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
PaddleHub
提交
e3ee127f
P
PaddleHub
项目概览
PaddlePaddle
/
PaddleHub
大约 2 年 前同步成功
通知
285
Star
12117
Fork
2091
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
200
列表
看板
标记
里程碑
合并请求
4
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
PaddleHub
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
200
Issue
200
列表
看板
标记
里程碑
合并请求
4
合并请求
4
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
未验证
提交
e3ee127f
编写于
12月 29, 2022
作者:
jm_12138
提交者:
GitHub
12月 29, 2022
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
add gradio app (#2161)
上级
32dbb01a
变更
3
隐藏空白更改
内联
并排
Showing
3 changed file
with
67 addition
and
43 deletion
+67
-43
modules/text/machine_translation/transformer/zh-en/README.md
modules/text/machine_translation/transformer/zh-en/README.md
+12
-4
modules/text/machine_translation/transformer/zh-en/module.py
modules/text/machine_translation/transformer/zh-en/module.py
+47
-30
modules/text/machine_translation/transformer/zh-en/utils.py
modules/text/machine_translation/transformer/zh-en/utils.py
+8
-9
未找到文件。
modules/text/machine_translation/transformer/zh-en/README.md
浏览文件 @
e3ee127f
# transformer_zh-en
# transformer_zh-en
|模型名称|transformer_zh-en|
|模型名称|transformer_zh-en|
| :--- | :---: |
| :--- | :---: |
|类别|文本-机器翻译|
|类别|文本-机器翻译|
|网络|Transformer|
|网络|Transformer|
|数据集|CWMT2021|
|数据集|CWMT2021|
...
@@ -24,7 +24,7 @@
...
@@ -24,7 +24,7 @@
-
### 1、环境依赖
-
### 1、环境依赖
-
paddlepaddle >= 2.1.0
-
paddlepaddle >= 2.1.0
-
paddlehub >= 2.1.0 |
[
如何安装PaddleHub
](
../../../../docs/docs_ch/get_start/installation.rst
)
-
paddlehub >= 2.1.0 |
[
如何安装PaddleHub
](
../../../../docs/docs_ch/get_start/installation.rst
)
-
### 2、安装
-
### 2、安装
...
@@ -55,7 +55,7 @@
...
@@ -55,7 +55,7 @@
print('-'*30)
print('-'*30)
print(f'src: {st}')
print(f'src: {st}')
for i in range(n_best):
for i in range(n_best):
print(f'trg[{i+1}]: {trg_texts[idx*n_best+i]}')
print(f'trg[{i+1}]: {trg_texts[idx*n_best+i]}')
```
```
-
### 2、API
-
### 2、API
...
@@ -132,6 +132,9 @@
...
@@ -132,6 +132,9 @@
-
关于PaddleHub Serving更多信息参考:
[
服务部署
](
../../../../docs/docs_ch/tutorial/serving.md
)
-
关于PaddleHub Serving更多信息参考:
[
服务部署
](
../../../../docs/docs_ch/tutorial/serving.md
)
-
### Gradio APP 支持
从 PaddleHub 2.3.1 开始支持使用链接 http://127.0.0.1:8866/gradio/transformer_zh-en 在浏览器中访问 transformer_zh-en 的 Gradio APP。
## 五、更新历史
## 五、更新历史
*
1.0.0
*
1.0.0
...
@@ -141,6 +144,11 @@
...
@@ -141,6 +144,11 @@
*
1.0.1
*
1.0.1
修复模型初始化的兼容性问题
修复模型初始化的兼容性问题
*
1.1.0
添加 Gradio APP 支持
-
```shell
-
```shell
$ hub install transformer_zh-en==1.
0.1
$ hub install transformer_zh-en==1.
1.0
```
```
modules/text/machine_translation/transformer/zh-en/module.py
浏览文件 @
e3ee127f
...
@@ -11,24 +11,26 @@
...
