未验证 提交 e3ee127f 编写于 作者: jm_12138's avatar jm_12138 提交者: GitHub

add gradio app (#2161)

上级 32dbb01a
# 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
``` ```
...@@ -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
...@@ -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 MosesTokenizer, MosesDetokenizer 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):
""" """
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册