未验证 提交 b68da8e6 编写于 作者: K KP 提交者: GitHub

Add electra modules (#1173)

* Add electra modules

* Amend module versions of chinese-bert-wwm, chinese-bert-wwm-ext, rbt3 and rbtl3

* Updata demo README.md
上级 790e05dd
......@@ -60,10 +60,10 @@ ERNIE, Chinese | `hub.Module(name='ernie')`
ERNIE tiny, Chinese | `hub.Module(name='ernie_tiny')`
ERNIE 2.0 Base, English | `hub.Module(name='ernie_v2_eng_base')`
ERNIE 2.0 Large, English | `hub.Module(name='ernie_v2_eng_large')`
BERT-Base, Cased | `hub.Module(name='bert-base-cased')`
BERT-Base, Uncased | `hub.Module(name='bert-base-uncased')`
BERT-Large, Cased | `hub.Module(name='bert-large-cased')`
BERT-Large, Uncased | `hub.Module(name='bert-large-uncased')`
BERT-Base, English Cased | `hub.Module(name='bert-base-cased')`
BERT-Base, English Uncased | `hub.Module(name='bert-base-uncased')`
BERT-Large, English Cased | `hub.Module(name='bert-large-cased')`
BERT-Large, English Uncased | `hub.Module(name='bert-large-uncased')`
BERT-Base, Multilingual Cased | `hub.Module(nane='bert-base-multilingual-cased')`
BERT-Base, Multilingual Uncased | `hub.Module(nane='bert-base-multilingual-uncased')`
BERT-Base, Chinese | `hub.Module(name='bert-base-chinese')`
......@@ -73,6 +73,11 @@ RoBERTa-wwm-ext, Chinese | `hub.Module(name='roberta-wwm-ext')`
RoBERTa-wwm-ext-large, Chinese | `hub.Module(name='roberta-wwm-ext-large')`
RBT3, Chinese | `hub.Module(name='rbt3')`
RBTL3, Chinese | `hub.Module(name='rbtl3')`
ELECTRA-Small, English | `hub.Module(name='electra-small')`
ELECTRA-Base, English | `hub.Module(name='electra-base')`
ELECTRA-Large, English | `hub.Module(name='electra-large')`
ELECTRA-Base, Chinese | `hub.Module(name='chinese-electra-base')`
ELECTRA-Small, Chinese | `hub.Module(name='chinese-electra-small')`
通过以上的一行代码,`model`初始化为一个适用于序列标注任务的模型,为ERNIE Tiny的预训练模型后拼接上一个输出token共享的全连接网络(Full Connected)。
![](https://ss1.bdstatic.com/70cFuXSh_Q1YnxGkpoWK1HF6hhy/it/u=224484727,3049769188&fm=15&gp=0.jpg)
......
......@@ -49,10 +49,10 @@ ERNIE, Chinese | `hub.Module(name='ernie')`
ERNIE tiny, Chinese | `hub.Module(name='ernie_tiny')`
ERNIE 2.0 Base, English | `hub.Module(name='ernie_v2_eng_base')`
ERNIE 2.0 Large, English | `hub.Module(name='ernie_v2_eng_large')`
BERT-Base, Cased | `hub.Module(name='bert-base-cased')`
BERT-Base, Uncased | `hub.Module(name='bert-base-uncased')`
BERT-Large, Cased | `hub.Module(name='bert-large-cased')`
BERT-Large, Uncased | `hub.Module(name='bert-large-uncased')`
BERT-Base, English Cased | `hub.Module(name='bert-base-cased')`
BERT-Base, English Uncased | `hub.Module(name='bert-base-uncased')`
BERT-Large, English Cased | `hub.Module(name='bert-large-cased')`
BERT-Large, English Uncased | `hub.Module(name='bert-large-uncased')`
BERT-Base, Multilingual Cased | `hub.Module(nane='bert-base-multilingual-cased')`
BERT-Base, Multilingual Uncased | `hub.Module(nane='bert-base-multilingual-uncased')`
BERT-Base, Chinese | `hub.Module(name='bert-base-chinese')`
......@@ -62,6 +62,11 @@ RoBERTa-wwm-ext, Chinese | `hub.Module(name='roberta-wwm-ext')`
RoBERTa-wwm-ext-large, Chinese | `hub.Module(name='roberta-wwm-ext-large')`
RBT3, Chinese | `hub.Module(name='rbt3')`
RBTL3, Chinese | `hub.Module(name='rbtl3')`
ELECTRA-Small, English | `hub.Module(name='electra-small')`
ELECTRA-Base, English | `hub.Module(name='electra-base')`
ELECTRA-Large, English | `hub.Module(name='electra-large')`
ELECTRA-Base, Chinese | `hub.Module(name='chinese-electra-base')`
ELECTRA-Small, Chinese | `hub.Module(name='chinese-electra-small')`
通过以上的一行代码,`model`初始化为一个适用于文本分类任务的模型,为ERNIE Tiny的预训练模型后拼接上一个全连接网络(Full Connected)。
![](https://ai-studio-static-online.cdn.bcebos.com/f9e1bf9d56c6412d939960f2e3767c2f13b93eab30554d738b137ab2b98e328c)
......
```shell
$ hub install chinese-bert-wwm==2.0.1
$ hub install chinese-bert-wwm==2.0.0
```
<p align="center">
<img src="https://bj.bcebos.com/paddlehub/paddlehub-img/bert_network.png" hspace='10'/> <br />
......@@ -82,7 +82,7 @@ label_map = {0: 'negative', 1: 'positive'}
model = hub.Module(
name='chinese-bert-wwm',
version='2.0.1',
version='2.0.0',
task='seq-cls',
load_checkpoint='/path/to/parameters',
label_map=label_map)
......@@ -153,6 +153,6 @@ paddlehub >= 2.0.0
初始发布
* 2.0.1
* 2.0.0
全面升级动态图,接口有所变化。任务名称调整,增加序列标注任务`token-cls`
......@@ -29,7 +29,7 @@ from paddlehub.utils.log import logger
@moduleinfo(
name="chinese-bert-wwm",
version="2.0.1",
version="2.0.0",
summary=
"chinese-bert-wwm, 12-layer, 768-hidden, 12-heads, 110M parameters. The module is executed as paddle.dygraph.",
author="ymcui",
......
```shell
$ hub install chinese-bert-wwm-ext==2.0.1
$ hub install chinese-bert-wwm-ext==2.0.0
```
<p align="center">
<img src="https://bj.bcebos.com/paddlehub/paddlehub-img/bert_network.png" hspace='10'/> <br />
......@@ -82,7 +82,7 @@ label_map = {0: 'negative', 1: 'positive'}
model = hub.Module(
name='chinese-bert-wwm-ext',
version='2.0.1',
version='2.0.0',
task='seq-cls',
load_checkpoint='/path/to/parameters',
label_map=label_map)
......@@ -153,6 +153,6 @@ paddlehub >= 2.0.0
初始发布
* 2.0.1
* 2.0.0
全面升级动态图,接口有所变化。任务名称调整,增加序列标注任务`token-cls`
......@@ -29,7 +29,7 @@ from paddlehub.utils.log import logger
@moduleinfo(
name="chinese-bert-wwm-ext",
version="2.0.1",
version="2.0.0",
summary=
"chinese-bert-wwm-ext, 12-layer, 768-hidden, 12-heads, 110M parameters. The module is executed as paddle.dygraph.",
author="ymcui",
......
```shell
$ hub install chinese-electra-base==1.0.0
$ hub install chinese-electra-base==2.0.0
```
<p align="center">
<img src="https://github.com/ymcui/Chinese-ELECTRA/blob/master/pics/model.png" hspace='10'/> <br />
<img src="http://bj.bcebos.com/ibox-thumbnail98/1a5578bfbe1ad629035f7ad1eb3d0bce?authorization=bce-auth-v1%2Ffbe74140929444858491fbf2b6bc0935%2F2020-03-31T06%3A45%3A51Z%2F1800%2F%2F02b8749292f8ba1c606410d0e4e5dbabdf1d367d80da395887775d36424ac13e" hspace='10'/> <br />
</p>
更多详情请参考[ELECTRA论文](https://openreview.net/pdf?id=r1xMH1BtvB)
## API
```python
def context(
trainable=True,
max_seq_len=128
def __init__(
task=None,
load_checkpoint=None,
label_map=None,
num_classes=2,
**kwargs,
)
```
用于获取Module的上下文信息,得到输入、输出以及预训练的Paddle Program副本
**参数**
> trainable:设置为True时,Module中的参数在Fine-tune时也会随之训练,否则保持不变。
> max_seq_len:BERT模型的最大序列长度,若序列长度不足,会通过padding方式补到**max_seq_len**, 若序列长度大于该值,则会以截断方式让序列长度为**max_seq_len**,max_seq_len可取值范围为0~512;
创建Module对象(动态图组网版本)。
**返回**
> inputs:dict类型,有以下字段:
> >**input_ids**存放输入文本tokenize后各token对应BERT词汇表的word ids, shape为\[batch_size, max_seq_len\],int64类型;
> >**position_ids**存放输入文本tokenize后各token所在该文本的位置,shape为\[batch_size, max_seq_len\],int64类型;
> >**segment_ids**存放各token所在文本的标识(token属于文本1或者文本2),shape为\[batch_size, max_seq_len\],int64类型;
> >**input_mask**存放token是否为padding的标识,shape为\[batch_size, max_seq_len\],int64类型;
>
> outputs:dict类型,Module的输出特征,有以下字段:
> >**pooled_output**字段存放句子粒度的特征,可用于文本分类等任务,shape为 \[batch_size, 768\],int64类型;
> >**sequence_output**字段存放字粒度的特征,可用于序列标注等任务,shape为 \[batch_size, seq_len, 768\],int64类型;
>
> program:包含该Module计算图的Program。
**参数**
* `task`: 任务名称,可为`seq-cls`(文本分类任务,原来的`sequence_classification`在未来会被弃用)或`token-cls`(序列标注任务)。
* `load_checkpoint`:使用PaddleHub Fine-tune api训练保存的模型参数文件路径。
* `label_map`:预测时的类别映射表。
* `num_classes`:分类任务的类别数,如果指定了`label_map`,此参数可不传,默认2分类。
* `**kwargs`:用户额外指定的关键字字典类型的参数。
```python
def get_embedding(
texts,
use_gpu=False,
batch_size=1
def predict(
data,
max_seq_len=128,
batch_size=1,
use_gpu=False
)
```
用于获取输入文本的句子粒度特征与字粒度特征
**参数**
> texts:输入文本列表,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。
> use_gpu:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
* `data`: 待预测数据,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。每个样例文本数量(1个或者2个)需和训练时保持一致。
* `max_seq_len`:模型处理文本的最大长度
* `batch_size`:模型批处理大小
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
> results:list类型,格式为\[\[sample\_a\_pooled\_feature, sample\_a\_seq\_feature\], \[sample\_b\_pooled\_feature, sample\_b\_seq\_feature\],…,\],其中每个元素都是对应样例的特征输出,每个样例都有句子粒度特征pooled\_feature与字粒度特征seq\_feature。
>
* `results`:list类型,不同任务类型的返回结果如下
* 文本分类:列表里包含每个句子的预测标签,格式为\[label\_1, label\_2, …,\]
* 序列标注:列表里包含每个句子每个token的预测标签,格式为\[\[token\_1, token\_2, …,\], \[token\_1, token\_2, …,\], …,\]
```python
def get_params_layer()
def get_embedding(
data,
use_gpu=False
)
```
用于获取参数层信息,该方法与ULMFiTStrategy联用可以严格按照层数设置分层学习率与逐层解冻。
用于获取输入文本的句子粒度特征与字粒度特征
**参数**
> 无
* `data`:输入文本列表,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
> params_layer:dict类型,key为参数名,值为参数所在层数
* `results`:list类型,格式为\[\[sample\_a\_pooled\_feature, sample\_a\_seq\_feature\], \[sample\_b\_pooled\_feature, sample\_b\_seq\_feature\],…,\],其中每个元素都是对应样例的特征输出,每个样例都有句子粒度特征pooled\_feature与字粒度特征seq\_feature。
**代码示例**
......@@ -75,45 +75,83 @@ def get_params_layer()
```python
import paddlehub as hub
# Load $ hub install chinese-electra-base pretrained model
module = hub.Module(name="chinese-electra-base")
inputs, outputs, program = module.context(trainable=True, max_seq_len=128)
data = [
['这个宾馆比较陈旧了,特价的房间也很一般。总体来说一般'],
['怀着十分激动的心情放映,可是看着看着发现,在放映完毕后,出现一集米老鼠的动画片'],
['作为老的四星酒店,房间依然很整洁,相当不错。机场接机服务很好,可以在车上办理入住手续,节省时间。'],
]
label_map = {0: 'negative', 1: 'positive'}
model = hub.Module(
name='chinese-electra-base',
version='2.0.0',
task='seq-cls',
load_checkpoint='/path/to/parameters',
label_map=label_map)
results = model.predict(data, max_seq_len=50, batch_size=1, use_gpu=False)
for idx, text in enumerate(data):
print('Data: {} \t Lable: {}'.format(text, results[idx]))
```
详情可参考PaddleHub示例:
- [文本分类](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/text_classification)
- [序列标注](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/sequence_labeling)
# Must feed all the tensor of chinese-electra-base's module need
input_ids = inputs["input_ids"]
position_ids = inputs["position_ids"]
segment_ids = inputs["segment_ids"]
input_mask = inputs["input_mask"]
