Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Replace DataProviderWrapperConverter with DataProviderConverter demos #50

Merged
merged 3 commits into from
Sep 9, 2016
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 4 additions & 5 deletions demo/image_classification/prediction.py
Original file line number Diff line number Diff line change
Expand Up @@ -20,9 +20,8 @@

import paddle.utils.image_util as image_util

from py_paddle import swig_paddle, util
from py_paddle import DataProviderWrapperConverter
from paddle.trainer.PyDataProviderWrapper import DenseSlot
from py_paddle import swig_paddle, DataProviderConverter
from paddle.trainer.PyDataProvider2 import dense_vector
from paddle.trainer.config_parser import parse_config

logging.basicConfig(format='[%(levelname)s %(asctime)s %(filename)s:%(lineno)s] %(message)s')
Expand Down Expand Up @@ -75,8 +74,8 @@ def __init__(self,
self.network.loadParameters(self.model_dir)

data_size = 3 * self.crop_dims[0] * self.crop_dims[1]
slots = [DenseSlot(data_size)]
self.converter = util.DataProviderWrapperConverter(False, slots)
slots = [dense_vector(data_size)]
self.converter = DataProviderConverter(slots)

def get_data(self, img_path):
"""
Expand Down
10 changes: 4 additions & 6 deletions demo/model_zoo/resnet/classify.py
Original file line number Diff line number Diff line change
Expand Up @@ -22,9 +22,8 @@

import paddle.utils.image_util as image_util

from py_paddle import swig_paddle, util
from py_paddle import DataProviderWrapperConverter
from paddle.trainer.PyDataProviderWrapper import DenseSlot
from py_paddle import swig_paddle, DataProviderConverter
from paddle.trainer.PyDataProvider2 import dense_vector
from paddle.trainer.config_parser import parse_config

logging.basicConfig(format='[%(levelname)s %(asctime)s %(filename)s:%(lineno)s] %(message)s')
Expand Down Expand Up @@ -85,9 +84,8 @@ def __init__(self, train_conf, model_dir=None,
self.network.loadParameters(self.model_dir)

data_size = 3 * self.crop_dims[0] * self.crop_dims[1]
slots = [DenseSlot(data_size)]
is_sequence = False
self.converter = util.DataProviderWrapperConverter(is_sequence, slots)
slots = [dense_vector(data_size)]
self.converter = DataProviderConverter(slots)

def get_data(self, img_path):
"""
Expand Down
22 changes: 14 additions & 8 deletions demo/semantic_role_labeling/predict.py
Original file line number Diff line number Diff line change
Expand Up @@ -15,12 +15,12 @@
import os
import numpy as np
from optparse import OptionParser
from py_paddle import swig_paddle, util, DataProviderWrapperConverter
from paddle.trainer.PyDataProviderWrapper import IndexSlot
from py_paddle import swig_paddle, DataProviderConverter
from paddle.trainer.PyDataProvider2 import integer_value_sequence
from paddle.trainer.config_parser import parse_config
"""
Usage: run following command to show help message.
python predict.py -h
python predict.py -h
"""
UNK_IDX = 0

Expand All @@ -43,16 +43,22 @@ def __init__(self, train_conf, dict_file, model_dir, label_file):

conf = parse_config(
train_conf,
'dict_len=' + str(len_dict) +
'dict_len=' + str(len_dict) +
',label_len=' + str(len_label) +
',is_predict=True')
self.network = swig_paddle.GradientMachine.createFromConfigProto(
conf.model_config)
self.network.loadParameters(model_dir)

slots = [IndexSlot(len_dict), IndexSlot(len_dict), IndexSlot(len_dict),
IndexSlot(len_dict), IndexSlot(len_dict), IndexSlot(2)]
self.converter = util.DataProviderWrapperConverter(True, slots)
slots = [
integer_value_sequence(len_dict),
integer_value_sequence(len_dict),
integer_value_sequence(len_dict),
integer_value_sequence(len_dict),
integer_value_sequence(len_dict),
integer_value_sequence(2)
]
self.converter = DataProviderConverter(slots)

def load_dict_label(self, dict_file, label_file):
"""
Expand Down Expand Up @@ -109,7 +115,7 @@ def predict(self, data_file):


def option_parser():
usage = ("python predict.py -c config -w model_dir "
usage = ("python predict.py -c config -w model_dir "
"-d word dictionary -l label_file -i input_file")
parser = OptionParser(usage="usage: %s [options]" % usage)
parser.add_option(
Expand Down
10 changes: 5 additions & 5 deletions demo/sentiment/predict.py
Original file line number Diff line number Diff line change
Expand Up @@ -15,13 +15,13 @@
import os
import numpy as np
from optparse import OptionParser
from py_paddle import swig_paddle, util, DataProviderWrapperConverter
from paddle.trainer.PyDataProviderWrapper import IndexSlot
from py_paddle import swig_paddle, DataProviderConverter
from paddle.trainer.PyDataProvider2 import integer_value_sequence
from paddle.trainer.config_parser import parse_config

