Merge branch 'develop' of https://github.com/baidu/Paddle into cmrnorm

.gitignore

@@ -9,3 +9,6 @@ build/
 .pydevproject
 Makefile
 .test_env/
+
+*~
+bazel-*

.pre-commit-config.yaml

@@ -6,7 +6,8 @@
 -   repo: https://github.com/reyoung/mirrors-yapf.git
     sha: v0.13.2
     hooks:
-    -   id: yapf
+    - id: yapf
+      files: (.*\.(py|bzl)|BUILD|.*\.BUILD|WORKSPACE)$  # Bazel BUILD files follow Python syntax.
 -   repo: https://github.com/pre-commit/pre-commit-hooks
     sha: 7539d8bd1a00a3c1bfd34cdb606d3a6372e83469
     hooks:

.travis.yml

@@ -8,10 +8,13 @@ os:
 env:
   - JOB=DOCS
   - JOB=BUILD_AND_TEST
+  - JOB=PRE_COMMIT
 matrix:
   exclude:
     - os: osx
-      env: JOB=DOCS  # Only generate documentation in linux
+      env: JOB=DOCS  # Only generate documentation in linux.
+    - os: osx
+      env: JOB=PRE_COMMIT # Only check pre-commit hook in linux
 
 addons:
   apt:
@@ -39,18 +42,23 @@ addons:
       - lcov
       - graphviz
       - swig
+      - clang-format-3.8
 before_install:
   - |
     if [ ${JOB} == "BUILD_AND_TEST" ]; then
-      if ! git diff --name-only $TRAVIS_COMMIT_RANGE | grep -qvE '(\.md$)|(\.rst$)|(\.jpg$)|(\.png$)'
-      then
-        echo "Only markdown docs were updated, stopping build process."
-        exit
+      local change_list=`git diff --name-only $TRAVIS_COMMIT_RANGE`
+      if [ $? -eq 0 ]; then  # if git diff return no zero, then rerun unit test.
+        if ! echo ${change_list} | grep -qvE '(\.md$)|(\.rst$)|(\.jpg$)|(\.png$)'
+        then
+          echo "Only markdown docs were updated, stopping build process."
+          exit
+        fi
       fi
     fi
   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then sudo paddle/scripts/travis/before_install.linux.sh; fi
   - if [[ "$TRAVIS_OS_NAME" == "osx" ]]; then paddle/scripts/travis/before_install.osx.sh; fi
-  - pip install wheel protobuf sphinx breathe recommonmark virtualenv numpy sphinx_rtd_theme
+  - if [[ "$JOB" == "PRE_COMMIT" ]]; then sudo ln -s /usr/bin/clang-format-3.8 /usr/bin/clang-format; fi
+  - pip install wheel protobuf sphinx recommonmark virtualenv numpy sphinx_rtd_theme pre-commit
 script:
   - paddle/scripts/travis/main.sh
 notifications:

CMakeLists.txt

@@ -11,7 +11,7 @@ find_package(Protobuf REQUIRED)
 
 # Check protobuf library version.
 execute_process(COMMAND ${PROTOBUF_PROTOC_EXECUTABLE} --version
-	OUTPUT_VARIABLE PROTOBUF_VERSION)
+    OUTPUT_VARIABLE PROTOBUF_VERSION)
 string(REPLACE "libprotoc " "" PROTOBUF_VERSION ${PROTOBUF_VERSION})
 
 set(PROTOBUF_3 OFF)
@@ -25,8 +25,8 @@ find_package(ZLIB REQUIRED)
 find_package(NumPy REQUIRED)
 find_package(Threads REQUIRED)
 find_package(AVX QUIET)
-find_package(Glog)
-find_package(Gflags QUIET)
+find_package(Glog REQUIRED)
+find_package(Gflags REQUIRED)
 find_package(GTest)
 find_package(Sphinx)
 find_package(Doxygen)
@@ -40,8 +40,6 @@ option(WITH_AVX "Compile PaddlePaddle with avx intrinsics" ${AVX_FOUND})
 option(WITH_PYTHON "Compile PaddlePaddle with python interpreter" ON)
 option(WITH_STYLE_CHECK "Style Check for PaddlePaddle" ${PYTHONINTERP_FOUND})
 option(WITH_RDMA "Compile PaddlePaddle with rdma support" OFF)
-option(WITH_GLOG "Compile PaddlePaddle use glog, otherwise use a log implement internally" ${LIBGLOG_FOUND})
-option(WITH_GFLAGS "Compile PaddlePaddle use gflags, otherwise use a flag implement internally" ${GFLAGS_FOUND})
 option(WITH_TIMER "Compile PaddlePaddle use timer" OFF)
 option(WITH_PROFILER "Compile PaddlePaddle use gpu profiler" OFF)
 option(WITH_TESTING "Compile and run unittest for PaddlePaddle" ${GTEST_FOUND})
@@ -136,16 +134,12 @@ else(WITH_RDMA)
   add_definitions(-DPADDLE_DISABLE_RDMA)
 endif(WITH_RDMA)
 
-if(WITH_GLOG)
-    add_definitions(-DPADDLE_USE_GLOG)
-    include_directories(${LIBGLOG_INCLUDE_DIR})
-endif()
+# glog
+include_directories(${LIBGLOG_INCLUDE_DIR})
 
-if(WITH_GFLAGS)
-    add_definitions(-DPADDLE_USE_GFLAGS)
-    add_definitions(-DGFLAGS_NS=${GFLAGS_NAMESPACE})
-    include_directories(${GFLAGS_INCLUDE_DIRS})
-endif()
+#gflags
+add_definitions(-DGFLAGS_NS=${GFLAGS_NAMESPACE})
+include_directories(${GFLAGS_INCLUDE_DIRS})
 
 if(WITH_TESTING)
     enable_testing()
@@ -169,5 +163,4 @@ add_subdirectory(paddle)
 add_subdirectory(python)
 if(WITH_DOC)
     add_subdirectory(doc)
-    add_subdirectory(doc_cn)
 endif()