## 服务部署
# Use "pooled_output" for sentence-level output.
pooled_output = outputs["pooled_output"]
PaddleHub Serving可以部署一个在线获取预训练词向量。
# Use "sequence_output" for token-level output.
sequence_output = outputs["sequence_output"]
### Step1: 启动PaddleHub Serving
# Use "get_embedding" to get embedding result.
embedding_result = module.get_embedding(texts=[["Sample1_text_a"],["Sample2_text_a","Sample2_text_b"]], use_gpu=True)
运行启动命令:
# Use "get_params_layer" to get params layer and used to ULMFiTStrategy.
params_layer = module.get_params_layer()
strategy = hub.finetune.strategy.ULMFiTStrategy(frz_params_layer=params_layer, dis_params_layer=params_layer)
```shell
$ hub serving start -m chinese-electra-base
```
这样就完成了一个获取预训练词向量服务化API的部署,默认端口号为8866。
**NOTE:** 如使用GPU预测,则需要在启动服务之前,请设置CUDA_VISIBLE_DEVICES环境变量,否则不用设置。
### Step2: 发送预测请求
配置好服务端,以下数行代码即可实现发送预测请求,获取预测结果
```python
import requests
import json
# 指定用于获取embedding的文本[[text_1], [text_2], ... ]}
text = [["今天是个好日子"], ["天气预报说今天要下雨"]]
# 以key的方式指定text传入预测方法的时的参数,此例中为"data"
# 对应本地部署,则为module.get_embedding(data=text)
data = {"data": text}
# 发送post请求,content-type类型应指定json方式,url中的ip地址需改为对应机器的ip
url = "http://10.12.121.132:8866/predict/chinese-electra-base"
# 指定post请求的headers为application/json方式
headers = {"Content-Type": "application/json"}
r = requests.post(url=url, headers=headers, data=json.dumps(data))
print(r.json())
```
## 查看代码
https://github.com/ymcui/Chinese-ELECTRA
## 依赖
paddlepaddle >= 1.6.2
paddlepaddle >= 2.0.0
paddlehub >= 1.6.0
paddlehub >= 2.0.0
## 更新历史
* 1.0.0
初始发布
* 2.0.0
全面升级动态图,接口有所变化。任务名称调整,增加序列标注任务`token-cls`
# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""ELECTRA model."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import six
import json
import paddle.fluid as fluid
from chinese_electra_base.model.transformer_encoder import encoder, pre_process_layer
class ElectraConfig(object):
def __init__(self, config_path):
self._config_dict = self._parse(config_path)
def _parse(self, config_path):
try:
with open(config_path) as json_file:
config_dict = json.load(json_file)
except Exception:
raise IOError("Error in parsing electra model config file '%s'" % config_path)
else:
return config_dict
def __getitem__(self, key):
return self._config_dict[key]
def print_config(self):
for arg, value in sorted(six.iteritems(self._config_dict)):
print('%s: %s' % (arg, value))
print('------------------------------------------------')
class ElectraModel(object):
def __init__(self, src_ids, position_ids, sentence_ids, input_mask, config, weight_sharing=True, use_fp16=False):
self._emb_size = config['hidden_size']
self._n_layer = config['num_hidden_layers']
self._n_head = config['num_attention_heads']
self._voc_size = config['vocab_size']
self._max_position_seq_len = config['max_position_embeddings']
self._sent_types = config['type_vocab_size']
self._hidden_act = config['hidden_act']
self._prepostprocess_dropout = config['hidden_dropout_prob']
self._attention_dropout = config['attention_probs_dropout_prob']
self._weight_sharing = weight_sharing
self._word_emb_name = "word_embedding"
self._pos_emb_name = "pos_embedding"
self._sent_emb_name = "sent_embedding"
self._dtype = "float16" if use_fp16 else "float32"
# Initialize all weigths by truncated normal initializer, and all biases
# will be initialized by constant zero by default.
self._param_initializer = fluid.initializer.TruncatedNormal(scale=config['initializer_range'])
self._build_model(src_ids, position_ids, sentence_ids, input_mask)
def _build_model(self, src_ids, position_ids, sentence_ids, input_mask):
# padding id in vocabulary must be set to 0
emb_out = fluid.layers.embedding(input=src_ids,
size=[self._voc_size, self._emb_size],
dtype=self._dtype,
param_attr=fluid.ParamAttr(name=self._word_emb_name,
initializer=self._param_initializer),
is_sparse=False)
position_emb_out = fluid.layers.embedding(input=position_ids,
size=[self._max_position_seq_len, self._emb_size],
dtype=self._dtype,
param_attr=fluid.ParamAttr(name=self._pos_emb_name,
initializer=self._param_initializer))
sent_emb_out = fluid.layers.embedding(sentence_ids,
size=[self._sent_types, self._emb_size],
dtype=self._dtype,
param_attr=fluid.ParamAttr(name=self._sent_emb_name,
initializer=self._param_initializer))
emb_out = emb_out + position_emb_out
emb_out = emb_out + sent_emb_out
emb_out = pre_process_layer(emb_out, 'nd', self._prepostprocess_dropout, name='pre_encoder')
if self._dtype == "float16":
input_mask = fluid.layers.cast(x=input_mask, dtype=self._dtype)
self_attn_mask = fluid.layers.matmul(x=input_mask, y=input_mask, transpose_y=True)
self_attn_mask = fluid.layers.scale(x=self_attn_mask, scale=10000.0, bias=-1.0, bias_after_scale=False)
n_head_self_attn_mask = fluid.layers.stack(x=[self_attn_mask] * self._n_head, axis=1)
n_head_self_attn_mask.stop_gradient = True
self._enc_out = encoder(enc_input=emb_out,
attn_bias=n_head_self_attn_mask,
n_layer=self._n_layer,
n_head=self._n_head,
d_key=self._emb_size // self._n_head,
d_value=self._emb_size // self._n_head,
d_model=self._emb_size,
d_inner_hid=self._emb_size * 4,
prepostprocess_dropout=self._prepostprocess_dropout,
attention_dropout=self._attention_dropout,
relu_dropout=0,
hidden_act=self._hidden_act,
preprocess_cmd="",
postprocess_cmd="dan",
param_initializer=self._param_initializer,
name='encoder')
def get_sequence_output(self):
return self._enc_out
def get_pooled_output(self):
"""Get the first feature of each sequence for classification"""
next_sent_feat = fluid.layers.slice(input=self._enc_out, axes=[1], starts=[0], ends=[1])
return next_sent_feat
def get_pretraining_output(self, mask_label, mask_pos, labels):
"""Get the loss & accuracy for pretraining"""
mask_pos = fluid.layers.cast(x=mask_pos, dtype='int32')
# extract the first token feature in each sentence
next_sent_feat = self.get_pooled_output()
reshaped_emb_out = fluid.layers.reshape(x=self._enc_out, shape=[-1, self._emb_size])
# extract masked tokens' feature
mask_feat = fluid.layers.gather(input=reshaped_emb_out, index=mask_pos)
# transform: fc
mask_trans_feat = fluid.layers.fc(input=mask_feat,
size=self._emb_size,
act=self._hidden_act,
param_attr=fluid.ParamAttr(name='mask_lm_trans_fc.w_0',
initializer=self._param_initializer),
bias_attr=fluid.ParamAttr(name='mask_lm_trans_fc.b_0'))
# transform: layer norm
mask_trans_feat = pre_process_layer(mask_trans_feat, 'n', name='mask_lm_trans')
mask_lm_out_bias_attr = fluid.ParamAttr(name="mask_lm_out_fc.b_0",
initializer=fluid.initializer.Constant(value=0.0))
if self._weight_sharing:
fc_out = fluid.layers.matmul(x=mask_trans_feat,
y=fluid.default_main_program().global_block().var(self._word_emb_name),
transpose_y=True)
fc_out += fluid.layers.create_parameter(shape=[self._voc_size],
dtype=self._dtype,
attr=mask_lm_out_bias_attr,
is_bias=True)
else:
fc_out = fluid.layers.fc(input=mask_trans_feat,
size=self._voc_size,
param_attr=fluid.ParamAttr(name="mask_lm_out_fc.w_0",
initializer=self._param_initializer),
bias_attr=mask_lm_out_bias_attr)
mask_lm_loss = fluid.layers.softmax_with_cross_entropy(logits=fc_out, label=mask_label)
mean_mask_lm_loss = fluid.layers.mean(mask_lm_loss)
next_sent_fc_out = fluid.layers.fc(input=next_sent_feat,
size=2,
param_attr=fluid.ParamAttr(name="next_sent_fc.w_0",
initializer=self._param_initializer),
bias_attr="next_sent_fc.b_0")
next_sent_loss, next_sent_softmax = fluid.layers.softmax_with_cross_entropy(logits=next_sent_fc_out,
label=labels,
return_softmax=True)
next_sent_acc = fluid.layers.accuracy(input=next_sent_softmax, label=labels)
mean_next_sent_loss = fluid.layers.mean(next_sent_loss)
loss = mean_next_sent_loss + mean_mask_lm_loss
return next_sent_acc, mean_mask_lm_loss, loss
# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Transformer encoder."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from functools import partial
import paddle.fluid as fluid
import paddle.fluid.layers as layers
def multi_head_attention(queries,
keys,
values,
attn_bias,
d_key,
d_value,
d_model,
n_head=1,
dropout_rate=0.,
cache=None,
param_initializer=None,
name='multi_head_att'):
"""
Multi-Head Attention. Note that attn_bias is added to the logit before
computing softmax activiation to mask certain selected positions so that
they will not considered in attention weights.
"""
keys = queries if keys is None else keys
values = keys if values is None else values
if not (len(queries.shape) == len(keys.shape) == len(values.shape) == 3):
raise ValueError("Inputs: quries, keys and values should all be 3-D tensors.")
def __compute_qkv(queries, keys, values, n_head, d_key, d_value):
"""
Add linear projection to queries, keys, and values.
