Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
PaddlePaddle
DeepSpeech
提交
38174c70
D
DeepSpeech
项目概览
PaddlePaddle
/
DeepSpeech
大约 1 年 前同步成功
通知
207
Star
8425
Fork
1598
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
245
列表
看板
标记
里程碑
合并请求
3
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
D
DeepSpeech
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
245
Issue
245
列表
看板
标记
里程碑
合并请求
3
合并请求
3
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
体验新版 GitCode,发现更多精彩内容 >>
未验证
提交
38174c70
编写于
9月 02, 2021
作者:
J
Jackwaterveg
提交者:
GitHub
9月 02, 2021
浏览文件
操作
浏览文件
下载
差异文件
Merge pull request #802 from PaddlePaddle/fix_ds2_bw_bug
fix the bug of sharing cell in BiGRU and BIRNN
上级
0d90d3f9
f54dc983
变更
6
隐藏空白更改
内联
并排
Showing
6 changed file
with
22 addition
and
15 deletion
+22
-15
README.md
README.md
+1
-1
README_cn.md
README_cn.md
+1
-1
deepspeech/models/ds2/rnn.py
deepspeech/models/ds2/rnn.py
+7
-7
deepspeech/modules/ctc.py
deepspeech/modules/ctc.py
+7
-3
examples/aishell/s0/README.md
examples/aishell/s0/README.md
+4
-1
utils/avg.sh
utils/avg.sh
+2
-2
未找到文件。
README.md
浏览文件 @
38174c70
...
...
@@ -18,7 +18,7 @@
All tested under:
*
Ubuntu 16.04
*
python>=3.7
*
paddlepaddle>=2.
1.2
*
paddlepaddle>=2.
2.0rc
Please see
[
install
](
doc/src/install.md
)
.
...
...
README_cn.md
浏览文件 @
38174c70
...
...
@@ -20,7 +20,7 @@
*
Ubuntu 16.04
*
python>=3.7
*
paddlepaddle>=2.
1.2
*
paddlepaddle>=2.
2.0rc
参看
[
安装
](
doc/src/install.md
)
。
...
...
deepspeech/models/ds2/rnn.py
浏览文件 @
38174c70
...
...
@@ -29,13 +29,13 @@ __all__ = ['RNNStack']
class
RNNCell
(
nn
.
RNNCellBase
):
r
"""
Elman RNN (SimpleRNN) cell. Given the inputs and previous states, it
Elman RNN (SimpleRNN) cell. Given the inputs and previous states, it
computes the outputs and updates states.
The formula used is as follows:
.. math::
h_{t} & = act(x_{t} + b_{ih} + W_{hh}h_{t-1} + b_{hh})
y_{t} & = h_{t}
where :math:`act` is for :attr:`activation`.
"""
...
...
@@ -92,7 +92,7 @@ class RNNCell(nn.RNNCellBase):
class
GRUCell
(
nn
.
RNNCellBase
):
r
"""
Gated Recurrent Unit (GRU) RNN cell. Given the inputs and previous states,
Gated Recurrent Unit (GRU) RNN cell. Given the inputs and previous states,
it computes the outputs and updates states.
The formula for GRU used is as follows:
.. math::
...
...
@@ -101,8 +101,8 @@ class GRUCell(nn.RNNCellBase):
\widetilde{h}_{t} & = \tanh(W_{ic}x_{t} + b_{ic} + r_{t} * (W_{hc}h_{t-1} + b_{hc}))
h_{t} & = z_{t} * h_{t-1} + (1 - z_{t}) * \widetilde{h}_{t}
y_{t} & = h_{t}
where :math:`\sigma` is the sigmoid fucntion, and * is the elemetwise
where :math:`\sigma` is the sigmoid fucntion, and * is the elemetwise
multiplication operator.
"""
...
...
@@ -202,7 +202,7 @@ class BiRNNWithBN(nn.Layer):
self
.
fw_rnn
=
nn
.
RNN
(
self
.
fw_cell
,
is_reverse
=
False
,
time_major
=
False
)
#[B, T, D]
self
.
bw_rnn
=
nn
.
