Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into multiclass_nms_op

emailweixu-patch-1
dangqingqing 7 years ago
commit 1d9a7e10da

@ -39,7 +39,7 @@ option(WITH_GPU "Compile PaddlePaddle with NVIDIA GPU" ${CUDA_F
option(WITH_AVX "Compile PaddlePaddle with AVX intrinsics" ${AVX_FOUND})
option(WITH_MKL "Compile PaddlePaddle with MKL support." ${AVX_FOUND})
option(WITH_DSO "Compile PaddlePaddle with dynamic linked CUDA" ON)
option(WITH_TESTING "Compile PaddlePaddle with unit testing" ON)
option(WITH_TESTING "Compile PaddlePaddle with unit testing" OFF)
option(WITH_SWIG_PY "Compile PaddlePaddle with inference api" ON)
option(WITH_STYLE_CHECK "Compile PaddlePaddle with style check" ON)
option(WITH_PYTHON "Compile PaddlePaddle with python interpreter" ON)

@ -186,6 +186,11 @@ function(cc_library TARGET_NAME)
add_library(${TARGET_NAME} STATIC ${cc_library_SRCS})
endif()
if (cc_library_DEPS)
# Don't need link libwarpctc.so
if ("${cc_library_DEPS};" MATCHES "warpctc;")
list(REMOVE_ITEM cc_library_DEPS warpctc)
add_dependencies(${TARGET_NAME} warpctc)
endif()
add_dependencies(${TARGET_NAME} ${cc_library_DEPS})
target_link_libraries(${TARGET_NAME} ${cc_library_DEPS})
endif()
@ -224,12 +229,18 @@ function(cc_test TARGET_NAME)
if(WITH_TESTING)
set(options "")
set(oneValueArgs "")
set(multiValueArgs SRCS DEPS)
set(multiValueArgs SRCS DEPS ARGS)
cmake_parse_arguments(cc_test "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
add_executable(${TARGET_NAME} ${cc_test_SRCS})
target_link_libraries(${TARGET_NAME} ${cc_test_DEPS} paddle_gtest_main paddle_memory gtest gflags)
# Support linking flags: --whole-archive (Linux) / -force_load (MacOS)
target_circle_link_libraries(${TARGET_NAME} ${cc_test_DEPS} paddle_gtest_main paddle_memory gtest gflags)
if("${cc_test_DEPS}" MATCHES "ARCHIVE_START")
list(REMOVE_ITEM cc_test_DEPS ARCHIVE_START ARCHIVE_END)
endif()
add_dependencies(${TARGET_NAME} ${cc_test_DEPS} paddle_gtest_main paddle_memory gtest gflags)
add_test(NAME ${TARGET_NAME} COMMAND ${TARGET_NAME} WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR})
add_test(NAME ${TARGET_NAME}
COMMAND ${TARGET_NAME} ${cc_test_ARGS}
WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR})
endif()
endfunction(cc_test)
@ -457,12 +468,12 @@ endfunction()
function(py_test TARGET_NAME)
if(WITH_TESTING)
set(options STATIC static SHARED shared)
set(options "")
set(oneValueArgs "")
set(multiValueArgs SRCS DEPS ARGS)
set(multiValueArgs SRCS DEPS ARGS ENVS)
cmake_parse_arguments(py_test "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
add_test(NAME ${TARGET_NAME}
COMMAND env PYTHONPATH=${PADDLE_PYTHON_BUILD_DIR}/lib-python
COMMAND env PYTHONPATH=${PADDLE_PYTHON_BUILD_DIR}/lib-python ${py_test_ENVS}
${PYTHON_EXECUTABLE} -u ${py_test_SRCS} ${py_test_ARGS}
WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR})
endif()

@ -1,9 +1,14 @@
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
===========
DataFeeder
data_feeder
===========
DataFeeder
-----------
.. automodule:: paddle.v2.fluid.data_feeder
:members: DataFeeder
----------
.. autoclass:: paddle.v2.fluid.data_feeder.DataFeeder
:members:
:noindex:

@ -1,9 +1,21 @@
===========
Evaluator
===========
Evaluator
-----------
.. automodule:: paddle.v2.fluid.evaluator
:members: Evaluator
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
=========
evaluator
=========
Accuracy
--------
.. autoclass:: paddle.v2.fluid.evaluator.Accuracy
:members:
:noindex:
ChunkEvaluator
--------------
.. autoclass:: paddle.v2.fluid.evaluator.ChunkEvaluator
:members:
:noindex:

@ -1,9 +1,32 @@
===========
Executor
===========
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
========
executor
========
Executor
--------
.. autoclass:: paddle.v2.fluid.executor.Executor
:members:
:noindex:
global_scope
------------
.. autofunction:: paddle.v2.fluid.executor.global_scope
:noindex:
scope_guard
-----------
.. automodule:: paddle.v2.fluid.executor
:members: Executor
.. autofunction:: paddle.v2.fluid.executor.scope_guard
:noindex:
switch_scope
------------
.. autofunction:: paddle.v2.fluid.executor.switch_scope
:noindex:

@ -0,0 +1,109 @@
# Copyright (c) 2018 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 __future__ import print_function
import argparse
import sys
import types
import paddle.v2.fluid as fluid
def parse_arg():
parser = argparse.ArgumentParser()
parser.add_argument('--submodules', nargs="*")
parser.add_argument(
'module', type=str, help='Generate the documentation of which module')
return parser.parse_args()
class DocGenerator(object):
def __init__(self, module_name, stream=sys.stdout):
self.stream = stream
self.module_name = module_name
if not hasattr(fluid, module_name):
raise ValueError("Cannot find fluid.{0}".format(module_name))
else:
self.module = getattr(fluid, module_name)
self.stream.write('''.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
''')
self._print_header_(module_name, dot='=', is_title=True)
def print_submodule(self, submodule_name):
submodule = getattr(self.module, submodule_name)
if submodule is None:
raise ValueError("Cannot find submodule {0}".format(submodule_name))
self.print_section(submodule_name)
for item in submodule.__all__:
self.print_item(item)
def print_current_module(self):
for item in self.module.__all__:
self.print_item(item)
def print_section(self, name):
self._print_header_(name, dot='=', is_title=False)
def print_item(self, name):
item = getattr(self.module, name)
if isinstance(item, types.TypeType):
self.print_class(name)
elif isinstance(item, types.FunctionType):
self.print_method(name)
else:
raise RuntimeError("Unsupported item {0}".format(name))
def print_class(self, name):
self._print_header_(name, dot='-', is_title=False)
self.stream.write('''.. autoclass:: paddle.v2.fluid.{0}.{1}
:members:
:noindex:
'''.format(self.module_name, name))
def print_method(self, name):
self._print_header_(name, dot='-', is_title=False)
self.stream.write('''.. autofunction:: paddle.v2.fluid.{0}.{1}
:noindex:
'''.format(self.module_name, name))
def _print_header_(self, name, dot, is_title):
dot_line = dot * len(name)
if is_title:
self.stream.write(dot_line)
self.stream.write('\n')
self.stream.write(name)
self.stream.write('\n')
self.stream.write(dot_line)
self.stream.write('\n')
self.stream.write('\n')
def main():
args = parse_arg()
gen = DocGenerator(args.module)
if args.submodules is None:
gen.print_current_module()
else:
for submodule_name in args.submodules:
gen.print_submodule(submodule_name)
if __name__ == '__main__':
main()

@ -0,0 +1,7 @@
#!/bin/bash
python gen_doc.py layers --submodules control_flow device io nn ops tensor > layers.rst
for module in io data_feeder evaluator executor initializer io nets optimizer param_attr profiler regularizer
do
python gen_doc.py ${module} > ${module}.rst
done

@ -1,50 +1,35 @@
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
===========
Initializer
initializer
===========
Constant
--------
Initializer
-----------
.. automodule:: paddle.v2.fluid.initializer
:members: Initializer
:noindex:
ConstantInitializer
-------------------
.. automodule:: paddle.v2.fluid.initializer
:members: ConstantInitializer
.. autoclass:: paddle.v2.fluid.initializer.Constant
:members:
:noindex:
Uniform
-------
UniformInitializer
------------------
.. automodule:: paddle.v2.fluid.initializer
:members: UniformInitializer
:noindex:
NormalInitializer
-----------------
.. automodule:: paddle.v2.fluid.initializer
:members: NormalInitializer
.. autoclass:: paddle.v2.fluid.initializer.Uniform
:members:
:noindex:
Normal
------
XavierInitializer
-----------------
.. automodule:: paddle.v2.fluid.initializer
:members: XavierInitializer
.. autoclass:: paddle.v2.fluid.initializer.Normal
:members:
:noindex:
Xavier
------
MSRAInitializer
---------------
.. automodule:: paddle.v2.fluid.initializer
:members: MSRAInitializer
.. autoclass:: paddle.v2.fluid.initializer.Xavier
:members:
:noindex:

@ -1,10 +1,61 @@
===========
IO
===========
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
==
io
==
save_vars
---------
is_parameter
.. autofunction:: paddle.v2.fluid.io.save_vars
:noindex:
save_params
-----------
.. autofunction:: paddle.v2.fluid.io.is_parameter
.. autofunction:: paddle.v2.fluid.io.save_params
:noindex:
save_persistables
-----------------
.. autofunction:: paddle.v2.fluid.io.save_persistables
:noindex:
load_vars
---------
.. autofunction:: paddle.v2.fluid.io.load_vars
:noindex:
load_params
-----------
.. autofunction:: paddle.v2.fluid.io.load_params
:noindex:
load_persistables
-----------------
.. autofunction:: paddle.v2.fluid.io.load_persistables
:noindex:
save_inference_model
--------------------
.. autofunction:: paddle.v2.fluid.io.save_inference_model
:noindex:
load_inference_model
--------------------
.. autofunction:: paddle.v2.fluid.io.load_inference_model
:noindex:
get_inference_program
---------------------
.. autofunction:: paddle.v2.fluid.io.get_inference_program
:noindex:

File diff suppressed because it is too large Load Diff

@ -1,33 +1,31 @@
===========
Nets
===========
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
====
nets
====
simple_img_conv_pool
--------------------
.. autofunction:: paddle.v2.fluid.nets.simple_img_conv_pool
:noindex:
img_conv_group
---------------
.. autofunction:: paddle.v2.fluid.nets.img_conv_group
.. autofunction:: paddle.v2.fluid.nets.simple_img_conv_pool
:noindex:
sequence_conv_pool
------------------
.. autofunction:: paddle.v2.fluid.nets.sequence_conv_pool
:noindex:
glu
---
.. autofunction:: paddle.v2.fluid.nets.glu
:noindex:
scaled_dot_product_attention
----------------------------
.. autofunction:: paddle.v2.fluid.nets.scaled_dot_product_attention
:noindex:

@ -1,54 +1,49 @@
===========
Optimizer
===========
Optimizer
-----------
.. automodule:: paddle.v2.fluid.optimizer
:members: Optimizer
:noindex:
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
=========
optimizer
=========
SGDOptimizer
-----------
.. automodule:: paddle.v2.fluid.optimizer
:members: SGDOptimizer
:noindex:
SGD
---
.. autoclass:: paddle.v2.fluid.optimizer.SGD
:members:
:noindex:
Momentum
--------
MomentumOptimizer
-----------------
.. automodule:: paddle.v2.fluid.optimizer
:members: MomentumOptimizer
.. autoclass:: paddle.v2.fluid.optimizer.Momentum
:members:
:noindex:
Adagrad
-------
AdagradOptimizer
----------------
.. automodule:: paddle.v2.fluid.optimizer
:members: AdagradOptimizer
.. autoclass:: paddle.v2.fluid.optimizer.Adagrad
:members:
:noindex:
Adam
----
AdamOptimizer
-------------
.. automodule:: paddle.v2.fluid.optimizer
:members: AdamOptimizer
.. autoclass:: paddle.v2.fluid.optimizer.Adam
:members:
:noindex:
Adamax
------
AdamaxOptimizer
-----------
.. automodule:: paddle.v2.fluid.optimizer
:members: AdamaxOptimizer
.. autoclass:: paddle.v2.fluid.optimizer.Adamax
:members:
:noindex:
DecayedAdagrad
--------------
DecayedAdagradOptimizer
-----------------------
.. automodule:: paddle.v2.fluid.optimizer
:members: DecayedAdagradOptimizer
.. autoclass:: paddle.v2.fluid.optimizer.DecayedAdagrad
:members:
:noindex:

@ -1,11 +1,21 @@
===========
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
==========
param_attr
==========
ParamAttr
===========
---------
.. autoclass:: paddle.v2.fluid.param_attr.ParamAttr
:members:
:noindex:
WeightNormParamAttr
-------------------
ParamAttr
-----------
.. automodule:: paddle.v2.fluid.param_attr
:members: ParamAttr
.. autoclass:: paddle.v2.fluid.param_attr.WeightNormParamAttr
:members:
:noindex:

@ -1,10 +1,25 @@
===========
Profiler
===========
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
========
profiler
========
cuda_profiler
-------------
Profiler
-----------
.. autofunction:: paddle.v2.fluid.profiler.cuda_profiler
:noindex:
reset_profiler
--------------
.. autofunction:: paddle.v2.fluid.profiler.reset_profiler
:noindex:
profiler
--------
.. autofunction:: paddle.v2.fluid.profiler.profiler
:noindex:

@ -1,25 +1,27 @@
.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
!DO NOT EDIT THIS FILE MANUALLY!
===========
Regularizer
regularizer
===========
WeightDecayRegularizer
----------------------
.. automodule:: paddle.v2.fluid.regularizer
:members: WeightDecayRegularizer
:noindex:
append_regularization_ops
-------------------------
L2DecayRegularizer
------------------
.. automodule:: paddle.v2.fluid.regularizer
:members: L2DecayRegularizer
.. autofunction:: paddle.v2.fluid.regularizer.append_regularization_ops
:noindex:
L1Decay
-------
.. autoclass:: paddle.v2.fluid.regularizer.L1Decay
:members:
:noindex:
L1DecayRegularizer
-------------------
.. automodule:: paddle.v2.fluid.regularizer
:members: L1DecayRegularizer
L2Decay
-------
.. autoclass:: paddle.v2.fluid.regularizer.L2Decay
:members:
:noindex:

@ -140,7 +140,19 @@ TODO by Assignees
### Beam Search with CTC and LM
TODO by Assignees
<div align="center">
<img src="image/beam_search.png" width=600><br/>
Figure 2. Algorithm for CTC Beam Search Decoder.
</div>
- The **Beam Search Decoder** for DS2 CTC-trained network follows the similar approach in \[[3](#references)\] as shown in Figure 2, with two important modifications for the ambiguous parts:
- 1) in the iterative computation of probabilities, the assignment operation is changed to accumulation for one prefix may comes from different paths;
- 2) the if condition ```if l^+ not in A_prev then``` after probabilities' computation is deprecated for it is hard to understand and seems unnecessary.
- An **external scorer** would be passed into the decoder to evaluate a candidate prefix during decoding whenever a white space appended in English decoding and any character appended in Mandarin decoding.
- Such external scorer consists of language model, word count or any other custom scorers.
- The **language model** is built from Task 5, with parameters should be carefully tuned to achieve minimum WER/CER (c.f. Task 7)
- This decoder needs to perform with **high efficiency** for the convenience of parameters tuning and speech recognition in reality.
## Future Work
@ -153,3 +165,4 @@ TODO by Assignees
1. Dario Amodei, etc., [Deep Speech 2 : End-to-End Speech Recognition in English and Mandarin](http://proceedings.mlr.press/v48/amodei16.pdf). ICML 2016.
2. Dario Amodei, etc., [Deep Speech 2 : End-to-End Speech Recognition in English and Mandarin](https://arxiv.org/abs/1512.02595). arXiv:1512.02595.
3. Awni Y. Hannun, etc. [First-Pass Large Vocabulary Continuous Speech Recognition using Bi-Directional Recurrent DNNs](https://arxiv.org/abs/1408.2873). arXiv:1408.2873

Binary file not shown.

After

Width:  |  Height:  |  Size: 464 KiB

@ -115,7 +115,7 @@ PaddlePaddle的编译选项包括生成CPU/GPU二进制文件、链接何种B
"WITH_AVX", "是否编译含有AVX指令集的PaddlePaddle二进制文件", "ON"
"WITH_PYTHON", "是否内嵌PYTHON解释器", "ON"
"WITH_STYLE_CHECK", "是否编译时进行代码风格检查", "ON"
"WITH_TESTING", "是否开启单元测试", "ON"
"WITH_TESTING", "是否开启单元测试", "OFF"
"WITH_DOC", "是否编译中英文文档", "OFF"
"WITH_SWIG_PY", "是否编译PYTHON的SWIG接口该接口可用于预测和定制化训练", "Auto"
"WITH_GOLANG", "是否编译go语言的可容错parameter server", "ON"