CONTRIBUTING.md

+./doc/howto/contribute_to_paddle_en.md
\ No newline at end of file

WORKSPACE

+# External dependency to Google protobuf.
+http_archive(
+    name="protobuf",
+    url="http://github.com/google/protobuf/archive/v3.1.0.tar.gz",
+    sha256="0a0ae63cbffc274efb573bdde9a253e3f32e458c41261df51c5dbc5ad541e8f7",
+    strip_prefix="protobuf-3.1.0")
+
+# External dependency to gtest 1.7.0.  This method comes from
+# https://www.bazel.io/versions/master/docs/tutorial/cpp.html.
+new_http_archive(
+    name="gtest",
+    url="https://github.com/google/googletest/archive/release-1.7.0.zip",
+    sha256="b58cb7547a28b2c718d1e38aee18a3659c9e3ff52440297e965f5edffe34b6d0",
+    build_file="third_party/gtest.BUILD",
+    strip_prefix="googletest-release-1.7.0")
+
+# External dependency to gflags.  This method comes from
+# https://github.com/gflags/example/blob/master/WORKSPACE.
+new_git_repository(
+    name="gflags",
+    tag="v2.2.0",
+    remote="https://github.com/gflags/gflags.git",
+    build_file="third_party/gflags.BUILD")
+
+# External dependency to glog.  This method comes from
+# https://github.com/reyoung/bazel_playground/blob/master/WORKSPACE
+new_git_repository(
+    name="glog",
+    remote="https://github.com/google/glog.git",
+    commit="b6a5e0524c28178985f0d228e9eaa43808dbec3c",
+    build_file="third_party/glog.BUILD")

benchmark/tensorflow/rnn/run_multi.sh

@@ -25,4 +25,3 @@ test 4 2 256 512
 test 4 2 512 128 
 test 4 2 512 256 
 test 4 2 512 512 
-

cmake/FindSphinx.cmake

@@ -72,6 +72,7 @@ function( Sphinx_add_target target_name builder conf cache source destination )
     ${source}
     ${destination}
     COMMENT "Generating sphinx documentation: ${builder}"
+    COMMAND ln -s ${destination}/index_*.html ${destination}/index.html
     )
 
   set_property(
@@ -143,4 +144,4 @@ function( Sphinx_add_targets target_base_name conf source base_destination )
 
     add_dependencies( ${target_base_name}_linkcheck ${_dependencies} )
   endif()
-endfunction()
\ No newline at end of file
+endfunction()

cmake/check_packages.cmake

@@ -14,13 +14,9 @@ if(WITH_STYLE_CHECK)
   find_package(PythonInterp REQUIRED)
 endif()
 
-if(WITH_GLOG)
-  find_package(Glog REQUIRED)
-endif()
+find_package(Glog REQUIRED)
 
-if(WITH_GFLAGS)
-  find_package(Gflags REQUIRED)
-endif()
+find_package(Gflags REQUIRED)
 
 if(WITH_TESTING)
   find_package(GTest REQUIRED)
@@ -30,7 +26,6 @@ if(WITH_DOC)
   find_package(Sphinx REQUIRED)
   find_package(Doxygen REQUIRED)
   find_python_module(recommonmark REQUIRED)
-  find_python_module(breathe REQUIRED)
 endif()
 
 if(WITH_SWIG_PY)

cmake/util.cmake

@@ -65,7 +65,7 @@ endmacro()
 # link_paddle_exe
 # add paddle library for a paddle executable, such as trainer, pserver.
 #
-# It will handle WITH_PYTHON/WITH_GLOG etc.
+# It will handle WITH_PYTHON etc.
 function(link_paddle_exe TARGET_NAME)
     if(WITH_RDMA)
         generate_rdma_links()
@@ -108,6 +108,8 @@ function(link_paddle_exe TARGET_NAME)
         paddle_cuda
         ${METRIC_LIBS}
         ${PROTOBUF_LIBRARY}
+        ${LIBGLOG_LIBRARY}
+        ${GFLAGS_LIBRARIES}
         ${CMAKE_THREAD_LIBS_INIT}
         ${CBLAS_LIBS}
         ${ZLIB_LIBRARIES}
@@ -119,27 +121,17 @@ function(link_paddle_exe TARGET_NAME)
             ${RDMA_LD_FLAGS}
             ${RDMA_LIBS})
     endif()
-    
+
     if(WITH_PYTHON)
         target_link_libraries(${TARGET_NAME}
             ${PYTHON_LIBRARIES})
     endif()
 
-    if(WITH_GLOG)
-        target_link_libraries(${TARGET_NAME}
-            ${LIBGLOG_LIBRARY})
-    endif()
-
-    if(WITH_GFLAGS)
-        target_link_libraries(${TARGET_NAME}
-            ${GFLAGS_LIBRARIES})
-    endif()
-
     if(WITH_GPU)
-        if(NOT WITH_DSO OR WITH_METRIC) 
+        if(NOT WITH_DSO OR WITH_METRIC)
             target_link_libraries(${TARGET_NAME}
                 ${CUDNN_LIBRARY}
-                ${CUDA_curand_LIBRARY}) 
+                ${CUDA_curand_LIBRARY})
             CUDA_ADD_CUBLAS_TO_TARGET(${TARGET_NAME})
         endif()
 
@@ -206,5 +198,5 @@ function(create_resources res_file output)
     # Convert hex data for C compatibility
     string(REGEX REPLACE "([0-9a-f][0-9a-f])" "0x\\1," filedata ${filedata})
     # Append data to output file
-    file(APPEND ${output} "const unsigned char ${filename}[] = {${filedata}};\nconst unsigned ${filename}_size = sizeof(${filename});\n")
+    file(APPEND ${output} "const unsigned char ${filename}[] = {${filedata}0};\nconst unsigned ${filename}_size = sizeof(${filename});\n")
 endfunction()

demo/gan/README.md

@@ -10,4 +10,4 @@ Then you can run the command below. The flag -d specifies the training data (cif
 $python gan_trainer.py -d cifar --use_gpu 1
 
 The generated images will be stored in ./cifar_samples/
-The corresponding models will be stored in ./cifar_params/
\ No newline at end of file
+The corresponding models will be stored in ./cifar_params/

demo/gan/data/download_cifar.sh

@@ -15,4 +15,3 @@ set -e
 wget https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz
 tar zxf cifar-10-python.tar.gz
 rm cifar-10-python.tar.gz
-

demo/gan/data/get_mnist_data.sh

@@ -15,5 +15,3 @@ do
         gunzip ${fname}.gz
     fi
 done
-
-

demo/gan/gan_conf.py

@@ -14,10 +14,9 @@
 from paddle.trainer_config_helpers import *
 
 mode = get_config_arg("mode", str, "generator")
-assert mode in set(["generator",
-                    "discriminator",
-                    "generator_training",
-                    "discriminator_training"])
+assert mode in set([
+    "generator", "discriminator", "generator_training", "discriminator_training"
+])
 
 is_generator_training = mode == "generator_training"
 is_discriminator_training = mode == "discriminator_training"
@@ -38,8 +37,8 @@ sample_dim = 2
 settings(
     batch_size=128,
     learning_rate=1e-4,
-    learning_method=AdamOptimizer(beta1=0.5)
-)
+    learning_method=AdamOptimizer(beta1=0.5))
+
 