@@ -11,24 +11,26 @@
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# See the License for the specific language governing permissions and
# limitations under the License.
# limitations under the License.
import
os
import
os
from
typing
import
List
from
typing
import
List
import
paddle
import
paddle
import
paddle.nn
as
nn
import
paddle.nn
as
nn
from
paddlehub.env
import
MODULE_HOME
from
paddlenlp.data
import
Pad
from
paddlehub.module.module
import
moduleinfo
,
serving
from
paddlenlp.data
import
Vocab
import
paddlenlp
from
paddlenlp.transformers
import
InferTransformerModel
from
paddlenlp.data
import
Pad
,
Vocab
from
paddlenlp.transformers
import
position_encoding_init
from
paddlenlp.transformers
import
InferTransformerModel
,
position_encoding_init
from
transformer_zh_en.utils
import
MTTokenizer
from
transformer_zh_en.utils
import
post_process_seq
from
transformer_zh_en.utils
import
MTTokenizer
,
post_process_seq
from
paddlehub.env
import
MODULE_HOME
from
paddlehub.module.module
import
moduleinfo
from
paddlehub.module.module
import
serving
@
moduleinfo
(
@
moduleinfo
(
name
=
"transformer_zh-en"
,
name
=
"transformer_zh-en"
,
version
=
"1.
0.1
"
,
version
=
"1.
1.0
"
,
summary
=
""
,
summary
=
""
,
author
=
"PaddlePaddle"
,
author
=
"PaddlePaddle"
,
author_email
=
""
,
author_email
=
""
,
...
@@ -57,7 +59,7 @@ class MTTransformer(nn.Layer):
...
@@ -57,7 +59,7 @@ class MTTransformer(nn.Layer):
# Dropout rate
# Dropout rate
'dropout'
:
0
,
'dropout'
:
0
,
# Number of sub-layers to be stacked in the encoder and decoder.
# Number of sub-layers to be stacked in the encoder and decoder.
"num_encoder_layers"
:
6
,
"num_encoder_layers"
:
6
,
"num_decoder_layers"
:
6
"num_decoder_layers"
:
6
}
}
...
@@ -85,31 +87,29 @@ class MTTransformer(nn.Layer):
...
@@ -85,31 +87,29 @@ class MTTransformer(nn.Layer):
self
.
max_length
=
max_length
self
.
max_length
=
max_length
self
.
beam_size
=
beam_size
self
.
beam_size
=
beam_size
self
.
tokenizer
=
MTTokenizer
(
self
.
tokenizer
=
MTTokenizer
(
bpe_codes_file
=
bpe_codes_file
,
bpe_codes_file
=
bpe_codes_file
,
lang_src
=
self
.
lang_config
[
'source'
],
lang_trg
=
self
.
lang_config
[
'target'
])
lang_src
=
self
.
lang_config
[
'source'
],
self
.
src_vocab
=
Vocab
.
load_vocabulary
(
lang_trg
=
self
.
lang_config
[
'target'
])
filepath
=
src_vocab_file
,
self
.
src_vocab
=
Vocab
.
load_vocabulary
(
filepath
=
src_vocab_file
,
unk_token
=
self
.
vocab_config
[
'unk_token'
],
unk_token
=
self
.
vocab_config
[
'unk_token'
],
bos_token
=
self
.
vocab_config
[
'bos_token'
],
bos_token
=
self
.
vocab_config
[
'bos_token'
],
eos_token
=
self
.
vocab_config
[
'eos_token'
])
eos_token
=
self
.
vocab_config
[
'eos_token'
])
self
.
trg_vocab
=
Vocab
.
load_vocabulary
(
self
.
trg_vocab
=
Vocab
.
load_vocabulary
(
filepath
=
trg_vocab_file
,
filepath
=
trg_vocab_file
,
unk_token
=
self
.
vocab_config
[
'unk_token'
],
unk_token
=
self
.
vocab_config
[
'unk_token'
],
bos_token
=
self
.
vocab_config
[
'bos_token'
],
bos_token
=
self
.
vocab_config
[
'bos_token'
],
eos_token
=
self
.
vocab_config
[
'eos_token'
])
eos_token
=
self
.
vocab_config
[
'eos_token'
])
self
.
src_vocab_size
=
(
len
(
self
.
src_vocab
)
+
self
.
vocab_config
[
'pad_factor'
]
-
1
)
\
self
.
src_vocab_size
=
(
len
(
self
.
src_vocab
)
+
self
.
vocab_config
[
'pad_factor'
]
-
1
)
\
//
self
.
vocab_config
[
'pad_factor'
]
*
self
.
vocab_config
[
'pad_factor'
]
//
self
.
vocab_config
[
'pad_factor'
]
*
self
.
vocab_config
[
'pad_factor'
]
self
.
trg_vocab_size
=
(
len
(
self
.
trg_vocab
)
+
self
.
vocab_config
[
'pad_factor'
]
-
1
)
\
self
.