"""
q = layers.fc(input=queries,
size=d_key * n_head,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_query_fc.w_0', initializer=param_initializer),
bias_attr=name + '_query_fc.b_0')
k = layers.fc(input=keys,
size=d_key * n_head,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_key_fc.w_0', initializer=param_initializer),
bias_attr=name + '_key_fc.b_0')
v = layers.fc(input=values,
size=d_value * n_head,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_value_fc.w_0', initializer=param_initializer),
bias_attr=name + '_value_fc.b_0')
return q, k, v
def __split_heads(x, n_head):
"""
Reshape the last dimension of inpunt tensor x so that it becomes two
dimensions and then transpose. Specifically, input a tensor with shape
[bs, max_sequence_length, n_head * hidden_dim] then output a tensor
with shape [bs, n_head, max_sequence_length, hidden_dim].
"""
hidden_size = x.shape[-1]
# The value 0 in shape attr means copying the corresponding dimension
# size of the input as the output dimension size.
reshaped = layers.reshape(x=x, shape=[0, 0, n_head, hidden_size // n_head], inplace=True)
# permuate the dimensions into:
# [batch_size, n_head, max_sequence_len, hidden_size_per_head]
return layers.transpose(x=reshaped, perm=[0, 2, 1, 3])
def __combine_heads(x):
"""
Transpose and then reshape the last two dimensions of inpunt tensor x
so that it becomes one dimension, which is reverse to __split_heads.
"""
if len(x.shape) == 3: return x
if len(x.shape) != 4:
raise ValueError("Input(x) should be a 4-D Tensor.")
trans_x = layers.transpose(x, perm=[0, 2, 1, 3])
# The value 0 in shape attr means copying the corresponding dimension
# size of the input as the output dimension size.
return layers.reshape(x=trans_x, shape=[0, 0, trans_x.shape[2] * trans_x.shape[3]], inplace=True)
def scaled_dot_product_attention(q, k, v, attn_bias, d_key, dropout_rate):
"""
Scaled Dot-Product Attention
"""
scaled_q = layers.scale(x=q, scale=d_key**-0.5)
product = layers.matmul(x=scaled_q, y=k, transpose_y=True)
if attn_bias:
product += attn_bias
weights = layers.softmax(product)
if dropout_rate:
weights = layers.dropout(weights,
dropout_prob=dropout_rate,
dropout_implementation="upscale_in_train",
is_test=False)
out = layers.matmul(weights, v)
return out
q, k, v = __compute_qkv(queries, keys, values, n_head, d_key, d_value)
if cache is not None: # use cache and concat time steps
# Since the inplace reshape in __split_heads changes the shape of k and
# v, which is the cache input for next time step, reshape the cache
# input from the previous time step first.
k = cache["k"] = layers.concat([layers.reshape(cache["k"], shape=[0, 0, d_model]), k], axis=1)
v = cache["v"] = layers.concat([layers.reshape(cache["v"], shape=[0, 0, d_model]), v], axis=1)
q = __split_heads(q, n_head)
k = __split_heads(k, n_head)
v = __split_heads(v, n_head)
ctx_multiheads = scaled_dot_product_attention(q, k, v, attn_bias, d_key, dropout_rate)
out = __combine_heads(ctx_multiheads)
# Project back to the model size.
proj_out = layers.fc(input=out,
size=d_model,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_output_fc.w_0', initializer=param_initializer),
bias_attr=name + '_output_fc.b_0')
return proj_out
def positionwise_feed_forward(x, d_inner_hid, d_hid, dropout_rate, hidden_act, param_initializer=None, name='ffn'):
"""
Position-wise Feed-Forward Networks.
This module consists of two linear transformations with a ReLU activation
in between, which is applied to each position separately and identically.
"""
hidden = layers.fc(input=x,
size=d_inner_hid,
num_flatten_dims=2,
act=hidden_act,
param_attr=fluid.ParamAttr(name=name + '_fc_0.w_0', initializer=param_initializer),
bias_attr=name + '_fc_0.b_0')
if dropout_rate:
hidden = layers.dropout(hidden,
dropout_prob=dropout_rate,
dropout_implementation="upscale_in_train",
is_test=False)
out = layers.fc(input=hidden,
size=d_hid,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_fc_1.w_0', initializer=param_initializer),
bias_attr=name + '_fc_1.b_0')
return out
def pre_post_process_layer(prev_out, out, process_cmd, dropout_rate=0., name=''):
"""
Add residual connection, layer normalization and droput to the out tensor
optionally according to the value of process_cmd.
This will be used before or after multi-head attention and position-wise
feed-forward networks.
"""
for cmd in process_cmd:
if cmd == "a": # add residual connection
out = out + prev_out if prev_out else out
elif cmd == "n": # add layer normalization
out_dtype = out.dtype
if out_dtype == fluid.core.VarDesc.VarType.FP16:
out = layers.cast(x=out, dtype="float32")
out = layers.layer_norm(out,
begin_norm_axis=len(out.shape) - 1,
param_attr=fluid.ParamAttr(name=name + '_layer_norm_scale',
initializer=fluid.initializer.Constant(1.)),
bias_attr=fluid.ParamAttr(name=name + '_layer_norm_bias',
initializer=fluid.initializer.Constant(0.)))
if out_dtype == fluid.core.VarDesc.VarType.FP16:
out = layers.cast(x=out, dtype="float16")
elif cmd == "d": # add dropout
if dropout_rate:
out = layers.dropout(out,
dropout_prob=dropout_rate,
dropout_implementation="upscale_in_train",
is_test=False)
return out
pre_process_layer = partial(pre_post_process_layer, None)
post_process_layer = pre_post_process_layer
def encoder_layer(enc_input,
attn_bias,
n_head,
d_key,
d_value,
d_model,
d_inner_hid,
prepostprocess_dropout,
attention_dropout,
relu_dropout,
hidden_act,
preprocess_cmd="n",
postprocess_cmd="da",
param_initializer=None,
name=''):
"""The encoder layers that can be stacked to form a deep encoder.
This module consits of a multi-head (self) attention followed by
position-wise feed-forward networks and both the two components companied
with the post_process_layer to add residual connection, layer normalization
and droput.
"""
attn_output = multi_head_attention(pre_process_layer(enc_input,
preprocess_cmd,
prepostprocess_dropout,
name=name + '_pre_att'),
None,
None,
attn_bias,
d_key,
d_value,
d_model,
n_head,
attention_dropout,
param_initializer=param_initializer,
name=name + '_multi_head_att')
attn_output = post_process_layer(enc_input,
attn_output,
postprocess_cmd,
prepostprocess_dropout,
name=name + '_post_att')
ffd_output = positionwise_feed_forward(pre_process_layer(attn_output,
preprocess_cmd,
prepostprocess_dropout,
name=name + '_pre_ffn'),
d_inner_hid,
d_model,
relu_dropout,
hidden_act,
param_initializer=param_initializer,
name=name + '_ffn')
return post_process_layer(attn_output, ffd_output, postprocess_cmd, prepostprocess_dropout, name=name + '_post_ffn')
def encoder(enc_input,
attn_bias,
n_layer,
n_head,
d_key,
d_value,
d_model,
d_inner_hid,
prepostprocess_dropout,
attention_dropout,
relu_dropout,
hidden_act,
preprocess_cmd="n",
postprocess_cmd="da",
param_initializer=None,
name=''):
"""
The encoder is composed of a stack of identical layers returned by calling
encoder_layer.
"""
for i in range(n_layer):
enc_output = encoder_layer(enc_input,
attn_bias,
n_head,
d_key,
d_value,
d_model,
d_inner_hid,
prepostprocess_dropout,
attention_dropout,
relu_dropout,
hidden_act,
preprocess_cmd,
postprocess_cmd,
param_initializer=param_initializer,
name=name + '_layer_' + str(i))
enc_input = enc_output
enc_output = pre_process_layer(enc_output, preprocess_cmd, prepostprocess_dropout, name="post_encoder")
return enc_output
# coding:utf-8
# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License"
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
......@@ -12,62 +11,120 @@
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from typing import Dict
import os
from paddlehub import TransformerModule
from paddlehub.module.module import moduleinfo
import paddle
import paddle.nn as nn
import paddle.nn.functional as F
from chinese_electra_base.model.electra import ElectraConfig, ElectraModel
from paddlenlp.transformers.electra.modeling import ElectraForSequenceClassification, ElectraForTokenClassification, ElectraModel
from paddlenlp.transformers.electra.tokenizer import ElectraTokenizer
from paddlenlp.metrics import ChunkEvaluator
from paddlehub.module.module import moduleinfo
from paddlehub.module.nlp_module import TransformerModule
from paddlehub.utils.log import logger
@moduleinfo(
name="chinese-electra-base",
version="1.0.0",
summary="chinese-electra-base, 12-layer, 768-hidden, 12-heads, 102M parameters",
version="2.0.0",
summary=
"chinese-electra-base, 12-layer, 768-hidden, 12-heads, 102M parameters. The module is executed as paddle.dygraph.",
author="ymcui",
author_email="ymcui@ir.hit.edu.cn",
type="nlp/semantic_model",
meta=TransformerModule,
)
class Electra(TransformerModule):
def _initialize(self):
self.MAX_SEQ_LEN = 512
self.params_path = os.path.join(self.directory, "assets", "params")
self.vocab_path = os.path.join(self.directory, "assets", "vocab.txt")
class Electra(nn.Layer):
"""
Electra model
"""
electra_config_path = os.path.join(self.directory, "assets", "config.json")
self.electra_config = ElectraConfig(electra_config_path)
def __init__(
self,
task: str = None,
load_checkpoint: str = None,
label_map: Dict = None,
num_classes: int = 2,
**kwargs,
):
super(Electra, self).__init__()
if label_map:
self.label_map = label_map
self.num_classes = len(label_map)
else:
self.num_classes = num_classes
def net(self, input_ids, position_ids, segment_ids, input_mask):
"""
create neural network.
if task == 'sequence_classification':
task = 'seq-cls'
logger.warning(
"current task name 'sequence_classification' was renamed to 'seq-cls', "
"'sequence_classification' has been deprecated and will be removed in the future.",
)
if task == 'seq-cls':
self.model = ElectraForSequenceClassification.from_pretrained(
pretrained_model_name_or_path='chinese-electra-base',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = paddle.metric.Accuracy()
elif task == 'token-cls':
self.model = ElectraForTokenClassification.from_pretrained(
pretrained_model_name_or_path='chinese-electra-base',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = ChunkEvaluator(
label_list=[self.label_map[i] for i in sorted(self.label_map.keys())]
)
elif task is None:
self.model = ElectraModel.from_pretrained(pretrained_model_name_or_path='chinese-electra-base', **kwargs)
else:
raise RuntimeError("Unknown task {}, task should be one in {}".format(
task, self._tasks_supported))
Args:
input_ids (tensor): the word ids.
position_ids (tensor): the position ids.
segment_ids (tensor): the segment ids.
input_mask (tensor): the padding mask.
self.task = task
Returns:
pooled_output (tensor): sentence-level output for classification task.
sequence_output (tensor): token-level output for sequence task.