 def discriminator(sample):
     """
@@ -50,70 +49,87 @@ def discriminator(sample):
     of the sample is from real data.
     """
     param_attr = ParamAttr(is_static=is_generator_training)
-    bias_attr = ParamAttr(is_static=is_generator_training,
-                          initial_mean=1.0,
-                          initial_std=0)
-
-    hidden = fc_layer(input=sample, name="dis_hidden", size=hidden_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=ReluActivation())
-
-    hidden2 = fc_layer(input=hidden, name="dis_hidden2", size=hidden_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=LinearActivation())
-    
-    hidden_bn = batch_norm_layer(hidden2, 
-                     act=ReluActivation(), 
-                     name="dis_hidden_bn", 
-                     bias_attr=bias_attr, 
-                     param_attr=ParamAttr(is_static=is_generator_training,
-                           initial_mean=1.0,
-                           initial_std=0.02),
-                     use_global_stats=False)
-    
-    return fc_layer(input=hidden_bn, name="dis_prob", size=2,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=SoftmaxActivation())
+    bias_attr = ParamAttr(
+        is_static=is_generator_training, initial_mean=1.0, initial_std=0)
+
+    hidden = fc_layer(
+        input=sample,
+        name="dis_hidden",
+        size=hidden_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=ReluActivation())
+
+    hidden2 = fc_layer(
+        input=hidden,
+        name="dis_hidden2",
+        size=hidden_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=LinearActivation())
+
+    hidden_bn = batch_norm_layer(
+        hidden2,
+        act=ReluActivation(),
+        name="dis_hidden_bn",
+        bias_attr=bias_attr,
+        param_attr=ParamAttr(
+            is_static=is_generator_training, initial_mean=1.0,
+            initial_std=0.02),
+        use_global_stats=False)
+
+    return fc_layer(
+        input=hidden_bn,
+        name="dis_prob",
+        size=2,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=SoftmaxActivation())
+
 
 def generator(noise):
     """
     generator generates a sample given noise
     """
     param_attr = ParamAttr(is_static=is_discriminator_training)
-    bias_attr = ParamAttr(is_static=is_discriminator_training,
-                           initial_mean=1.0,
-                           initial_std=0)
-    
-    hidden = fc_layer(input=noise,
-                    name="gen_layer_hidden",
-                    size=hidden_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=ReluActivation())
-
-    hidden2 = fc_layer(input=hidden, name="gen_hidden2", size=hidden_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=LinearActivation())
-    
-    hidden_bn = batch_norm_layer(hidden2, 
-                     act=ReluActivation(), 
-                     name="gen_layer_hidden_bn", 
-                     bias_attr=bias_attr, 
-                     param_attr=ParamAttr(is_static=is_discriminator_training,
-                           initial_mean=1.0,
-                           initial_std=0.02),
-                     use_global_stats=False)
-    
-    return fc_layer(input=hidden_bn,
-                    name="gen_layer1",
-                    size=sample_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=LinearActivation())
+    bias_attr = ParamAttr(
+        is_static=is_discriminator_training, initial_mean=1.0, initial_std=0)
+
+    hidden = fc_layer(
+        input=noise,
+        name="gen_layer_hidden",
+        size=hidden_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=ReluActivation())
+
+    hidden2 = fc_layer(
+        input=hidden,
+        name="gen_hidden2",
+        size=hidden_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=LinearActivation())
+
+    hidden_bn = batch_norm_layer(
+        hidden2,
+        act=ReluActivation(),
+        name="gen_layer_hidden_bn",
+        bias_attr=bias_attr,
+        param_attr=ParamAttr(
+            is_static=is_discriminator_training,
+            initial_mean=1.0,
+            initial_std=0.02),
+        use_global_stats=False)
+
+    return fc_layer(
+        input=hidden_bn,
+        name="gen_layer1",
+        size=sample_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=LinearActivation())
+
 
 if is_generator_training:
     noise = data_layer(name="noise", size=noise_dim)
@@ -126,7 +142,8 @@ if is_generator_training or is_discriminator_training:
     label = data_layer(name="label", size=1)
     prob = discriminator(sample)
     cost = cross_entropy(input=prob, label=label)
-    classification_error_evaluator(input=prob, label=label, name=mode+'_error')
+    classification_error_evaluator(
+        input=prob, label=label, name=mode + '_error')
     outputs(cost)
 
 if is_generator:

demo/gan/gan_conf_image.py

@@ -15,10 +15,9 @@ from paddle.trainer_config_helpers import *
 
 mode = get_config_arg("mode", str, "generator")
 dataSource = get_config_arg("data", str, "mnist")
-assert mode in set(["generator",
-                    "discriminator",
-                    "generator_training",
-                    "discriminator_training"])
+assert mode in set([
+    "generator", "discriminator", "generator_training", "discriminator_training"
+])
 
 is_generator_training = mode == "generator_training"
 is_discriminator_training = mode == "discriminator_training"
@@ -36,24 +35,33 @@ noise_dim = 100
 gf_dim = 64
 df_dim = 64
 if dataSource == "mnist":
-    sample_dim = 28 # image dim
-    c_dim = 1 # image color
+    sample_dim = 28  # image dim
+    c_dim = 1  # image color
 else:
     sample_dim = 32
     c_dim = 3
-s2, s4 = int(sample_dim/2), int(sample_dim/4), 
-s8, s16 = int(sample_dim/8), int(sample_dim/16)
+s2, s4 = int(sample_dim / 2), int(sample_dim / 4),
+s8, s16 = int(sample_dim / 8), int(sample_dim / 16)
 
 settings(
     batch_size=128,
     learning_rate=2e-4,
-    learning_method=AdamOptimizer(beta1=0.5)
-)
+    learning_method=AdamOptimizer(beta1=0.5))
 
-def conv_bn(input, channels, imgSize, num_filters, output_x, stride, name, 
-                 param_attr, bias_attr, param_attr_bn, bn, trans=False, 
-                 act=ReluActivation()):
-    
+
+def conv_bn(input,
+            channels,
+            imgSize,
+            num_filters,
+            output_x,
+            stride,
+            name,
+            param_attr,
+            bias_attr,
+            param_attr_bn,
+            bn,
+            trans=False,
+            act=ReluActivation()):
     """
     conv_bn is a utility function that constructs a convolution/deconv layer 
     with an optional batch_norm layer
@@ -63,10 +71,10 @@ def conv_bn(input, channels, imgSize, num_filters, output_x, stride, name,
     :param trans: whether to use conv (False) or deconv (True)
     :type trans: bool
     """
-    
+
     # calculate the filter_size and padding size based on the given
     # imgSize and ouput size
-    tmp =  imgSize - (output_x - 1) * stride
+    tmp = imgSize - (output_x - 1) * stride
     if tmp <= 1 or tmp > 5:
         raise ValueError("conv input-output dimension does not fit")
     elif tmp <= 3:
@@ -76,111 +84,134 @@ def conv_bn(input, channels, imgSize, num_filters, output_x, stride, name,
         filter_size = tmp
         padding = 0
 