trg_vocab_size
=
(
len
(
self
.
trg_vocab
)
+
self
.
vocab_config
[
'pad_factor'
]
-
1
)
\
//
self
.
vocab_config
[
'pad_factor'
]
*
self
.
vocab_config
[
'pad_factor'
]
//
self
.
vocab_config
[
'pad_factor'
]
*
self
.
vocab_config
[
'pad_factor'
]
self
.
transformer
=
InferTransformerModel
(
self
.
transformer
=
InferTransformerModel
(
src_vocab_size
=
self
.
src_vocab_size
,
src_vocab_size
=
self
.
src_vocab_size
,
trg_vocab_size
=
self
.
trg_vocab_size
,
trg_vocab_size
=
self
.
trg_vocab_size
,
bos_id
=
self
.
vocab_config
[
'bos_id'
],
bos_id
=
self
.
vocab_config
[
'bos_id'
],
eos_id
=
self
.
vocab_config
[
'eos_id'
],
eos_id
=
self
.
vocab_config
[
'eos_id'
],
max_length
=
self
.
max_length
+
1
,
max_length
=
self
.
max_length
+
1
,
max_out_len
=
max_out_len
,
max_out_len
=
max_out_len
,
beam_size
=
self
.
beam_size
,
beam_size
=
self
.
beam_size
,
**
self
.
model_config
)
**
self
.
model_config
)
state_dict
=
paddle
.
load
(
checkpoint
)
state_dict
=
paddle
.
load
(
checkpoint
)
...
@@ -184,3 +184,20 @@ class MTTransformer(nn.Layer):
...
@@ -184,3 +184,20 @@ class MTTransformer(nn.Layer):
results
.
append
(
trg_sample_text
)
results
.
append
(
trg_sample_text
)
return
results
return
results
def
create_gradio_app
(
self
):
import
gradio
as
gr
def
inference
(
text
):
results
=
self
.
predict
(
data
=
[
text
])
return
results
[
0
]
examples
=
[[
'今天是个好日子'
]]
interface
=
gr
.
Interface
(
inference
,
"text"
,
[
gr
.
outputs
.
Textbox
(
label
=
"Translation"
)],
title
=
"transformer_zh-en"
,
examples
=
examples
,
allow_flagging
=
'never'
)
return
interface
modules/text/machine_translation/transformer/zh-en/utils.py
浏览文件 @
e3ee127f
...
@@ -11,29 +11,28 @@
...
@@ -11,29 +11,28 @@
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# See the License for the specific language governing permissions and
# limitations under the License.
# limitations under the License.
import
codecs
import
logging
import
logging
import
re
import
re
from
typing
import
List
from
typing
import
List
import
codecs
import
jieba
import
jieba
jieba
.
setLogLevel
(
logging
.
INFO
)
jieba
.
setLogLevel
(
logging
.
INFO
)
from
sacremoses
import
Moses
Tokenizer
,
Moses
Detokenizer
from
sacremoses
import
MosesDetokenizer
from
subword_nmt.apply_bpe
import
BPE
from
subword_nmt.apply_bpe
import
BPE
class
MTTokenizer
(
object
):
class
MTTokenizer
(
object
):
def
__init__
(
self
,
bpe_codes_file
:
str
,
lang_src
:
str
=
'zh'
,
lang_trg
:
str
=
'en'
,
separator
=
'@@'
):
def
__init__
(
self
,
bpe_codes_file
:
str
,
lang_src
:
str
=
'zh'
,
lang_trg
:
str
=
'en'
,
separator
=
'@@'
):
self
.
moses_detokenizer
=
MosesDetokenizer
(
lang
=
lang_trg
)
self
.
moses_detokenizer
=
MosesDetokenizer
(
lang
=
lang_trg
)
self
.
bpe_tokenizer
=
BPE
(
self
.
bpe_tokenizer
=
BPE
(
codes
=
codecs
.
open
(
bpe_codes_file
,
encoding
=
'utf-8'
),
codes
=
codecs
.
open
(
bpe_codes_file
,
encoding
=
'utf-8'
),
merges
=-
1
,
merges
=-
1
,
separator
=
separator
,
separator
=
separator
,
vocab
=
None
,
vocab
=
None
,
glossaries
=
None
)
glossaries
=
None
)
def
tokenize
(
self
,
text
:
str
):
def
tokenize
(
self
,
text
:
str
):
"""
"""
...
...
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