RNN
(
self
.
f
w_cell
,
is_reverse
=
True
,
time_major
=
False
)
#[B, T, D]
self
.
b
w_cell
,
is_reverse
=
True
,
time_major
=
False
)
#[B, T, D]
def
forward
(
self
,
x
:
paddle
.
Tensor
,
x_len
:
paddle
.
Tensor
):
# x, shape [B, T, D]
...
...
@@ -246,7 +246,7 @@ class BiGRUWithBN(nn.Layer):
self
.
fw_rnn
=
nn
.
RNN
(
self
.
fw_cell
,
is_reverse
=
False
,
time_major
=
False
)
#[B, T, D]
self
.
bw_rnn
=
nn
.
RNN
(
self
.
f
w_cell
,
is_reverse
=
True
,
time_major
=
False
)
#[B, T, D]
self
.
b
w_cell
,
is_reverse
=
True
,
time_major
=
False
)
#[B, T, D]
def
forward
(
self
,
x
,
x_len
):
# x, shape [B, T, D]
...
...
deepspeech/modules/ctc.py
浏览文件 @
38174c70
...
...
@@ -22,6 +22,13 @@ from deepspeech.utils.log import Log
logger
=
Log
(
__name__
).
getlog
()
try
:
from
deepspeech.decoders.swig_wrapper
import
ctc_beam_search_decoder_batch
# noqa: F401
from
deepspeech.decoders.swig_wrapper
import
ctc_greedy_decoder
# noqa: F401
from
deepspeech.decoders.swig_wrapper
import
Scorer
# noqa: F401
except
Exception
as
e
:
logger
.
info
(
"ctcdecoder not installed!"
)
__all__
=
[
'CTCDecoder'
]
...
...
@@ -216,9 +223,6 @@ class CTCDecoder(nn.Layer):
def
init_decode
(
self
,
beam_alpha
,
beam_beta
,
lang_model_path
,
vocab_list
,
decoding_method
):
from
deepspeech.decoders.swig_wrapper
import
ctc_beam_search_decoder_batch
# noqa: F401
from
deepspeech.decoders.swig_wrapper
import
ctc_greedy_decoder
# noqa: F401
from
deepspeech.decoders.swig_wrapper
import
Scorer
# noqa: F401
if
decoding_method
==
"ctc_beam_search"
:
self
.
_init_ext_scorer
(
beam_alpha
,
beam_beta
,
lang_model_path
,
...
...
examples/aishell/s0/README.md
浏览文件 @
38174c70
...
...
@@ -10,8 +10,11 @@
| Model | Params | Release | Config | Test set | Loss | CER |
| --- | --- | --- | --- | --- | --- | --- |
| DeepSpeech2 | 58.4M | 2.2.0 | conf/deepspeech2.yaml + spec aug | test | 5.71956205368042 | 0.064287 |
| DeepSpeech2 | 58.4M | 2.2.0 | conf/deepspeech2.yaml + spec aug | test | 6.016139030456543 | 0.066549 |
| --- | --- | --- | --- | --- | --- | --- |
| DeepSpeech2 | 58.4M | 7181e427 | conf/deepspeech2.yaml + spec aug | test | 5.71956205368042 | 0.064287 |
| DeepSpeech2 | 58.4M | 2.1.0 | conf/deepspeech2.yaml + spec aug | test | 7.483316898345947 | 0.077860 |
| DeepSpeech2 | 58.4M | 2.1.0 | conf/deepspeech2.yaml | test | 7.299022197723389 | 0.078671 |
| DeepSpeech2 | 58.4M | 2.0.0 | conf/deepspeech2.yaml | test | - | 0.078977 |
| --- | --- | --- | --- | --- | --- | --- |
| DeepSpeech2 | 58.4M | 1.8.5 | - | test | - | 0.080447 |
utils/avg.sh
浏览文件 @
38174c70
...
...
@@ -5,8 +5,8 @@ if [ $# != 3 ]; then
exit
-1
fi
ckpt_dir
=
${
1
}
avg_mode
=
${
2
}
# best,latest
avg_mode
=
${
1
}
# best,latest
ckpt_dir
=
${
2
}
average_num
=
${
3
}
decode_checkpoint
=
${
ckpt_dir
}
/avg_
${
average_num
}
.pdparams
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录