-    print (imgSize, output_x, stride, filter_size, padding)
-    
+    print(imgSize, output_x, stride, filter_size, padding)
+
     if trans:
         nameApx = "_conv"
     else:
         nameApx = "_convt"
-    
+
     if bn:
-        conv = img_conv_layer(input, filter_size=filter_size, 
-                   num_filters=num_filters,
-                   name=name + nameApx, num_channels=channels,
-                   act=LinearActivation(), groups=1, stride=stride, 
-                   padding=padding, bias_attr=bias_attr,
-                   param_attr=param_attr, shared_biases=True, layer_attr=None,
-                   filter_size_y=None, stride_y=None, padding_y=None, 
-                   trans=trans)
-        
-        conv_bn = batch_norm_layer(conv, 
-                         act=act, 
-                         name=name + nameApx + "_bn", 
-                         bias_attr=bias_attr, 
-                         param_attr=param_attr_bn,
-                         use_global_stats=False)
-        
+        conv = img_conv_layer(
+            input,
+            filter_size=filter_size,
+            num_filters=num_filters,
+            name=name + nameApx,
+            num_channels=channels,
+            act=LinearActivation(),
+            groups=1,
+            stride=stride,
+            padding=padding,
+            bias_attr=bias_attr,
+            param_attr=param_attr,
+            shared_biases=True,
+            layer_attr=None,
+            filter_size_y=None,
+            stride_y=None,
+            padding_y=None,
+            trans=trans)
+
+        conv_bn = batch_norm_layer(
+            conv,
+            act=act,
+            name=name + nameApx + "_bn",
+            bias_attr=bias_attr,
+            param_attr=param_attr_bn,
+            use_global_stats=False)
+
         return conv_bn
     else:
-        conv = img_conv_layer(input, filter_size=filter_size, 
-                   num_filters=num_filters,
-                   name=name + nameApx, num_channels=channels,
-                   act=act, groups=1, stride=stride, 
-                   padding=padding, bias_attr=bias_attr,
-                   param_attr=param_attr, shared_biases=True, layer_attr=None,
-                   filter_size_y=None, stride_y=None, padding_y=None,
-                   trans=trans)
+        conv = img_conv_layer(
+            input,
+            filter_size=filter_size,
+            num_filters=num_filters,
+            name=name + nameApx,
+            num_channels=channels,
+            act=act,
+            groups=1,
+            stride=stride,
+            padding=padding,
+            bias_attr=bias_attr,
+            param_attr=param_attr,
+            shared_biases=True,
+            layer_attr=None,
+            filter_size_y=None,
+            stride_y=None,
+            padding_y=None,
+            trans=trans)
         return conv
-    
+
+
 def generator(noise):
     """
     generator generates a sample given noise
     """
-    param_attr = ParamAttr(is_static=is_discriminator_training,
-                           initial_mean=0.0,
-                           initial_std=0.02)
-    bias_attr = ParamAttr(is_static=is_discriminator_training,
-                           initial_mean=0.0,
-                           initial_std=0.0)
-    
-    param_attr_bn=ParamAttr(is_static=is_discriminator_training,
-                           initial_mean=1.0,
-                           initial_std=0.02)
-    
-    h1 = fc_layer(input=noise,
-                    name="gen_layer_h1",
-                    size=s8 * s8 * gf_dim * 4,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=LinearActivation())
-    
-    h1_bn = batch_norm_layer(h1, 
-                     act=ReluActivation(), 
-                     name="gen_layer_h1_bn", 
-                     bias_attr=bias_attr, 
-                     param_attr=param_attr_bn,
-                     use_global_stats=False)
-    
-    h2_bn = conv_bn(h1_bn, 
-                    channels=gf_dim*4, 
-                    output_x=s8,
-                    num_filters=gf_dim*2, 
-                    imgSize=s4,
-                    stride=2,
-                    name="gen_layer_h2", 
-                    param_attr=param_attr, 
-                    bias_attr=bias_attr, 
-                    param_attr_bn=param_attr_bn,
-                    bn=True,
-                    trans=True)
-    
-    h3_bn = conv_bn(h2_bn, 
-                    channels=gf_dim*2, 
-                    output_x=s4,
-                    num_filters=gf_dim, 
-                    imgSize=s2,
-                    stride=2,
-                    name="gen_layer_h3", 
-                    param_attr=param_attr, 
-                    bias_attr=bias_attr, 
-                    param_attr_bn=param_attr_bn,
-                    bn=True,
-                    trans=True)
-     
-    
-    return conv_bn(h3_bn,
-                   channels=gf_dim, 
-                   output_x=s2,
-                   num_filters=c_dim, 
-                   imgSize=sample_dim,
-                   stride=2,
-                   name="gen_layer_h4", 
-                   param_attr=param_attr, 
-                   bias_attr=bias_attr, 
-                   param_attr_bn=param_attr_bn,
-                   bn=False,
-                   trans=True,
-                   act=TanhActivation())
+    param_attr = ParamAttr(
+        is_static=is_discriminator_training, initial_mean=0.0, initial_std=0.02)
+    bias_attr = ParamAttr(
+        is_static=is_discriminator_training, initial_mean=0.0, initial_std=0.0)
+
+    param_attr_bn = ParamAttr(
+        is_static=is_discriminator_training, initial_mean=1.0, initial_std=0.02)
+
+    h1 = fc_layer(
+        input=noise,
+        name="gen_layer_h1",
+        size=s8 * s8 * gf_dim * 4,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=LinearActivation())
+
+    h1_bn = batch_norm_layer(
+        h1,
+        act=ReluActivation(),
+        name="gen_layer_h1_bn",
+        bias_attr=bias_attr,
+        param_attr=param_attr_bn,
+        use_global_stats=False)
+
+    h2_bn = conv_bn(
+        h1_bn,
+        channels=gf_dim * 4,
+        output_x=s8,
+        num_filters=gf_dim * 2,
+        imgSize=s4,
+        stride=2,
+        name="gen_layer_h2",
+        param_attr=param_attr,
+        bias_attr=bias_attr,
+        param_attr_bn=param_attr_bn,
+        bn=True,
+        trans=True)
+
+    h3_bn = conv_bn(
+        h2_bn,
+        channels=gf_dim * 2,
+        output_x=s4,
+        num_filters=gf_dim,
+        imgSize=s2,
+        stride=2,
+        name="gen_layer_h3",
+        param_attr=param_attr,
+        bias_attr=bias_attr,
+        param_attr_bn=param_attr_bn,
+        bn=True,
+        trans=True)
+
+    return conv_bn(
+        h3_bn,
+        channels=gf_dim,
+        output_x=s2,
+        num_filters=c_dim,
+        imgSize=sample_dim,
+        stride=2,
+        name="gen_layer_h4",
+        param_attr=param_attr,
+        bias_attr=bias_attr,
+        param_attr_bn=param_attr_bn,
+        bn=False,
+        trans=True,
+        act=TanhActivation())
 