"""
electra = ElectraModel(src_ids=input_ids,
position_ids=position_ids,
sentence_ids=segment_ids,
input_mask=input_mask,
config=self.electra_config,
use_fp16=False)
pooled_output = electra.get_pooled_output()
sequence_output = electra.get_sequence_output()
return pooled_output, sequence_output
if load_checkpoint is not None and os.path.isfile(load_checkpoint):
state_dict = paddle.load(load_checkpoint)
self.set_state_dict(state_dict)
logger.info('Loaded parameters from %s' % os.path.abspath(load_checkpoint))
def forward(self, input_ids, token_type_ids=None, position_ids=None, attention_mask=None, seq_lengths=None, labels=None):
result = self.model(input_ids, token_type_ids, position_ids, attention_mask)
if self.task == 'seq-cls':
logits = result
probs = F.softmax(logits, axis=1)
if labels is not None:
loss = self.criterion(logits, labels)
correct = self.metric.compute(probs, labels)
acc = self.metric.update(correct)
return probs, loss, {'acc': acc}
return probs
elif self.task == 'token-cls':
logits = result
token_level_probs = F.softmax(logits, axis=-1)
preds = token_level_probs.argmax(axis=-1)
if labels is not None:
loss = self.criterion(logits, labels.unsqueeze(-1))
num_infer_chunks, num_label_chunks, num_correct_chunks = \
self.metric.compute(None, seq_lengths, preds, labels)
self.metric.update(
num_infer_chunks.numpy(), num_label_chunks.numpy(), num_correct_chunks.numpy())
_, _, f1_score = map(float, self.metric.accumulate())
return token_level_probs, loss, {'f1_score': f1_score}
return token_level_probs
else:
sequence_output, pooled_output = result
return sequence_output, pooled_output
if __name__ == '__main__':
test_module = Electra()
@staticmethod
def get_tokenizer(*args, **kwargs):
"""
Gets the tokenizer that is customized for this module.
"""
return ElectraTokenizer.from_pretrained(
pretrained_model_name_or_path='chinese-electra-base', *args, **kwargs)
```shell
$ hub install chinese-electra-small==1.0.0
$ hub install chinese-electra-small==2.0.0
```
<p align="center">
<img src="https://github.com/ymcui/Chinese-ELECTRA/blob/master/pics/model.png" hspace='10'/> <br />
<img src="http://bj.bcebos.com/ibox-thumbnail98/1a5578bfbe1ad629035f7ad1eb3d0bce?authorization=bce-auth-v1%2Ffbe74140929444858491fbf2b6bc0935%2F2020-03-31T06%3A45%3A51Z%2F1800%2F%2F02b8749292f8ba1c606410d0e4e5dbabdf1d367d80da395887775d36424ac13e" hspace='10'/> <br />
</p>
更多详情请参考[ELECTRA论文](https://openreview.net/pdf?id=r1xMH1BtvB)
## API
```python
def context(
trainable=True,
max_seq_len=128
def __init__(
task=None,
load_checkpoint=None,
label_map=None,
num_classes=2,
**kwargs,
)
```
用于获取Module的上下文信息,得到输入、输出以及预训练的Paddle Program副本
**参数**
> trainable:设置为True时,Module中的参数在Fine-tune时也会随之训练,否则保持不变。
> max_seq_len:BERT模型的最大序列长度,若序列长度不足,会通过padding方式补到**max_seq_len**, 若序列长度大于该值,则会以截断方式让序列长度为**max_seq_len**,max_seq_len可取值范围为0~512;
创建Module对象(动态图组网版本)。
**返回**
> inputs:dict类型,有以下字段:
> >**input_ids**存放输入文本tokenize后各token对应BERT词汇表的word ids, shape为\[batch_size, max_seq_len\],int64类型;
> >**position_ids**存放输入文本tokenize后各token所在该文本的位置,shape为\[batch_size, max_seq_len\],int64类型;
> >**segment_ids**存放各token所在文本的标识(token属于文本1或者文本2),shape为\[batch_size, max_seq_len\],int64类型;
> >**input_mask**存放token是否为padding的标识,shape为\[batch_size, max_seq_len\],int64类型;
>
> outputs:dict类型,Module的输出特征,有以下字段:
> >**pooled_output**字段存放句子粒度的特征,可用于文本分类等任务,shape为 \[batch_size, 768\],int64类型;
> >**sequence_output**字段存放字粒度的特征,可用于序列标注等任务,shape为 \[batch_size, seq_len, 768\],int64类型;
>
> program:包含该Module计算图的Program。
**参数**
* `task`: 任务名称,可为`seq-cls`(文本分类任务,原来的`sequence_classification`在未来会被弃用)或`token-cls`(序列标注任务)。
* `load_checkpoint`:使用PaddleHub Fine-tune api训练保存的模型参数文件路径。
* `label_map`:预测时的类别映射表。
* `num_classes`:分类任务的类别数,如果指定了`label_map`,此参数可不传,默认2分类。
* `**kwargs`:用户额外指定的关键字字典类型的参数。
```python
def get_embedding(
texts,
use_gpu=False,
batch_size=1
def predict(
data,
max_seq_len=128,
batch_size=1,
use_gpu=False
)
```
用于获取输入文本的句子粒度特征与字粒度特征
**参数**
> texts:输入文本列表,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。
> use_gpu:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
* `data`: 待预测数据,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。每个样例文本数量(1个或者2个)需和训练时保持一致。
* `max_seq_len`:模型处理文本的最大长度
* `batch_size`:模型批处理大小
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
> results:list类型,格式为\[\[sample\_a\_pooled\_feature, sample\_a\_seq\_feature\], \[sample\_b\_pooled\_feature, sample\_b\_seq\_feature\],…,\],其中每个元素都是对应样例的特征输出,每个样例都有句子粒度特征pooled\_feature与字粒度特征seq\_feature。
>
* `results`:list类型,不同任务类型的返回结果如下
* 文本分类:列表里包含每个句子的预测标签,格式为\[label\_1, label\_2, …,\]
* 序列标注:列表里包含每个句子每个token的预测标签,格式为\[\[token\_1, token\_2, …,\], \[token\_1, token\_2, …,\], …,\]
```python
def get_params_layer()
def get_embedding(
data,
use_gpu=False
)
```
用于获取参数层信息,该方法与ULMFiTStrategy联用可以严格按照层数设置分层学习率与逐层解冻。
用于获取输入文本的句子粒度特征与字粒度特征
**参数**
> 无
* `data`:输入文本列表,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
> params_layer:dict类型,key为参数名,值为参数所在层数
* `results`:list类型,格式为\[\[sample\_a\_pooled\_feature, sample\_a\_seq\_feature\], \[sample\_b\_pooled\_feature, sample\_b\_seq\_feature\],…,\],其中每个元素都是对应样例的特征输出,每个样例都有句子粒度特征pooled\_feature与字粒度特征seq\_feature。
**代码示例**
......@@ -75,45 +75,83 @@ def get_params_layer()
```python
import paddlehub as hub
# Load $ hub install chinese-electra-small pretrained model
module = hub.Module(name="chinese-electra-small")
inputs, outputs, program = module.context(trainable=True, max_seq_len=128)
data = [
['这个宾馆比较陈旧了,特价的房间也很一般。总体来说一般'],
['怀着十分激动的心情放映,可是看着看着发现,在放映完毕后,出现一集米老鼠的动画片'],
['作为老的四星酒店,房间依然很整洁,相当不错。机场接机服务很好,可以在车上办理入住手续,节省时间。'],
]
label_map = {0: 'negative', 1: 'positive'}
model = hub.Module(
name='chinese-electra-small',
version='2.0.0',
task='seq-cls',
load_checkpoint='/path/to/parameters',
label_map=label_map)
results = model.predict(data, max_seq_len=50, batch_size=1, use_gpu=False)
for idx, text in enumerate(data):
print('Data: {} \t Lable: {}'.format(text, results[idx]))
```
详情可参考PaddleHub示例:
- [文本分类](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/text_classification)
- [序列标注](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/sequence_labeling)
# Must feed all the tensor of chinese-electra-small's module need
input_ids = inputs["input_ids"]
position_ids = inputs["position_ids"]
segment_ids = inputs["segment_ids"]
input_mask = inputs["input_mask"]
## 服务部署
# Use "pooled_output" for sentence-level output.
pooled_output = outputs["pooled_output"]
PaddleHub Serving可以部署一个在线获取预训练词向量。
# Use "sequence_output" for token-level output.
sequence_output = outputs["sequence_output"]
### Step1: 启动PaddleHub Serving
# Use "get_embedding" to get embedding result.
embedding_result = module.get_embedding(texts=[["Sample1_text_a"],["Sample2_text_a","Sample2_text_b"]], use_gpu=True)
运行启动命令:
# Use "get_params_layer" to get params layer and used to ULMFiTStrategy.
params_layer = module.get_params_layer()
strategy = hub.finetune.strategy.ULMFiTStrategy(frz_params_layer=params_layer, dis_params_layer=params_layer)
```shell
$ hub serving start -m chinese-electra-small
```
这样就完成了一个获取预训练词向量服务化API的部署,默认端口号为8866。
**NOTE:** 如使用GPU预测,则需要在启动服务之前,请设置CUDA_VISIBLE_DEVICES环境变量,否则不用设置。
### Step2: 发送预测请求
配置好服务端,以下数行代码即可实现发送预测请求,获取预测结果
```python
import requests
import json
# 指定用于获取embedding的文本[[text_1], [text_2], ... ]}
text = [["今天是个好日子"], ["天气预报说今天要下雨"]]
# 以key的方式指定text传入预测方法的时的参数,此例中为"data"
# 对应本地部署,则为module.get_embedding(data=text)
data = {"data": text}
# 发送post请求,content-type类型应指定json方式,url中的ip地址需改为对应机器的ip
url = "http://10.12.121.132:8866/predict/chinese-electra-small"
# 指定post请求的headers为application/json方式
headers = {"Content-Type": "application/json"}
r = requests.post(url=url, headers=headers, data=json.dumps(data))
print(r.json())
```
## 查看代码
https://github.com/ymcui/Chinese-ELECTRA
## 依赖
paddlepaddle >= 1.6.2
paddlepaddle >= 2.0.0
paddlehub >= 1.6.0
paddlehub >= 2.0.0
## 更新历史
* 1.0.0
初始发布
* 2.0.0
全面升级动态图,接口有所变化。任务名称调整,增加序列标注任务`token-cls`
# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""ELECTRA model."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import six
import json
import paddle.fluid as fluid
from chinese_electra_small.model.transformer_encoder import encoder, pre_process_layer
class ElectraConfig(object):
def __init__(self, config_path):
self._config_dict = self._parse(config_path)
def _parse(self, config_path):
try:
with open(config_path) as json_file:
config_dict = json.load(json_file)
except Exception:
raise IOError("Error in parsing electra model config file '%s'" % config_path)
else:
return config_dict
def __getitem__(self, key):
return self._config_dict[key]
def print_config(self):
for arg, value in sorted(six.iteritems(self._config_dict)):
print('%s: %s' % (arg, value))
print('------------------------------------------------')
class ElectraModel(object):
def __init__(self, src_ids, position_ids, sentence_ids, input_mask, config, weight_sharing=True, use_fp16=False):
self._emb_size = 128
self._hidden_size = config['hidden_size']
self._n_layer = config['num_hidden_layers']
self._n_head = config['num_attention_heads']
self._voc_size = config['vocab_size']
self._max_position_seq_len = config['max_position_embeddings']
self._sent_types = config['type_vocab_size']
self._hidden_act = config['hidden_act']
self._prepostprocess_dropout = config['hidden_dropout_prob']
self._attention_dropout = config['attention_probs_dropout_prob']
self._weight_sharing = weight_sharing
self._word_emb_name = "word_embedding"
self._pos_emb_name = "pos_embedding"
self._sent_emb_name = "sent_embedding"
self._dtype = "float16" if use_fp16 else "float32"