@ -126,7 +126,7 @@ You can add :code:`-D` argument to pass such options, like:
"WITH_AVX", "Build with AVX support", "ON"
"WITH_PYTHON", "Build with integrated Python interpreter", "ON"
"WITH_STYLE_CHECK", "Check code style when building", "ON"
"WITH_TESTING", "Build unit tests", "ON"
"WITH_TESTING", "Build unit tests", "OFF"
"WITH_DOC", "Build documentations", "OFF"
"WITH_SWIG_PY", "Build Python SWIG interface for V2 API", "Auto"
"WITH_GOLANG", "Build fault-tolerant parameter server written in go", "ON"

@ -95,6 +95,12 @@ PaddlePaddle Book是为用户和开发者制作的一个交互式的Jupyter Note
docker run -p 8888:8888 paddlepaddle/book
国内用户可以使用下面的镜像源来加速访问:
.. code-block: bash
docker run -p 8888:8888 docker.paddlepaddlehub.com/book
然后在浏览器中输入以下网址:
.. code-block:: text

@ -102,6 +102,12 @@ We provide a packaged book image, simply issue the command:
docker run -p 8888:8888 paddlepaddle/book
For users in China, we provide a faster mirror:
.. code-block: bash
docker run -p 8888:8888 docker.paddlepaddlehub.com/book
Then, you would back and paste the address into the local browser:
.. code-block:: text

@ -92,11 +92,11 @@ paddle.init(
参数说明
- use_gpu **可选默认False**是否启用GPU训练
- trainer_count**必选默认1**,当前训练任务trainer总个数
- trainer_count**必选默认1**,当前trainer的线程数目
- port**必选默认7164**连接到pserver的端口
- ports_num**必选默认1**连接到pserver的端口个数
- ports_num_for_sparse**必选默认0**和pserver之间用于稀疏类型参数通信的端口个数
- num_gradient_servers**必选默认1**,当前训练任务pserver总数
- num_gradient_servers**必选默认1**,当前训练任务trainer总数
- trainer_id**必选默认0**每个trainer的唯一ID从0开始的整数
- pservers**必选默认127.0.0.1**当前训练任务启动的pserver的IP列表多个IP使用“,”隔开

@ -95,11 +95,11 @@ paddle.init(
Parameter Description
- use_gpu: **optional, default False**, set to "True" to enable GPU training.
- trainer_count: **required, default 1**, total count of trainers in the training job.
- trainer_count: **required, default 1**, number of threads in current trainer.
- port: **required, default 7164**, port to connect to parameter server.
- ports_num: **required, default 1**, number of ports for communication.
- ports_num_for_sparse: **required, default 0**, number of ports for sparse type caculation.
- num_gradient_servers: **required, default 1**, total number of gradient server.
- num_gradient_servers: **required, default 1**, number of trainers in current job.
- trainer_id: **required, default 0**, ID for every trainer, start from 0.
- pservers: **required, default 127.0.0.1**, list of IPs of parameter servers, separated by ",".

@ -22,11 +22,11 @@ cc_test(eigen_test SRCS eigen_test.cc DEPS tensor)
cc_library(lod_tensor SRCS lod_tensor.cc DEPS ddim place tensor framework_proto)
cc_test(lod_tensor_test SRCS lod_tensor_test.cc DEPS lod_tensor paddle_memory)
nv_test(lod_tensor_gpu_test SRCS lod_tensor_test.cu DEPS lod_tensor)
nv_test(lod_tensor_gpu_test SRCS lod_tensor_test.cu DEPS lod_tensor init)
cc_test(variable_test SRCS variable_test.cc)
cc_library(threadpool SRCS threadpool.cc)
cc_library(threadpool SRCS threadpool.cc DEPS enforce)
cc_test(threadpool_test SRCS threadpool_test.cc DEPS threadpool)
cc_library(scope SRCS scope.cc DEPS glog threadpool)

@ -26,9 +26,7 @@ class Channel {
virtual void Send(T*) = 0;
virtual void Receive(T*) = 0;
virtual size_t Cap() = 0;
// Don't delete channels; instead, call Channel::Close.
protected:
virtual void Close() = 0;
virtual ~Channel() {}
};
@ -50,11 +48,7 @@ Channel<T>* MakeChannel(size_t buffer_size) {
template <typename T>
void CloseChannel(Channel<T>* ch) {
if (ch->Cap() > 0) {
delete dynamic_cast<details::Buffered<T>*>(ch);
} else {
delete dynamic_cast<details::UnBuffered<T>*>(ch);
}
ch->Close();
}
} // namespace framework

Some files were not shown because too many files have changed in this diff Show More

Loading…
Cancel
Save