 
 def discriminator(sample):
@@ -191,58 +222,60 @@ def discriminator(sample):
     of the sample is from generator and dimension 1 is the probabblity
     of the sample is from real data.
     """
-    param_attr = ParamAttr(is_static=is_generator_training,
-                           initial_mean=0.0,
-                           initial_std=0.02)
-    bias_attr = ParamAttr(is_static=is_generator_training,
-                          initial_mean=0.0,
-                          initial_std=0.0)
-    
-    param_attr_bn=ParamAttr(is_static=is_generator_training,
-                           initial_mean=1.0,
-                           initial_std=0.02)
-    
-    h0 = conv_bn(sample, 
-                 channels=c_dim, 
-                 imgSize=sample_dim,
-                 num_filters=df_dim, 
-                 output_x=s2, 
-                 stride=2, 
-                 name="dis_h0", 
-                 param_attr=param_attr, 
-                 bias_attr=bias_attr, 
-                 param_attr_bn=param_attr_bn, 
-                 bn=False)
-    
-    h1_bn = conv_bn(h0, 
-                 channels=df_dim,
-                 imgSize=s2,
-                 num_filters=df_dim*2, 
-                 output_x=s4, 
-                 stride=2, 
-                 name="dis_h1", 
-                 param_attr=param_attr, 
-                 bias_attr=bias_attr, 
-                 param_attr_bn=param_attr_bn, 
-                 bn=True)
-
-    h2_bn = conv_bn(h1_bn, 
-                 channels=df_dim*2,
-                 imgSize=s4,
-                 num_filters=df_dim*4, 
-                 output_x=s8, 
-                 stride=2, 
-                 name="dis_h2", 
-                 param_attr=param_attr, 
-                 bias_attr=bias_attr, 
-                 param_attr_bn=param_attr_bn, 
-                 bn=True)
-        
-    return fc_layer(input=h2_bn, name="dis_prob", size=2,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=SoftmaxActivation())
+    param_attr = ParamAttr(
+        is_static=is_generator_training, initial_mean=0.0, initial_std=0.02)
+    bias_attr = ParamAttr(
+        is_static=is_generator_training, initial_mean=0.0, initial_std=0.0)
+
+    param_attr_bn = ParamAttr(
+        is_static=is_generator_training, initial_mean=1.0, initial_std=0.02)
+
+    h0 = conv_bn(
+        sample,
+        channels=c_dim,
+        imgSize=sample_dim,
+        num_filters=df_dim,
+        output_x=s2,
+        stride=2,
+        name="dis_h0",
+        param_attr=param_attr,
+        bias_attr=bias_attr,
+        param_attr_bn=param_attr_bn,
+        bn=False)
+
+    h1_bn = conv_bn(
+        h0,
+        channels=df_dim,
+        imgSize=s2,
+        num_filters=df_dim * 2,
+        output_x=s4,
+        stride=2,
+        name="dis_h1",
+        param_attr=param_attr,
+        bias_attr=bias_attr,
+        param_attr_bn=param_attr_bn,
+        bn=True)
+
+    h2_bn = conv_bn(
+        h1_bn,
+        channels=df_dim * 2,
+        imgSize=s4,
+        num_filters=df_dim * 4,
+        output_x=s8,
+        stride=2,
+        name="dis_h2",
+        param_attr=param_attr,
+        bias_attr=bias_attr,
+        param_attr_bn=param_attr_bn,
+        bn=True)
 
+    return fc_layer(
+        input=h2_bn,
+        name="dis_prob",
+        size=2,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=SoftmaxActivation())
 
 
 if is_generator_training:
@@ -250,13 +283,14 @@ if is_generator_training:
     sample = generator(noise)
 
 if is_discriminator_training:
-    sample = data_layer(name="sample", size=sample_dim * sample_dim*c_dim)
+    sample = data_layer(name="sample", size=sample_dim * sample_dim * c_dim)
 
 if is_generator_training or is_discriminator_training:
     label = data_layer(name="label", size=1)
     prob = discriminator(sample)
     cost = cross_entropy(input=prob, label=label)
-    classification_error_evaluator(input=prob, label=label, name=mode+'_error')
+    classification_error_evaluator(
+        input=prob, label=label, name=mode + '_error')
     outputs(cost)
 
 if is_generator:

demo/gan/gan_trainer.py

@@ -16,7 +16,7 @@ import argparse
 import random
 import numpy
 import cPickle
-import sys,os
+import sys, os
 from PIL import Image
 
 from paddle.trainer.config_parser import parse_config
@@ -24,6 +24,7 @@ from paddle.trainer.config_parser import logger
 import py_paddle.swig_paddle as api
 import matplotlib.pyplot as plt
 
+
 def plot2DScatter(data, outputfile):
     '''
     Plot the data as a 2D scatter plot and save to outputfile
@@ -41,9 +42,11 @@ def plot2DScatter(data, outputfile):
     plt.scatter(x, y)
     plt.savefig(outputfile, bbox_inches='tight')
 
+
 def CHECK_EQ(a, b):
     assert a == b, "a=%s, b=%s" % (a, b)
 
+
 def copy_shared_parameters(src, dst):
     '''
     copy the parameters from src to dst
@@ -52,11 +55,9 @@ def copy_shared_parameters(src, dst):
     :param dst: the destination of the parameters
     :type dst: GradientMachine
     '''
-    src_params = [src.getParameter(i)
-               for i in xrange(src.getParameterSize())]
+    src_params = [src.getParameter(i) for i in xrange(src.getParameterSize())]
     src_params = dict([(p.getName(), p) for p in src_params])
 
-
     for i in xrange(dst.getParameterSize()):
         dst_param = dst.getParameter(i)
         src_param = src_params.get(dst_param.getName(), None)
@@ -67,15 +68,17 @@ def copy_shared_parameters(src, dst):
         CHECK_EQ(len(src_value), len(dst_value))
         dst_value.copyFrom(src_value)
         dst_param.setValueUpdated()
-        
+
+
 def print_parameters(src):
-    src_params = [src.getParameter(i)
-               for i in xrange(src.getParameterSize())]
+    src_params = [src.getParameter(i) for i in xrange(src.getParameterSize())]
 
     print "***************"
     for p in src_params:
         print "Name is %s" % p.getName()
-        print "value is %s \n" % p.getBuf(api.PARAMETER_VALUE).copyToNumpyArray()
+        print "value is %s \n" % p.getBuf(api.PARAMETER_VALUE).copyToNumpyArray(
+        )
+
 
 def load_mnist_data(imageFile):
     f = open(imageFile, "rb")
@@ -86,33 +89,36 @@ def load_mnist_data(imageFile):
         n = 60000
     else:
         n = 10000
-    
-    data = numpy.fromfile(f, 'ubyte', count=n*28*28).reshape((n, 28*28))
+
+    data = numpy.fromfile(f, 'ubyte', count=n * 28 * 28).reshape((n, 28 * 28))
     data = data / 255.0 * 2.0 - 1.0
 
     f.close()
     return data.astype('float32')
 
+
 def load_cifar_data(cifar_path):
     batch_size = 10000
-    data = numpy.zeros((5*batch_size, 32*32*3), dtype = "float32")
+    data = numpy.zeros((5 * batch_size, 32 * 32 * 3), dtype="float32")
     for i in range(1, 6):
         file = cifar_path + "/data_batch_" + str(i)
         fo = open(file, 'rb')
         dict = cPickle.load(fo)
         fo.close()
-        data[(i - 1)*batch_size:(i*batch_size), :] = dict["data"]
-    
+        data[(i - 1) * batch_size:(i * batch_size), :] = dict["data"]
+
     data = data / 255.0 * 2.0 - 1.0
     return data
 