# Initialize all weigths by truncated normal initializer, and all biases
# will be initialized by constant zero by default.
self._param_initializer = fluid.initializer.TruncatedNormal(scale=config['initializer_range'])
self._build_model(src_ids, position_ids, sentence_ids, input_mask)
def _build_model(self, src_ids, position_ids, sentence_ids, input_mask):
# padding id in vocabulary must be set to 0
emb_out = fluid.layers.embedding(input=src_ids,
size=[self._voc_size, self._emb_size],
dtype=self._dtype,
param_attr=fluid.ParamAttr(name=self._word_emb_name,
initializer=self._param_initializer),
is_sparse=False)
position_emb_out = fluid.layers.embedding(input=position_ids,
size=[self._max_position_seq_len, self._emb_size],
dtype=self._dtype,
param_attr=fluid.ParamAttr(name=self._pos_emb_name,
initializer=self._param_initializer))
sent_emb_out = fluid.layers.embedding(sentence_ids,
size=[self._sent_types, self._emb_size],
dtype=self._dtype,
param_attr=fluid.ParamAttr(name=self._sent_emb_name,
initializer=self._param_initializer))
emb_out = emb_out + position_emb_out
emb_out = emb_out + sent_emb_out
emb_out = pre_process_layer(emb_out, 'nd', self._prepostprocess_dropout, name='pre_encoder')
if self._emb_size != self._hidden_size:
emb_out = fluid.layers.fc(input=emb_out,
size=self._hidden_size,
act=None,
param_attr=fluid.ParamAttr(name="embeddings_project.w_0",
initializer=self._param_initializer),
num_flatten_dims=2,
bias_attr="embeddings_project.b_0")
if self._dtype == "float16":
input_mask = fluid.layers.cast(x=input_mask, dtype=self._dtype)
self_attn_mask = fluid.layers.matmul(x=input_mask, y=input_mask, transpose_y=True)
self_attn_mask = fluid.layers.scale(x=self_attn_mask, scale=10000.0, bias=-1.0, bias_after_scale=False)
n_head_self_attn_mask = fluid.layers.stack(x=[self_attn_mask] * self._n_head, axis=1)
n_head_self_attn_mask.stop_gradient = True
self._enc_out = encoder(enc_input=emb_out,
attn_bias=n_head_self_attn_mask,
n_layer=self._n_layer,
n_head=self._n_head,
d_key=self._hidden_size // self._n_head,
d_value=self._hidden_size // self._n_head,
d_model=self._hidden_size,
d_inner_hid=self._hidden_size * 4,
prepostprocess_dropout=self._prepostprocess_dropout,
attention_dropout=self._attention_dropout,
relu_dropout=0,
hidden_act=self._hidden_act,
preprocess_cmd="",
postprocess_cmd="dan",
param_initializer=self._param_initializer,
name='encoder')
def get_sequence_output(self):
return self._enc_out
def get_pooled_output(self):
"""Get the first feature of each sequence for classification"""
next_sent_feat = fluid.layers.slice(input=self._enc_out, axes=[1], starts=[0], ends=[1])
return next_sent_feat
def get_pretraining_output(self, mask_label, mask_pos, labels):
"""Get the loss & accuracy for pretraining"""
mask_pos = fluid.layers.cast(x=mask_pos, dtype='int32')
# extract the first token feature in each sentence
next_sent_feat = self.get_pooled_output()
reshaped_emb_out = fluid.layers.reshape(x=self._enc_out, shape=[-1, self._hidden_size])
# extract masked tokens' feature
mask_feat = fluid.layers.gather(input=reshaped_emb_out, index=mask_pos)
# transform: fc
mask_trans_feat = fluid.layers.fc(input=mask_feat,
size=self._hidden_size,
act=self._hidden_act,
param_attr=fluid.ParamAttr(name='mask_lm_trans_fc.w_0',
initializer=self._param_initializer),
bias_attr=fluid.ParamAttr(name='mask_lm_trans_fc.b_0'))
# transform: layer norm
mask_trans_feat = pre_process_layer(mask_trans_feat, 'n', name='mask_lm_trans')
mask_lm_out_bias_attr = fluid.ParamAttr(name="mask_lm_out_fc.b_0",
initializer=fluid.initializer.Constant(value=0.0))
if self._weight_sharing:
fc_out = fluid.layers.matmul(x=mask_trans_feat,
y=fluid.default_main_program().global_block().var(self._word_emb_name),
transpose_y=True)
fc_out += fluid.layers.create_parameter(shape=[self._voc_size],
dtype=self._dtype,
attr=mask_lm_out_bias_attr,
is_bias=True)
else:
fc_out = fluid.layers.fc(input=mask_trans_feat,
size=self._voc_size,
param_attr=fluid.ParamAttr(name="mask_lm_out_fc.w_0",
initializer=self._param_initializer),
bias_attr=mask_lm_out_bias_attr)
mask_lm_loss = fluid.layers.softmax_with_cross_entropy(logits=fc_out, label=mask_label)
mean_mask_lm_loss = fluid.layers.mean(mask_lm_loss)
next_sent_fc_out = fluid.layers.fc(input=next_sent_feat,
size=2,
param_attr=fluid.ParamAttr(name="next_sent_fc.w_0",
initializer=self._param_initializer),
bias_attr="next_sent_fc.b_0")
next_sent_loss, next_sent_softmax = fluid.layers.softmax_with_cross_entropy(logits=next_sent_fc_out,
label=labels,
return_softmax=True)
next_sent_acc = fluid.layers.accuracy(input=next_sent_softmax, label=labels)
mean_next_sent_loss = fluid.layers.mean(next_sent_loss)
loss = mean_next_sent_loss + mean_mask_lm_loss
return next_sent_acc, mean_mask_lm_loss, loss
# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Transformer encoder."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from functools import partial
import paddle.fluid as fluid
import paddle.fluid.layers as layers
def multi_head_attention(queries,
keys,
values,
attn_bias,
d_key,
d_value,
d_model,
n_head=1,
dropout_rate=0.,
cache=None,
param_initializer=None,
name='multi_head_att'):
"""
Multi-Head Attention. Note that attn_bias is added to the logit before
computing softmax activiation to mask certain selected positions so that
they will not considered in attention weights.
"""
keys = queries if keys is None else keys
values = keys if values is None else values
if not (len(queries.shape) == len(keys.shape) == len(values.shape) == 3):
raise ValueError("Inputs: quries, keys and values should all be 3-D tensors.")
def __compute_qkv(queries, keys, values, n_head, d_key, d_value):
"""
Add linear projection to queries, keys, and values.
"""
q = layers.fc(input=queries,
size=d_key * n_head,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_query_fc.w_0', initializer=param_initializer),
bias_attr=name + '_query_fc.b_0')
k = layers.fc(input=keys,
size=d_key * n_head,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_key_fc.w_0', initializer=param_initializer),
bias_attr=name + '_key_fc.b_0')
v = layers.fc(input=values,
size=d_value * n_head,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_value_fc.w_0', initializer=param_initializer),
bias_attr=name + '_value_fc.b_0')
return q, k, v
def __split_heads(x, n_head):
"""
Reshape the last dimension of inpunt tensor x so that it becomes two
dimensions and then transpose. Specifically, input a tensor with shape
[bs, max_sequence_length, n_head * hidden_dim] then output a tensor
with shape [bs, n_head, max_sequence_length, hidden_dim].
"""
hidden_size = x.shape[-1]
# The value 0 in shape attr means copying the corresponding dimension
# size of the input as the output dimension size.
reshaped = layers.reshape(x=x, shape=[0, 0, n_head, hidden_size // n_head], inplace=True)
# permuate the dimensions into:
# [batch_size, n_head, max_sequence_len, hidden_size_per_head]
return layers.transpose(x=reshaped, perm=[0, 2, 1, 3])
def __combine_heads(x):
"""
Transpose and then reshape the last two dimensions of inpunt tensor x
so that it becomes one dimension, which is reverse to __split_heads.
"""
if len(x.shape) == 3: return x
if len(x.shape) != 4:
raise ValueError("Input(x) should be a 4-D Tensor.")
trans_x = layers.transpose(x, perm=[0, 2, 1, 3])
# The value 0 in shape attr means copying the corresponding dimension
# size of the input as the output dimension size.
return layers.reshape(x=trans_x, shape=[0, 0, trans_x.shape[2] * trans_x.shape[3]], inplace=True)
def scaled_dot_product_attention(q, k, v, attn_bias, d_key, dropout_rate):
"""
Scaled Dot-Product Attention
"""
scaled_q = layers.scale(x=q, scale=d_key**-0.5)
product = layers.matmul(x=scaled_q, y=k, transpose_y=True)
if attn_bias:
product += attn_bias
weights = layers.softmax(product)
if dropout_rate:
weights = layers.dropout(weights,
dropout_prob=dropout_rate,
dropout_implementation="upscale_in_train",
is_test=False)
out = layers.matmul(weights, v)
return out
q, k, v = __compute_qkv(queries, keys, values, n_head, d_key, d_value)
if cache is not None: # use cache and concat time steps
# Since the inplace reshape in __split_heads changes the shape of k and
# v, which is the cache input for next time step, reshape the cache
# input from the previous time step first.
k = cache["k"] = layers.concat([layers.reshape(cache["k"], shape=[0, 0, d_model]), k], axis=1)
v = cache["v"] = layers.concat([layers.reshape(cache["v"], shape=[0, 0, d_model]), v], axis=1)
q = __split_heads(q, n_head)
k = __split_heads(k, n_head)
v = __split_heads(v, n_head)
ctx_multiheads = scaled_dot_product_attention(q, k, v, attn_bias, d_key, dropout_rate)
out = __combine_heads(ctx_multiheads)
# Project back to the model size.
proj_out = layers.fc(input=out,
size=d_model,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_output_fc.w_0', initializer=param_initializer),
bias_attr=name + '_output_fc.b_0')
return proj_out
def positionwise_feed_forward(x, d_inner_hid, d_hid, dropout_rate, hidden_act, param_initializer=None, name='ffn'):
"""
Position-wise Feed-Forward Networks.
This module consists of two linear transformations with a ReLU activation
in between, which is applied to each position separately and identically.
"""
hidden = layers.fc(input=x,
size=d_inner_hid,
num_flatten_dims=2,
act=hidden_act,
param_attr=fluid.ParamAttr(name=name + '_fc_0.w_0', initializer=param_initializer),
bias_attr=name + '_fc_0.b_0')
if dropout_rate:
hidden = layers.dropout(hidden,
dropout_prob=dropout_rate,
dropout_implementation="upscale_in_train",
is_test=False)
out = layers.fc(input=hidden,
size=d_hid,
num_flatten_dims=2,
param_attr=fluid.ParamAttr(name=name + '_fc_1.w_0', initializer=param_initializer),
bias_attr=name + '_fc_1.b_0')
return out
def pre_post_process_layer(prev_out, out, process_cmd, dropout_rate=0., name=''):
"""
Add residual connection, layer normalization and droput to the out tensor
optionally according to the value of process_cmd.
This will be used before or after multi-head attention and position-wise
feed-forward networks.
"""
for cmd in process_cmd:
if cmd == "a": # add residual connection
out = out + prev_out if prev_out else out
elif cmd == "n": # add layer normalization
out_dtype = out.dtype
if out_dtype == fluid.core.VarDesc.VarType.FP16:
out = layers.cast(x=out, dtype="float32")
out = layers.layer_norm(out,
begin_norm_axis=len(out.shape) - 1,
param_attr=fluid.ParamAttr(name=name + '_layer_norm_scale',
initializer=fluid.initializer.Constant(1.)),
bias_attr=fluid.ParamAttr(name=name + '_layer_norm_bias',
initializer=fluid.initializer.Constant(0.)))
if out_dtype == fluid.core.VarDesc.VarType.FP16:
out = layers.cast(x=out, dtype="float16")
elif cmd == "d": # add dropout
if dropout_rate:
out = layers.dropout(out,
dropout_prob=dropout_rate,
dropout_implementation="upscale_in_train",
is_test=False)
return out
pre_process_layer = partial(pre_post_process_layer, None)
post_process_layer = pre_post_process_layer
def encoder_layer(enc_input,
attn_bias,
n_head,
d_key,
d_value,
d_model,
d_inner_hid,
prepostprocess_dropout,
attention_dropout,
relu_dropout,
hidden_act,
preprocess_cmd="n",
postprocess_cmd="da",
param_initializer=None,
name=''):
"""The encoder layers that can be stacked to form a deep encoder.