"""
Usage: run following command to show help message.
python predict.py -h
python predict.py -h
"""

class SentimentPrediction():
Expand All @@ -46,8 +46,8 @@ def __init__(self, train_conf, dict_file, model_dir=None, label_file = None):
conf = parse_config(train_conf, "is_predict=1")
self.network = swig_paddle.GradientMachine.createFromConfigProto(conf.model_config)
self.network.loadParameters(self.model_dir)
slots = [IndexSlot(self.dict_dim)]
self.converter = util.DataProviderWrapperConverter(True, slots)
slots = [integer_value_sequence(self.dict_dim)]
self.converter = DataProviderConverter(slots)

def load_dict(self):
"""
Expand Down
34 changes: 21 additions & 13 deletions doc_cn/ui/predict/swig_py_paddle.rst
Original file line number Diff line number Diff line change
Expand Up @@ -9,22 +9,30 @@ PaddlePaddle目前使用Swig对其常用的预测接口进行了封装,使在P
* 准备数据
* 预测

典型的预测代码如下,使用mnist手写识别作为样例。
典型的预测代码如下,使用mnist手写识别作为样例, 完整代码见
:code:`src_root/doc/ui/predict/predict_sample.py` 。

.. literalinclude:: ../../../doc/ui/predict/predict_sample.py
:language: python
:linenos:

主要的软件包为py_paddle.swig_paddle,这个软件包文档相对完善。可以使用python的 :code:`help()` 函数查询文档。主要步骤为:

* 在程序开始阶段,使用命令行参数初始化PaddlePaddle
* 在98行载入PaddlePaddle的训练文件。读取config
* 在100行创建神经网络,并在83行载入参数。
* 103行创建一个从工具类,用来转换数据。
:lines: 15-18,90-100,101-104

主要的软件包为py_paddle.swig_paddle,这个软件包文档相对完善。可以使用python的
:code:`help()` 函数查询文档。主要步骤为:

* 在程序开始阶段,使用 :code:`swig_paddle.initPaddle()` 传入命令行参数初始化
PaddlePaddle。详细的命令行参数请参考
`命令行参数 <../cmd_argument/detail_introduction.html>`_ 。
* 接下来使用 :code:`parse_config()` 解析训练时的配置文件。这里要注意预测数据通常
不包含label, 而且预测网络通常直接输出最后一层的结果而不是像训练时一样以cost
layer作为输出,所以用于预测的配置文件要做相应的修改。
* 使用 :code:`swig_paddle.GradientMachine.createFromConfigproto()` 根据上一步解
析好的配置创建神经网络。
* 创建一个 :code:`DataProviderConverter` 对象converter。
- swig_paddle接受的原始数据是C++的Matrix,也就是直接写内存的float数组。
- 这个接口并不用户友好。所以,我们提供了一个工具类DataProviderWrapperConverter.
- 这个工具类接收和PyDataProviderWrapper一样的输入数据,请参考PyDataProviderWrapper的文档。
* 在第105行执行预测。forwardTest是一个工具类,直接提取出神经网络Output层的输出结果。典型的输出结果为\:
这个接口并不用户友好。所以,我们提供了一个工具类DataProviderConverter。
这个工具类接收和PyDataProvider2一样的输入数据,详情请参考
`PyDataProvider2文档 <../../../doc/ui/data_provider/pydataprovider2.html>`_ 。
* 最后使用 :code:`forwardTest()` 直接提取出神经网络Output层的输出结果。典型的输出结果为\:

.. code-block:: text

Expand All @@ -37,4 +45,4 @@ PaddlePaddle目前使用Swig对其常用的预测接口进行了封装,使在P
2.70634608e-08, 3.48565123e-08, 5.25639710e-09,
4.48684503e-08]], dtype=float32)}]

其中,value即为softmax层的输出。由于数据是两个,所以输出的value
其中,value即为softmax层的输出。由于数据是两条,所以输出的value包含两个向量