+
 # synthesize 2-D uniform data
 def load_uniform_data():
     data = numpy.random.rand(1000000, 2).astype('float32')
     return data
 
+
 def merge(images, size):
-    if images.shape[1] == 28*28:
+    if images.shape[1] == 28 * 28:
         h, w, c = 28, 28, 1
     else:
         h, w, c = 32, 32, 3
@@ -124,6 +130,7 @@ def merge(images, size):
           ((images[idx, :].reshape((h, w, c), order="F").transpose(1, 0, 2) + 1.0) / 2.0 * 255.0)
     return img.astype('uint8')
 
+
 def save_images(images, path):
     merged_img = merge(images, [8, 8])
     if merged_img.shape[2] == 1:
@@ -131,14 +138,17 @@ def save_images(images, path):
     else:
         im = Image.fromarray(merged_img, mode="RGB")
     im.save(path)
-    
+
+
 def get_real_samples(batch_size, data_np):
-    return data_np[numpy.random.choice(data_np.shape[0], batch_size, 
-                                       replace=False),:]
-    
+    return data_np[numpy.random.choice(
+        data_np.shape[0], batch_size, replace=False), :]
+
+
 def get_noise(batch_size, noise_dim):
     return numpy.random.normal(size=(batch_size, noise_dim)).astype('float32')
 
+
 def get_fake_samples(generator_machine, batch_size, noise):
     gen_inputs = api.Arguments.createArguments(1)
     gen_inputs.setSlotValue(0, api.Matrix.createDenseFromNumpy(noise))
@@ -147,12 +157,14 @@ def get_fake_samples(generator_machine, batch_size, noise):
     fake_samples = gen_outputs.getSlotValue(0).copyToNumpyMat()
     return fake_samples
 
+
 def get_training_loss(training_machine, inputs):
     outputs = api.Arguments.createArguments(0)
     training_machine.forward(inputs, outputs, api.PASS_TEST)
     loss = outputs.getSlotValue(0).copyToNumpyMat()
     return numpy.mean(loss)
 
+
 def prepare_discriminator_data_batch_pos(batch_size, data_np):
     real_samples = get_real_samples(batch_size, data_np)
     labels = numpy.ones(batch_size, dtype='int32')
@@ -161,6 +173,7 @@ def prepare_discriminator_data_batch_pos(batch_size, data_np):
     inputs.setSlotIds(1, api.IVector.createVectorFromNumpy(labels))
     return inputs
 
+
 def prepare_discriminator_data_batch_neg(generator_machine, batch_size, noise):
     fake_samples = get_fake_samples(generator_machine, batch_size, noise)
     labels = numpy.zeros(batch_size, dtype='int32')
@@ -169,6 +182,7 @@ def prepare_discriminator_data_batch_neg(generator_machine, batch_size, noise):
     inputs.setSlotIds(1, api.IVector.createVectorFromNumpy(labels))
     return inputs
 
+
 def prepare_generator_data_batch(batch_size, noise):
     label = numpy.ones(batch_size, dtype='int32')
     inputs = api.Arguments.createArguments(2)
@@ -193,10 +207,9 @@ def get_layer_size(model_conf, layer_name):
 def main():
     parser = argparse.ArgumentParser()
     parser.add_argument("-d", "--data_source", help="mnist or cifar or uniform")
-    parser.add_argument("--use_gpu", default="1", 
-                        help="1 means use gpu for training")
-    parser.add_argument("--gpu_id", default="0", 
-                        help="the gpu_id parameter")
+    parser.add_argument(
+        "--use_gpu", default="1", help="1 means use gpu for training")
+    parser.add_argument("--gpu_id", default="0", help="the gpu_id parameter")
     args = parser.parse_args()
     data_source = args.data_source
     use_gpu = args.use_gpu
@@ -208,30 +221,32 @@ def main():
 
     if not os.path.exists("./%s_params/" % data_source):
         os.makedirs("./%s_params/" % data_source)
-        
-    api.initPaddle('--use_gpu=' + use_gpu, '--dot_period=10', '--log_period=100', 
-                   '--gpu_id=' + args.gpu_id, '--save_dir=' + "./%s_params/" % data_source)
-    
+
+    api.initPaddle('--use_gpu=' + use_gpu, '--dot_period=10',
+                   '--log_period=100', '--gpu_id=' + args.gpu_id,
+                   '--save_dir=' + "./%s_params/" % data_source)
+
     if data_source == "uniform":
         conf = "gan_conf.py"
         num_iter = 10000
     else:
         conf = "gan_conf_image.py"
         num_iter = 1000
-        
+
     gen_conf = parse_config(conf, "mode=generator_training,data=" + data_source)
-    dis_conf = parse_config(conf, "mode=discriminator_training,data=" + data_source)
+    dis_conf = parse_config(conf,
+                            "mode=discriminator_training,data=" + data_source)
     generator_conf = parse_config(conf, "mode=generator,data=" + data_source)
     batch_size = dis_conf.opt_config.batch_size
     noise_dim = get_layer_size(gen_conf.model_config, "noise")
-    
+
     if data_source == "mnist":
         data_np = load_mnist_data("./data/mnist_data/train-images-idx3-ubyte")
     elif data_source == "cifar":
         data_np = load_cifar_data("./data/cifar-10-batches-py/")
     else:
         data_np = load_uniform_data()
-    
+
     # this creates a gradient machine for discriminator
     dis_training_machine = api.GradientMachine.createFromConfigProto(
         dis_conf.model_config)
@@ -244,26 +259,24 @@ def main():
     logger.info(str(generator_conf.model_config))
     generator_machine = api.GradientMachine.createFromConfigProto(
         generator_conf.model_config)
-    
-    dis_trainer = api.Trainer.create(
-        dis_conf, dis_training_machine)
 