This module consits of a multi-head (self) attention followed by
position-wise feed-forward networks and both the two components companied
with the post_process_layer to add residual connection, layer normalization
and droput.
"""
attn_output = multi_head_attention(pre_process_layer(enc_input,
preprocess_cmd,
prepostprocess_dropout,
name=name + '_pre_att'),
None,
None,
attn_bias,
d_key,
d_value,
d_model,
n_head,
attention_dropout,
param_initializer=param_initializer,
name=name + '_multi_head_att')
attn_output = post_process_layer(enc_input,
attn_output,
postprocess_cmd,
prepostprocess_dropout,
name=name + '_post_att')
ffd_output = positionwise_feed_forward(pre_process_layer(attn_output,
preprocess_cmd,
prepostprocess_dropout,
name=name + '_pre_ffn'),
d_inner_hid,
d_model,
relu_dropout,
hidden_act,
param_initializer=param_initializer,
name=name + '_ffn')
return post_process_layer(attn_output, ffd_output, postprocess_cmd, prepostprocess_dropout, name=name + '_post_ffn')
def encoder(enc_input,
attn_bias,
n_layer,
n_head,
d_key,
d_value,
d_model,
d_inner_hid,
prepostprocess_dropout,
attention_dropout,
relu_dropout,
hidden_act,
preprocess_cmd="n",
postprocess_cmd="da",
param_initializer=None,
name=''):
"""
The encoder is composed of a stack of identical layers returned by calling
encoder_layer.
"""
for i in range(n_layer):
enc_output = encoder_layer(enc_input,
attn_bias,
n_head,
d_key,
d_value,
d_model,
d_inner_hid,
prepostprocess_dropout,
attention_dropout,
relu_dropout,
hidden_act,
preprocess_cmd,
postprocess_cmd,
param_initializer=param_initializer,
name=name + '_layer_' + str(i))
enc_input = enc_output
enc_output = pre_process_layer(enc_output, preprocess_cmd, prepostprocess_dropout, name="post_encoder")
return enc_output
# coding:utf-8
# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License"
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
......@@ -12,62 +11,120 @@
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from typing import Dict
import os
from paddlehub import TransformerModule
from paddlehub.module.module import moduleinfo
import paddle
import paddle.nn as nn
import paddle.nn.functional as F
from chinese_electra_small.model.electra import ElectraConfig, ElectraModel
from paddlenlp.transformers.electra.modeling import ElectraForSequenceClassification, ElectraForTokenClassification, ElectraModel
from paddlenlp.transformers.electra.tokenizer import ElectraTokenizer
from paddlenlp.metrics import ChunkEvaluator
from paddlehub.module.module import moduleinfo
from paddlehub.module.nlp_module import TransformerModule
from paddlehub.utils.log import logger
@moduleinfo(
name="chinese-electra-small",
version="1.0.0",
summary="chinese-electra-small, 12-layer, 256-hidden, 4-heads, 12M parameters",
version="2.0.0",
summary=
"chinese-electra-small, 12-layer, 256-hidden, 4-heads, 12M parameters. The module is executed as paddle.dygraph.",
author="ymcui",
author_email="ymcui@ir.hit.edu.cn",
type="nlp/semantic_model",
meta=TransformerModule,
)
class Electra(TransformerModule):
def _initialize(self):
self.MAX_SEQ_LEN = 512
self.params_path = os.path.join(self.directory, "assets", "params")
self.vocab_path = os.path.join(self.directory, "assets", "vocab.txt")
class Electra(nn.Layer):
"""
Electra model
"""
electra_config_path = os.path.join(self.directory, "assets", "config.json")
self.electra_config = ElectraConfig(electra_config_path)
def __init__(
self,
task: str = None,
load_checkpoint: str = None,
label_map: Dict = None,
num_classes: int = 2,
**kwargs,
):
super(Electra, self).__init__()
if label_map:
self.label_map = label_map
self.num_classes = len(label_map)
else:
self.num_classes = num_classes
def net(self, input_ids, position_ids, segment_ids, input_mask):
"""
create neural network.
if task == 'sequence_classification':
task = 'seq-cls'
logger.warning(
"current task name 'sequence_classification' was renamed to 'seq-cls', "
"'sequence_classification' has been deprecated and will be removed in the future.",
)
if task == 'seq-cls':
self.model = ElectraForSequenceClassification.from_pretrained(
pretrained_model_name_or_path='chinese-electra-small',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = paddle.metric.Accuracy()
elif task == 'token-cls':
self.model = ElectraForTokenClassification.from_pretrained(
pretrained_model_name_or_path='chinese-electra-small',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = ChunkEvaluator(
label_list=[self.label_map[i] for i in sorted(self.label_map.keys())]
)
elif task is None:
self.model = ElectraModel.from_pretrained(pretrained_model_name_or_path='chinese-electra-small', **kwargs)
else:
raise RuntimeError("Unknown task {}, task should be one in {}".format(
task, self._tasks_supported))
Args:
input_ids (tensor): the word ids.
position_ids (tensor): the position ids.
segment_ids (tensor): the segment ids.
input_mask (tensor): the padding mask.
self.task = task
Returns:
pooled_output (tensor): sentence-level output for classification task.
sequence_output (tensor): token-level output for sequence task.
"""
electra = ElectraModel(src_ids=input_ids,
position_ids=position_ids,
sentence_ids=segment_ids,
input_mask=input_mask,
config=self.electra_config,
use_fp16=False)
pooled_output = electra.get_pooled_output()
sequence_output = electra.get_sequence_output()
return pooled_output, sequence_output
if load_checkpoint is not None and os.path.isfile(load_checkpoint):
state_dict = paddle.load(load_checkpoint)
self.set_state_dict(state_dict)
logger.info('Loaded parameters from %s' % os.path.abspath(load_checkpoint))
def forward(self, input_ids, token_type_ids=None, position_ids=None, attention_mask=None, seq_lengths=None, labels=None):
result = self.model(input_ids, token_type_ids, position_ids, attention_mask)
if self.task == 'seq-cls':
logits = result
probs = F.softmax(logits, axis=1)
if labels is not None:
loss = self.criterion(logits, labels)
correct = self.metric.compute(probs, labels)
acc = self.metric.update(correct)
return probs, loss, {'acc': acc}
return probs
elif self.task == 'token-cls':
logits = result
token_level_probs = F.softmax(logits, axis=-1)
preds = token_level_probs.argmax(axis=-1)
if labels is not None:
loss = self.criterion(logits, labels.unsqueeze(-1))
num_infer_chunks, num_label_chunks, num_correct_chunks = \
self.metric.compute(None, seq_lengths, preds, labels)
self.metric.update(
num_infer_chunks.numpy(), num_label_chunks.numpy(), num_correct_chunks.numpy())
_, _, f1_score = map(float, self.metric.accumulate())
return token_level_probs, loss, {'f1_score': f1_score}
return token_level_probs
else:
sequence_output, pooled_output = result
return sequence_output, pooled_output
if __name__ == '__main__':
test_module = Electra()
@staticmethod
def get_tokenizer(*args, **kwargs):
"""
Gets the tokenizer that is customized for this module.
"""
return ElectraTokenizer.from_pretrained(
pretrained_model_name_or_path='chinese-electra-small', *args, **kwargs)
```shell
$ hub install electra-base==1.0.0
```
<p align="center">
<img src="http://bj.bcebos.com/ibox-thumbnail98/1a5578bfbe1ad629035f7ad1eb3d0bce?authorization=bce-auth-v1%2Ffbe74140929444858491fbf2b6bc0935%2F2020-03-31T06%3A45%3A51Z%2F1800%2F%2F02b8749292f8ba1c606410d0e4e5dbabdf1d367d80da395887775d36424ac13e" hspace='10'/> <br />
</p>
更多详情请参考[ELECTRA论文](https://openreview.net/pdf?id=r1xMH1BtvB)
## API
```python
def __init__(
task=None,
load_checkpoint=None,
label_map=None,
num_classes=2,
**kwargs,
)
```
创建Module对象(动态图组网版本)。
**参数**
* `task`: 任务名称,可为`seq-cls`(文本分类任务,原来的`sequence_classification`在未来会被弃用)或`token-cls`(序列标注任务)。
* `load_checkpoint`:使用PaddleHub Fine-tune api训练保存的模型参数文件路径。
* `label_map`:预测时的类别映射表。
* `num_classes`:分类任务的类别数,如果指定了`label_map`,此参数可不传,默认2分类。
* `**kwargs`:用户额外指定的关键字字典类型的参数。
```python
def predict(
data,
max_seq_len=128,
batch_size=1,
use_gpu=False
)
```
**参数**
* `data`: 待预测数据,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。每个样例文本数量(1个或者2个)需和训练时保持一致。
* `max_seq_len`:模型处理文本的最大长度
* `batch_size`:模型批处理大小
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
* `results`:list类型,不同任务类型的返回结果如下
* 文本分类:列表里包含每个句子的预测标签,格式为\[label\_1, label\_2, …,\]
* 序列标注:列表里包含每个句子每个token的预测标签,格式为\[\[token\_1, token\_2, …,\], \[token\_1, token\_2, …,\], …,\]
```python
def get_embedding(
data,
use_gpu=False
)
```
用于获取输入文本的句子粒度特征与字粒度特征
**参数**
* `data`:输入文本列表,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
* `results`:list类型,格式为\[\[sample\_a\_pooled\_feature, sample\_a\_seq\_feature\], \[sample\_b\_pooled\_feature, sample\_b\_seq\_feature\],…,\],其中每个元素都是对应样例的特征输出,每个样例都有句子粒度特征pooled\_feature与字粒度特征seq\_feature。
**代码示例**
```python
import paddlehub as hub
data = [
['这个宾馆比较陈旧了,特价的房间也很一般。总体来说一般'],
['怀着十分激动的心情放映,可是看着看着发现,在放映完毕后,出现一集米老鼠的动画片'],
['作为老的四星酒店,房间依然很整洁,相当不错。机场接机服务很好,可以在车上办理入住手续,节省时间。'],
]
label_map = {0: 'negative', 1: 'positive'}
model = hub.Module(
name='electra-base',
version='1.0.0',
task='seq-cls',
load_checkpoint='/path/to/parameters',
label_map=label_map)
results = model.predict(data, max_seq_len=50, batch_size=1, use_gpu=False)
for idx, text in enumerate(data):
print('Data: {} \t Lable: {}'.format(text, results[idx]))
```
详情可参考PaddleHub示例:
- [文本分类](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/text_classification)
- [序列标注](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/sequence_labeling)
## 服务部署
PaddleHub Serving可以部署一个在线获取预训练词向量。
### Step1: 启动PaddleHub Serving
运行启动命令:
```shell
$ hub serving start -m electra-base
```
这样就完成了一个获取预训练词向量服务化API的部署,默认端口号为8866。
**NOTE:** 如使用GPU预测,则需要在启动服务之前,请设置CUDA_VISIBLE_DEVICES环境变量,否则不用设置。
### Step2: 发送预测请求
配置好服务端,以下数行代码即可实现发送预测请求,获取预测结果
```python
import requests
import json
# 指定用于获取embedding的文本[[text_1], [text_2], ... ]}
text = [["今天是个好日子"], ["天气预报说今天要下雨"]]
# 以key的方式指定text传入预测方法的时的参数,此例中为"data"
# 对应本地部署,则为module.get_embedding(data=text)
data = {"data": text}
# 发送post请求,content-type类型应指定json方式,url中的ip地址需改为对应机器的ip
url = "http://10.12.121.132:8866/predict/electra-base"
# 指定post请求的headers为application/json方式
headers = {"Content-Type": "application/json"}
r = requests.post(url=url, headers=headers, data=json.dumps(data))
print(r.json())
```
## 查看代码
https://github.com/google-research/electra
## 依赖
paddlepaddle >= 2.0.0
paddlehub >= 2.0.0
## 更新历史
* 1.0.0
初始发布,动态图版本模型,支持文本分类`seq-cls`和序列标注`token-cls`任务的fine-tune
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import Dict
import os
import paddle
