rate)
-        # linear transformation for positional ecoding
+        # linear transformation for positional encoding
         self.linear_pos = nn.Linear(n_feat, n_feat, bias=False)
         # these two learnable bias are used in matrix c and matrix d
         # as described in https://arxiv.org/abs/1901.02860 Section 3.3
@@ -142,7 +141,6 @@ def rel_shift(self, x, zero_triu: bool = False):
         Returns:
             torch.Tensor: Output tensor.
         """
-        #print(x.size()[1])
 
         zero_pad = torch.zeros((x.size()[0], x.size()[1], x.size()[2], 1),
                                device=x.device,
diff --git a/wenet/transformer/encoder_layer.py b/wenet/transformer/encoder_layer.py
index e8fc3564e..bbf58d9ad 100644
--- a/wenet/transformer/encoder_layer.py
+++ b/wenet/transformer/encoder_layer.py
@@ -10,6 +10,7 @@
 import torch
 from torch import nn
 
+
 class TransformerEncoderLayer(nn.Module):
     """Encoder layer module.
 
@@ -29,7 +30,6 @@ class TransformerEncoderLayer(nn.Module):
             if False, no additional linear will be applied. i.e. x -> x + att(x)
 
     """
-
     def __init__(
         self,
         size: int,
@@ -125,7 +125,6 @@ class ConformerEncoderLayer(nn.Module):
             i.e. x -> x + linear(concat(x, att(x)))
             if False, no additional linear will be applied. i.e. x -> x + att(x)
     """
-
     def __init__(
         self,
         size: int,
@@ -151,7 +150,8 @@ def __init__(
         else:
             self.ff_scale = 1.0
         if self.conv_module is not None:
-            self.norm_conv = nn.LayerNorm(size, eps=1e-12)  # for the CNN module
+            self.norm_conv = nn.LayerNorm(size,
+                                          eps=1e-12)  # for the CNN module
             self.norm_final = nn.LayerNorm(
                 size, eps=1e-12)  # for the final output of the block
         self.dropout = nn.Dropout(dropout_rate)
@@ -241,6 +241,4 @@ def forward(
         if cache is not None:
             x = torch.cat([cache, x], dim=1)
 
-        # if pos_emb is not None:
         return (x, pos_emb), mask
-