-    gen_trainer = api.Trainer.create(
-        gen_conf, gen_training_machine)
-    
+    dis_trainer = api.Trainer.create(dis_conf, dis_training_machine)
+
+    gen_trainer = api.Trainer.create(gen_conf, gen_training_machine)
+
     dis_trainer.startTrain()
     gen_trainer.startTrain()
-    
+
     # Sync parameters between networks (GradientMachine) at the beginning
     copy_shared_parameters(gen_training_machine, dis_training_machine)
     copy_shared_parameters(gen_training_machine, generator_machine)
-    
+
     # constrain that either discriminator or generator can not be trained
     # consecutively more than MAX_strike times
     curr_train = "dis"
     curr_strike = 0
     MAX_strike = 5
-     
+
     for train_pass in xrange(100):
         dis_trainer.startTrainPass()
         gen_trainer.startTrainPass()
@@ -272,23 +285,25 @@ def main():
             noise = get_noise(batch_size, noise_dim)
             data_batch_dis_pos = prepare_discriminator_data_batch_pos(
                 batch_size, data_np)
-            dis_loss_pos = get_training_loss(dis_training_machine, data_batch_dis_pos)
-            
+            dis_loss_pos = get_training_loss(dis_training_machine,
+                                             data_batch_dis_pos)
+
             data_batch_dis_neg = prepare_discriminator_data_batch_neg(
                 generator_machine, batch_size, noise)
-            dis_loss_neg = get_training_loss(dis_training_machine, data_batch_dis_neg)            
-                         
+            dis_loss_neg = get_training_loss(dis_training_machine,
+                                             data_batch_dis_neg)
+
             dis_loss = (dis_loss_pos + dis_loss_neg) / 2.0
-            
+
             # Do forward pass in generator to get the gen_loss
-            data_batch_gen = prepare_generator_data_batch(
-                    batch_size, noise)
+            data_batch_gen = prepare_generator_data_batch(batch_size, noise)
             gen_loss = get_training_loss(gen_training_machine, data_batch_gen)
-             
+
             if i % 100 == 0:
-                print "d_pos_loss is %s     d_neg_loss is %s" % (dis_loss_pos, dis_loss_neg) 
+                print "d_pos_loss is %s     d_neg_loss is %s" % (dis_loss_pos,
+                                                                 dis_loss_neg)
                 print "d_loss is %s    g_loss is %s" % (dis_loss, gen_loss)
-            
+
             # Decide which network to train based on the training history
             # And the relative size of the loss        
             if (not (curr_train == "dis" and curr_strike == MAX_strike)) and \
@@ -297,11 +312,12 @@ def main():
                     curr_strike += 1
                 else:
                     curr_train = "dis"
-                    curr_strike = 1                
+                    curr_strike = 1
                 dis_trainer.trainOneDataBatch(batch_size, data_batch_dis_neg)
-                dis_trainer.trainOneDataBatch(batch_size, data_batch_dis_pos)               
-                copy_shared_parameters(dis_training_machine, gen_training_machine)
- 
+                dis_trainer.trainOneDataBatch(batch_size, data_batch_dis_pos)
+                copy_shared_parameters(dis_training_machine,
+                                       gen_training_machine)
+
             else:
                 if curr_train == "gen":
                     curr_strike += 1
@@ -311,19 +327,23 @@ def main():
                 gen_trainer.trainOneDataBatch(batch_size, data_batch_gen)
                 # TODO: add API for paddle to allow true parameter sharing between different GradientMachines 
                 # so that we do not need to copy shared parameters. 
-                copy_shared_parameters(gen_training_machine, dis_training_machine)
+                copy_shared_parameters(gen_training_machine,
+                                       dis_training_machine)
                 copy_shared_parameters(gen_training_machine, generator_machine)
- 
+
         dis_trainer.finishTrainPass()
         gen_trainer.finishTrainPass()
         # At the end of each pass, save the generated samples/images
         fake_samples = get_fake_samples(generator_machine, batch_size, noise)
         if data_source == "uniform":
-            plot2DScatter(fake_samples, "./%s_samples/train_pass%s.png" % (data_source, train_pass))
+            plot2DScatter(fake_samples, "./%s_samples/train_pass%s.png" %
+                          (data_source, train_pass))
         else:
-            save_images(fake_samples, "./%s_samples/train_pass%s.png" % (data_source, train_pass))
+            save_images(fake_samples, "./%s_samples/train_pass%s.png" %
+                        (data_source, train_pass))
     dis_trainer.finishTrain()
     gen_trainer.finishTrain()
 
+
 if __name__ == '__main__':
     main()

demo/quick_start/trainer_config.resnet-lstm.py

@@ -13,7 +13,6 @@
 # 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.
-
 """
 This configuration is a demonstration of how to implement the stacked LSTM
 with residual connections, i.e. an LSTM layer takes the sum of the hidden states
@@ -46,11 +45,12 @@ is_predict = get_config_arg('is_predict', bool, False)
 trn = 'data/train.list' if not is_predict else None
 tst = 'data/test.list' if not is_predict else 'data/pred.list'
 process = 'process' if not is_predict else 'process_predict'
-define_py_data_sources2(train_list=trn,
-                        test_list=tst,
-                        module="dataprovider_emb",
-                        obj=process,
-                        args={"dictionary": word_dict})
+define_py_data_sources2(
+    train_list=trn,
+    test_list=tst,
+    module="dataprovider_emb",
+    obj=process,
+    args={"dictionary": word_dict})
 
 batch_size = 128 if not is_predict else 1
 settings(
@@ -58,10 +58,9 @@ settings(
     learning_rate=2e-3,
     learning_method=AdamOptimizer(),
     regularization=L2Regularization(8e-4),
-    gradient_clipping_threshold=25
-)
+    gradient_clipping_threshold=25)
 
-bias_attr = ParamAttr(initial_std=0.,l2_rate=0.)
+bias_attr = ParamAttr(initial_std=0., l2_rate=0.)
 
 data = data_layer(name="word", size=len(word_dict))
 emb = embedding_layer(input=data, size=128)
@@ -73,17 +72,15 @@ for i in range(3):
     # The input to the current layer is the sum of the hidden state
     # and input of the previous layer.
     current_input = addto_layer(input=[previous_input, previous_hidden_state])
-    hidden_state = simple_lstm(input=current_input, size=128,
-                               lstm_cell_attr=ExtraAttr(drop_rate=0.1))
+    hidden_state = simple_lstm(
+        input=current_input, size=128, lstm_cell_attr=ExtraAttr(drop_rate=0.1))
     previous_input, previous_hidden_state = current_input, hidden_state
 
 lstm = previous_hidden_state
 
 lstm_last = pooling_layer(input=lstm, pooling_type=MaxPooling())
-output = fc_layer(input=lstm_last, size=2,
-                  bias_attr=bias_attr,
-                  act=SoftmaxActivation())
-
+output = fc_layer(
+    input=lstm_last, size=2, bias_attr=bias_attr, act=SoftmaxActivation())
 
 if is_predict:
     maxid = maxid_layer(output)

demo/semantic_role_labeling/data/extract_dict_feature.py

@@ -33,7 +33,7 @@ def extract_dict_features(pair_file, feature_file):
                 ctx_n1 = sentence_list[verb_index - 1]
             else:
                 ctx_n1 = 'bos'
-            
+
             if verb_index > 1:
                 mark[verb_index - 2] = 1
                 ctx_n2 = sentence_list[verb_index - 2]
@@ -43,13 +43,13 @@ def extract_dict_features(pair_file, feature_file):
             mark[verb_index] = 1
             ctx_0 = sentence_list[verb_index]
 
-            if verb_index < len(labels_list) - 2:
+            if verb_index < len(labels_list) - 1:
                 mark[verb_index + 1] = 1
                 ctx_p1 = sentence_list[verb_index + 1]
             else:
                 ctx_p1 = 'eos'
-            
-            if verb_index < len(labels_list) - 3:
+
+            if verb_index < len(labels_list) - 2:
                 mark[verb_index + 2] = 1
                 ctx_p2 = sentence_list[verb_index + 2]
             else:
@@ -69,7 +69,6 @@ def extract_dict_features(pair_file, feature_file):
             feature_out.write(feature_str + '\n')
 