import paddle.nn as nn
import paddle.nn.functional as F
from paddlenlp.transformers.electra.modeling import ElectraForSequenceClassification, ElectraForTokenClassification, ElectraModel
from paddlenlp.transformers.electra.tokenizer import ElectraTokenizer
from paddlenlp.metrics import ChunkEvaluator
from paddlehub.module.module import moduleinfo
from paddlehub.module.nlp_module import TransformerModule
from paddlehub.utils.log import logger
@moduleinfo(
name="electra-base",
version="1.0.0",
summary=
"electra-base, 12-layer, 768-hidden, 12-heads, 110M parameters. The module is executed as paddle.dygraph.",
author="paddlepaddle",
author_email="",
type="nlp/semantic_model",
meta=TransformerModule,
)
class Electra(nn.Layer):
"""
Electra model
"""
def __init__(
self,
task: str = None,
load_checkpoint: str = None,
label_map: Dict = None,
num_classes: int = 2,
**kwargs,
):
super(Electra, self).__init__()
if label_map:
self.label_map = label_map
self.num_classes = len(label_map)
else:
self.num_classes = num_classes
if task == 'sequence_classification':
task = 'seq-cls'
logger.warning(
"current task name 'sequence_classification' was renamed to 'seq-cls', "
"'sequence_classification' has been deprecated and will be removed in the future.",
)
if task == 'seq-cls':
self.model = ElectraForSequenceClassification.from_pretrained(
pretrained_model_name_or_path='electra-base',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = paddle.metric.Accuracy()
elif task == 'token-cls':
self.model = ElectraForTokenClassification.from_pretrained(
pretrained_model_name_or_path='electra-base',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = ChunkEvaluator(
label_list=[self.label_map[i] for i in sorted(self.label_map.keys())]
)
elif task is None:
self.model = ElectraModel.from_pretrained(pretrained_model_name_or_path='electra-base', **kwargs)
else:
raise RuntimeError("Unknown task {}, task should be one in {}".format(
task, self._tasks_supported))
self.task = task
if load_checkpoint is not None and os.path.isfile(load_checkpoint):
state_dict = paddle.load(load_checkpoint)
self.set_state_dict(state_dict)
logger.info('Loaded parameters from %s' % os.path.abspath(load_checkpoint))
def forward(self, input_ids, token_type_ids=None, position_ids=None, attention_mask=None, seq_lengths=None, labels=None):
result = self.model(input_ids, token_type_ids, position_ids, attention_mask)
if self.task == 'seq-cls':
logits = result
probs = F.softmax(logits, axis=1)
if labels is not None:
loss = self.criterion(logits, labels)
correct = self.metric.compute(probs, labels)
acc = self.metric.update(correct)
return probs, loss, {'acc': acc}
return probs
elif self.task == 'token-cls':
logits = result
token_level_probs = F.softmax(logits, axis=-1)
preds = token_level_probs.argmax(axis=-1)
if labels is not None:
loss = self.criterion(logits, labels.unsqueeze(-1))
num_infer_chunks, num_label_chunks, num_correct_chunks = \
self.metric.compute(None, seq_lengths, preds, labels)
self.metric.update(
num_infer_chunks.numpy(), num_label_chunks.numpy(), num_correct_chunks.numpy())
_, _, f1_score = map(float, self.metric.accumulate())
return token_level_probs, loss, {'f1_score': f1_score}
return token_level_probs
else:
sequence_output, pooled_output = result
return sequence_output, pooled_output
@staticmethod
def get_tokenizer(*args, **kwargs):
"""
Gets the tokenizer that is customized for this module.
"""
return ElectraTokenizer.from_pretrained(
pretrained_model_name_or_path='electra-base', *args, **kwargs)
```shell
$ hub install electra-large==1.0.0
```
<p align="center">
<img src="http://bj.bcebos.com/ibox-thumbnail98/1a5578bfbe1ad629035f7ad1eb3d0bce?authorization=bce-auth-v1%2Ffbe74140929444858491fbf2b6bc0935%2F2020-03-31T06%3A45%3A51Z%2F1800%2F%2F02b8749292f8ba1c606410d0e4e5dbabdf1d367d80da395887775d36424ac13e" hspace='10'/> <br />
</p>
更多详情请参考[ELECTRA论文](https://openreview.net/pdf?id=r1xMH1BtvB)
## API
```python
def __init__(
task=None,
load_checkpoint=None,
label_map=None,
num_classes=2,
**kwargs,
)
```
创建Module对象(动态图组网版本)。
**参数**
* `task`: 任务名称,可为`seq-cls`(文本分类任务,原来的`sequence_classification`在未来会被弃用)或`token-cls`(序列标注任务)。
* `load_checkpoint`:使用PaddleHub Fine-tune api训练保存的模型参数文件路径。
* `label_map`:预测时的类别映射表。
* `num_classes`:分类任务的类别数,如果指定了`label_map`,此参数可不传,默认2分类。
* `**kwargs`:用户额外指定的关键字字典类型的参数。
```python
def predict(
data,
max_seq_len=128,
batch_size=1,
use_gpu=False
)
```
**参数**
* `data`: 待预测数据,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。每个样例文本数量(1个或者2个)需和训练时保持一致。
* `max_seq_len`:模型处理文本的最大长度
* `batch_size`:模型批处理大小
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
* `results`:list类型,不同任务类型的返回结果如下
* 文本分类:列表里包含每个句子的预测标签,格式为\[label\_1, label\_2, …,\]
* 序列标注:列表里包含每个句子每个token的预测标签,格式为\[\[token\_1, token\_2, …,\], \[token\_1, token\_2, …,\], …,\]
```python
def get_embedding(
data,
use_gpu=False
)
```
用于获取输入文本的句子粒度特征与字粒度特征
**参数**
* `data`:输入文本列表,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
* `results`:list类型,格式为\[\[sample\_a\_pooled\_feature, sample\_a\_seq\_feature\], \[sample\_b\_pooled\_feature, sample\_b\_seq\_feature\],…,\],其中每个元素都是对应样例的特征输出,每个样例都有句子粒度特征pooled\_feature与字粒度特征seq\_feature。
**代码示例**
```python
import paddlehub as hub
data = [
['这个宾馆比较陈旧了,特价的房间也很一般。总体来说一般'],
['怀着十分激动的心情放映,可是看着看着发现,在放映完毕后,出现一集米老鼠的动画片'],
['作为老的四星酒店,房间依然很整洁,相当不错。机场接机服务很好,可以在车上办理入住手续,节省时间。'],
]
label_map = {0: 'negative', 1: 'positive'}
model = hub.Module(
name='electra-large',
version='1.0.0',
task='seq-cls',
load_checkpoint='/path/to/parameters',
label_map=label_map)
results = model.predict(data, max_seq_len=50, batch_size=1, use_gpu=False)
for idx, text in enumerate(data):
print('Data: {} \t Lable: {}'.format(text, results[idx]))
```
详情可参考PaddleHub示例:
- [文本分类](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/text_classification)
- [序列标注](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/sequence_labeling)
## 服务部署
PaddleHub Serving可以部署一个在线获取预训练词向量。
### Step1: 启动PaddleHub Serving
运行启动命令:
```shell
$ hub serving start -m electra-large
```
这样就完成了一个获取预训练词向量服务化API的部署,默认端口号为8866。
**NOTE:** 如使用GPU预测,则需要在启动服务之前,请设置CUDA_VISIBLE_DEVICES环境变量,否则不用设置。
### Step2: 发送预测请求
配置好服务端,以下数行代码即可实现发送预测请求,获取预测结果
```python
import requests
import json
# 指定用于获取embedding的文本[[text_1], [text_2], ... ]}
text = [["今天是个好日子"], ["天气预报说今天要下雨"]]
# 以key的方式指定text传入预测方法的时的参数,此例中为"data"
# 对应本地部署,则为module.get_embedding(data=text)
data = {"data": text}
# 发送post请求,content-type类型应指定json方式,url中的ip地址需改为对应机器的ip
url = "http://10.12.121.132:8866/predict/electra-large"
# 指定post请求的headers为application/json方式
headers = {"Content-Type": "application/json"}
r = requests.post(url=url, headers=headers, data=json.dumps(data))
print(r.json())
```
## 查看代码
https://github.com/google-research/electra
## 依赖
paddlepaddle >= 2.0.0
paddlehub >= 2.0.0
## 更新历史
* 1.0.0
初始发布,动态图版本模型,支持文本分类`seq-cls`和序列标注`token-cls`任务的fine-tune
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import Dict
import os
import paddle
import paddle.nn as nn
import paddle.nn.functional as F
from paddlenlp.transformers.electra.modeling import ElectraForSequenceClassification, ElectraForTokenClassification, ElectraModel
from paddlenlp.transformers.electra.tokenizer import ElectraTokenizer
from paddlenlp.metrics import ChunkEvaluator
from paddlehub.module.module import moduleinfo
from paddlehub.module.nlp_module import TransformerModule
from paddlehub.utils.log import logger
@moduleinfo(
name="electra-large",
version="1.0.0",
summary=
"electra-large, 24-layer, 1024-hidden, 16-heads, 335M parameters. The module is executed as paddle.dygraph.",
author="paddlepaddle",
author_email="",
type="nlp/semantic_model",
meta=TransformerModule,
)
class Electra(nn.Layer):
"""
Electra model
"""
def __init__(
self,
task: str = None,
load_checkpoint: str = None,
label_map: Dict = None,
num_classes: int = 2,
**kwargs,
):
super(Electra, self).__init__()
if label_map:
self.label_map = label_map
self.num_classes = len(label_map)
else:
self.num_classes = num_classes
if task == 'sequence_classification':
task = 'seq-cls'
logger.warning(
"current task name 'sequence_classification' was renamed to 'seq-cls', "
"'sequence_classification' has been deprecated and will be removed in the future.",
)
if task == 'seq-cls':
self.model = ElectraForSequenceClassification.from_pretrained(
pretrained_model_name_or_path='electra-large',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = paddle.metric.Accuracy()
elif task == 'token-cls':
self.model = ElectraForTokenClassification.from_pretrained(
pretrained_model_name_or_path='electra-large',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = ChunkEvaluator(
label_list=[self.label_map[i] for i in sorted(self.label_map.keys())]
)
elif task is None:
self.model = ElectraModel.from_pretrained(pretrained_model_name_or_path='electra-large', **kwargs)
else:
raise RuntimeError("Unknown task {}, task should be one in {}".format(
task, self._tasks_supported))
self.task = task
if load_checkpoint is not None and os.path.isfile(load_checkpoint):
state_dict = paddle.load(load_checkpoint)
self.set_state_dict(state_dict)
logger.info('Loaded parameters from %s' % os.path.abspath(load_checkpoint))
def forward(self, input_ids, token_type_ids=None, position_ids=None, attention_mask=None, seq_lengths=None, labels=None):
result = self.model(input_ids, token_type_ids, position_ids, attention_mask)
if self.task == 'seq-cls':
logits = result
probs = F.softmax(logits, axis=1)
if labels is not None:
loss = self.criterion(logits, labels)
correct = self.metric.compute(probs, labels)
acc = self.metric.update(correct)
return probs, loss, {'acc': acc}
return probs
elif self.task == 'token-cls':
logits = result
token_level_probs = F.softmax(logits, axis=-1)
preds = token_level_probs.argmax(axis=-1)
if labels is not None:
loss = self.criterion(logits, labels.unsqueeze(-1))
num_infer_chunks, num_label_chunks, num_correct_chunks = \
self.metric.compute(None, seq_lengths, preds, labels)
self.metric.update(
num_infer_chunks.numpy(), num_label_chunks.numpy(), num_correct_chunks.numpy())
_, _, f1_score = map(float, self.metric.accumulate())
return token_level_probs, loss, {'f1_score': f1_score}
return token_level_probs
else:
sequence_output, pooled_output = result
return sequence_output, pooled_output
@staticmethod
def get_tokenizer(*args, **kwargs):
"""
Gets the tokenizer that is customized for this module.
"""
return ElectraTokenizer.from_pretrained(
pretrained_model_name_or_path='electra-large', *args, **kwargs)
```shell
$ hub install electra-small==1.0.0
```
<p align="center">
<img src="http://bj.bcebos.com/ibox-thumbnail98/1a5578bfbe1ad629035f7ad1eb3d0bce?authorization=bce-auth-v1%2Ffbe74140929444858491fbf2b6bc0935%2F2020-03-31T06%3A45%3A51Z%2F1800%2F%2F02b8749292f8ba1c606410d0e4e5dbabdf1d367d80da395887775d36424ac13e" hspace='10'/> <br />
</p>
更多详情请参考[ELECTRA论文](https://openreview.net/pdf?id=r1xMH1BtvB)
## API
```python
def __init__(
task=None,
load_checkpoint=None,
label_map=None,
num_classes=2,
**kwargs,
)
```
创建Module对象(动态图组网版本)。
**参数**
* `task`: 任务名称,可为`seq-cls`(文本分类任务,原来的`sequence_classification`在未来会被弃用)或`token-cls`(序列标注任务)。
* `load_checkpoint`:使用PaddleHub Fine-tune api训练保存的模型参数文件路径。
* `label_map`:预测时的类别映射表。
* `num_classes`:分类任务的类别数,如果指定了`label_map`,此参数可不传,默认2分类。
* `**kwargs`:用户额外指定的关键字字典类型的参数。
```python
def predict(
data,
max_seq_len=128,
batch_size=1,
use_gpu=False
)
```
**参数**
* `data`: 待预测数据,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。每个样例文本数量(1个或者2个)需和训练时保持一致。
* `max_seq_len`:模型处理文本的最大长度
* `batch_size`:模型批处理大小
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
* `results`:list类型,不同任务类型的返回结果如下
* 文本分类:列表里包含每个句子的预测标签,格式为\[label\_1, label\_2, …,\]
* 序列标注:列表里包含每个句子每个token的预测标签,格式为\[\[token\_1, token\_2, …,\], \[token\_1, token\_2, …,\], …,\]
```python
def get_embedding(
data,
use_gpu=False
)
```
用于获取输入文本的句子粒度特征与字粒度特征
**参数**
* `data`:输入文本列表,格式为\[\[sample\_a\_text\_a, sample\_a\_text\_b\], \[sample\_b\_text\_a, sample\_b\_text\_b\],…,\],其中每个元素都是一个样例,每个样例可以包含text\_a与text\_b。
* `use_gpu`:是否使用gpu,默认为False。对于GPU用户,建议开启use_gpu。
**返回**
* `results`:list类型,格式为\[\[sample\_a\_pooled\_feature, sample\_a\_seq\_feature\], \[sample\_b\_pooled\_feature, sample\_b\_seq\_feature\],…,\],其中每个元素都是对应样例的特征输出,每个样例都有句子粒度特征pooled\_feature与字粒度特征seq\_feature。
**代码示例**
```python
import paddlehub as hub
data = [
['这个宾馆比较陈旧了,特价的房间也很一般。总体来说一般'],
['怀着十分激动的心情放映,可是看着看着发现,在放映完毕后,出现一集米老鼠的动画片'],
['作为老的四星酒店,房间依然很整洁,相当不错。机场接机服务很好,可以在车上办理入住手续,节省时间。'],
]
label_map = {0: 'negative', 1: 'positive'}
model = hub.Module(
name='electra-small',
version='1.0.0',
task='seq-cls',
load_checkpoint='/path/to/parameters',
label_map=label_map)
results = model.predict(data, max_seq_len=50, batch_size=1, use_gpu=False)
for idx, text in enumerate(data):
print('Data: {} \t Lable: {}'.format(text, results[idx]))
```
详情可参考PaddleHub示例:
- [文本分类](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/text_classification)
- [序列标注](https://github.com/PaddlePaddle/PaddleHub/tree/release/v2.0.0-beta/demo/sequence_labeling)
## 服务部署
PaddleHub Serving可以部署一个在线获取预训练词向量。
### Step1: 启动PaddleHub Serving
运行启动命令:
```shell
$ hub serving start -m electra-small
```
这样就完成了一个获取预训练词向量服务化API的部署,默认端口号为8866。
**NOTE:** 如使用GPU预测,则需要在启动服务之前,请设置CUDA_VISIBLE_DEVICES环境变量,否则不用设置。
### Step2: 发送预测请求
配置好服务端,以下数行代码即可实现发送预测请求,获取预测结果
```python
import requests
import json
# 指定用于获取embedding的文本[[text_1], [text_2], ... ]}
text = [["今天是个好日子"], ["天气预报说今天要下雨"]]
# 以key的方式指定text传入预测方法的时的参数,此例中为"data"
# 对应本地部署,则为module.get_embedding(data=text)
data = {"data": text}
# 发送post请求,content-type类型应指定json方式,url中的ip地址需改为对应机器的ip
url = "http://10.12.121.132:8866/predict/electra-small"
# 指定post请求的headers为application/json方式
headers = {"Content-Type": "application/json"}
r = requests.post(url=url, headers=headers, data=json.dumps(data))
print(r.json())
```
## 查看代码
https://github.com/google-research/electra
## 依赖
paddlepaddle >= 2.0.0
paddlehub >= 2.0.0
## 更新历史
* 1.0.0
初始发布,动态图版本模型,支持文本分类`seq-cls`和序列标注`token-cls`任务的fine-tune
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import Dict
import os
import paddle
import paddle.nn as nn
import paddle.nn.functional as F
from paddlenlp.transformers.electra.modeling import ElectraForSequenceClassification, ElectraForTokenClassification, ElectraModel
from paddlenlp.transformers.electra.tokenizer import ElectraTokenizer
from paddlenlp.metrics import ChunkEvaluator
from paddlehub.module.module import moduleinfo
from paddlehub.module.nlp_module import TransformerModule
from paddlehub.utils.log import logger
@moduleinfo(
name="electra-small",
version="1.0.0",
summary=
"electra-small, 12-layer, 256-hidden, 4-heads, 14M parameters. The module is executed as paddle.dygraph.",
author="paddlepaddle",
author_email="",
type="nlp/semantic_model",
meta=TransformerModule,
)
class Electra(nn.Layer):
"""
Electra model
"""
def __init__(
self,
task: str = None,
load_checkpoint: str = None,
label_map: Dict = None,
num_classes: int = 2,
**kwargs,
):
super(Electra, self).__init__()
if label_map:
self.label_map = label_map
self.num_classes = len(label_map)
else:
self.num_classes = num_classes
if task == 'sequence_classification':
task = 'seq-cls'
logger.warning(
"current task name 'sequence_classification' was renamed to 'seq-cls', "
"'sequence_classification' has been deprecated and will be removed in the future.",
)
if task == 'seq-cls':
self.model = ElectraForSequenceClassification.from_pretrained(
pretrained_model_name_or_path='electra-small',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = paddle.metric.Accuracy()
elif task == 'token-cls':
self.model = ElectraForTokenClassification.from_pretrained(
pretrained_model_name_or_path='electra-small',
num_classes=self.num_classes,
**kwargs
)
self.criterion = paddle.nn.loss.CrossEntropyLoss()
self.metric = ChunkEvaluator(
label_list=[self.label_map[i] for i in sorted(self.label_map.keys())]
)
elif task is None:
self.model = ElectraModel.from_pretrained(pretrained_model_name_or_path='electra-small', **kwargs)
else:
raise RuntimeError("Unknown task {}, task should be one in {}".format(
task, self._tasks_supported))
self.task = task
if load_checkpoint is not None and os.path.isfile(load_checkpoint):
state_dict = paddle.load(load_checkpoint)
self.set_state_dict(state_dict)
logger.info('Loaded parameters from %s' % os.path.abspath(load_checkpoint))
def forward(self, input_ids, token_type_ids=None, position_ids=None, attention_mask=None, seq_lengths=None, labels=None):
result = self.model(input_ids, token_type_ids, position_ids, attention_mask)
if self.task == 'seq-cls':
logits = result
probs = F.softmax(logits, axis=1)
if labels is not None:
loss = self.criterion(logits, labels)
correct = self.metric.compute(probs, labels)
acc = self.metric.update(correct)
return probs, loss, {'acc': acc}
return probs
elif self.task == 'token-cls':
logits = result
token_level_probs = F.softmax(logits, axis=-1)
preds = token_level_probs.argmax(axis=-1)
if labels is not None:
loss = self.criterion(logits, labels.unsqueeze(-1))
num_infer_chunks, num_label_chunks, num_correct_chunks = \
self.metric.compute(None, seq_lengths, preds, labels)
self.metric.update(
num_infer_chunks.numpy(), num_label_chunks.numpy(), num_correct_chunks.numpy())
_, _, f1_score = map(float, self.metric.accumulate())
return token_level_probs, loss, {'f1_score': f1_score}
return token_level_probs
else:
sequence_output, pooled_output = result
return sequence_output, pooled_output
@staticmethod
def get_tokenizer(*args, **kwargs):
"""
Gets the tokenizer that is customized for this module.
"""
return ElectraTokenizer.from_pretrained(
pretrained_model_name_or_path='electra-small', *args, **kwargs)
```shell
$ hub install rtb3==2.0.1
$ hub install rtb3==2.0.0
```
<p align="center">
<img src="https://bj.bcebos.com/paddlehub/paddlehub-img/bert_network.png" hspace='10'/> <br />
......@@ -82,7 +82,7 @@ label_map = {0: 'negative', 1: 'positive'}
model = hub.Module(
name='rtb3',
version='2.0.1',
version='2.0.0',
task='seq-cls',
load_checkpoint='/path/to/parameters',
label_map=label_map)
......@@ -153,6 +153,6 @@ paddlehub >= 2.0.0
初始发布
* 2.0.1
* 2.0.0
全面升级动态图,接口有所变化。任务名称调整,增加序列标注任务`token-cls`
......@@ -29,7 +29,7 @@ from paddlehub.utils.log import logger
@moduleinfo(
name="rbt3",
version="2.0.1",
version="2.0.0",
summary="rbt3, 3-layer, 768-hidden, 12-heads, 38M parameters ",
author="ymcui",
author_email="ymcui@ir.hit.edu.cn",
......
```shell
$ hub install rbtl3==2.0.1
$ hub install rbtl3==2.0.0
```
<p align="center">
<img src="https://bj.bcebos.com/paddlehub/paddlehub-img/bert_network.png" hspace='10'/> <br />
......@@ -82,7 +82,7 @@ label_map = {0: 'negative', 1: 'positive'}
model = hub.Module(
name='rbtl3',
version='2.0.1',
version='2.0.0',
task='seq-cls',
load_checkpoint='/path/to/parameters',
label_map=label_map)
......@@ -153,6 +153,6 @@ paddlehub >= 2.0.0
初始发布
* 2.0.1
* 2.0.0
全面升级动态图,接口有所变化。任务名称调整,增加序列标注任务`token-cls`
......@@ -29,7 +29,7 @@ from paddlehub.utils.log import logger
@moduleinfo(
name="rbtl3",
version="2.0.1",
version="2.0.0",
summary="rbtl3, 3-layer, 1024-hidden, 16-heads, 61M parameters ",
author="ymcui",
author_email="ymcui@ir.hit.edu.cn",
......
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