 
-
 if __name__ == '__main__':
 
     usage = '-p pair_file -f feature_file'

demo/semantic_role_labeling/data/extract_pairs.py

@@ -66,8 +66,8 @@ def transform_labels(sentences, labels):
         else:
             verb_list = []
             for x in labels[i][0]:
-                if x !='-':
-                   verb_list.append(x)
+                if x != '-':
+                    verb_list.append(x)
 
             for j in xrange(1, len(labels[i])):
                 label_list = labels[i][j]
@@ -93,7 +93,7 @@ def transform_labels(sentences, labels):
                         is_in_bracket = True
                     else:
                         print 'error:', ll
-                sen_lab_pair.append((sentences[i], verb_list[j-1], label_seq))
+                sen_lab_pair.append((sentences[i], verb_list[j - 1], label_seq))
     return sen_lab_pair
 
 
@@ -103,7 +103,7 @@ def write_file(sen_lab_pair, output_file):
             sentence = x[0]
             label_seq = ' '.join(x[2])
             assert len(sentence.split()) == len(x[2])
-            fout.write(sentence + '\t' + x[1]+'\t' +label_seq + '\n')
+            fout.write(sentence + '\t' + x[1] + '\t' + label_seq + '\n')
 
 
 if __name__ == '__main__':

demo/semantic_role_labeling/dataprovider.py

@@ -21,7 +21,7 @@ def hook(settings, word_dict, label_dict, predicate_dict, **kwargs):
     settings.word_dict = word_dict
     settings.label_dict = label_dict
     settings.predicate_dict = predicate_dict
-   
+
     #all inputs are integral and sequential type
     settings.slots = [
         integer_value_sequence(len(word_dict)),
@@ -29,25 +29,28 @@ def hook(settings, word_dict, label_dict, predicate_dict, **kwargs):
         integer_value_sequence(len(word_dict)),
         integer_value_sequence(len(word_dict)),
         integer_value_sequence(len(word_dict)),
-        integer_value_sequence(len(word_dict)), 
-        integer_value_sequence(len(predicate_dict)),
-        integer_value_sequence(2),
+        integer_value_sequence(len(word_dict)),
+        integer_value_sequence(len(predicate_dict)), integer_value_sequence(2),
         integer_value_sequence(len(label_dict))
     ]
 
 
 def get_batch_size(yeild_data):
     return len(yeild_data[0])
-    
 
-@provider(init_hook=hook, should_shuffle=True, calc_batch_size=get_batch_size, 
-          can_over_batch_size=False, cache=CacheType.CACHE_PASS_IN_MEM)
+
+@provider(
+    init_hook=hook,
+    should_shuffle=True,
+    calc_batch_size=get_batch_size,
+    can_over_batch_size=False,
+    cache=CacheType.CACHE_PASS_IN_MEM)
 def process(settings, file_name):
     with open(file_name, 'r') as fdata:
         for line in fdata:
             sentence, predicate, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2,  mark, label = \
                 line.strip().split('\t')
-           
+
             words = sentence.split()
             sen_len = len(words)
             word_slot = [settings.word_dict.get(w, UNK_IDX) for w in words]

demo/semantic_role_labeling/db_lstm.py

@@ -20,7 +20,7 @@ from paddle.trainer_config_helpers import *
 #file paths
 word_dict_file = './data/wordDict.txt'
 label_dict_file = './data/targetDict.txt'
-predicate_file= './data/verbDict.txt'
+predicate_file = './data/verbDict.txt'
 train_list_file = './data/train.list'
 test_list_file = './data/test.list'
 
@@ -47,7 +47,6 @@ if not is_predict:
             w = line.strip()
             predicate_dict[w] = i
 
-
     if is_test:
         train_list_file = None
 
@@ -57,9 +56,11 @@ if not is_predict:
         test_list=test_list_file,
         module='dataprovider',
         obj='process',
-        args={'word_dict': word_dict,
-              'label_dict': label_dict,
-              'predicate_dict': predicate_dict })
+        args={
+            'word_dict': word_dict,
+            'label_dict': label_dict,
+            'predicate_dict': predicate_dict
+        })
 
     word_dict_len = len(word_dict)
     label_dict_len = len(label_dict)
@@ -77,24 +78,16 @@ mark_dim = 5
 hidden_dim = 512
 depth = 8
 
-
-
 ########################### Optimizer #######################################
 
-
 settings(
     batch_size=150,
     learning_method=MomentumOptimizer(momentum=0),
     learning_rate=2e-2,
     regularization=L2Regularization(8e-4),
     is_async=False,
-    model_average=ModelAverage(average_window=0.5,
-                               max_average_window=10000),
-                               
-)
-
-
-
+    model_average=ModelAverage(
+        average_window=0.5, max_average_window=10000), )
 
 ####################################### network ##############################
 #8 features and 1 target
@@ -108,22 +101,28 @@ ctx_p1 = data_layer(name='ctx_p1_data', size=word_dict_len)
 ctx_p2 = data_layer(name='ctx_p2_data', size=word_dict_len)
 mark = data_layer(name='mark_data', size=mark_dict_len)
 
-
 if not is_predict:
     target = data_layer(name='target', size=label_dict_len)
 
-
-default_std=1/math.sqrt(hidden_dim)/3.0
+default_std = 1 / math.sqrt(hidden_dim) / 3.0
 
 emb_para = ParameterAttribute(name='emb', initial_std=0., learning_rate=0.)
 std_0 = ParameterAttribute(initial_std=0.)
-std_default = ParameterAttribute(initial_std=default_std) 
-
-predicate_embedding = embedding_layer(size=word_dim, input=predicate, param_attr=ParameterAttribute(name='vemb',initial_std=default_std))
-mark_embedding = embedding_layer(name='word_ctx-in_embedding', size=mark_dim, input=mark, param_attr=std_0)
-
-word_input=[word, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2]
-emb_layers = [embedding_layer(size=word_dim, input=x, param_attr=emb_para) for x in word_input]
+std_default = ParameterAttribute(initial_std=default_std)
+
+predicate_embedding = embedding_layer(
+    size=word_dim,
+    input=predicate,
+    param_attr=ParameterAttribute(
+        name='vemb', initial_std=default_std))
+mark_embedding = embedding_layer(
+    name='word_ctx-in_embedding', size=mark_dim, input=mark, param_attr=std_0)
+
+word_input = [word, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2]
+emb_layers = [
+    embedding_layer(
+        size=word_dim, input=x, param_attr=emb_para) for x in word_input
+]
 emb_layers.append(predicate_embedding)
 emb_layers.append(mark_embedding)
 
@@ -131,84 +130,89 @@ hidden_0 = mixed_layer(
     name='hidden0',
     size=hidden_dim,
     bias_attr=std_default,
-    input=[ full_matrix_projection(input=emb, param_attr=std_default ) for emb in emb_layers ])
-
+    input=[
+        full_matrix_projection(
+            input=emb, param_attr=std_default) for emb in emb_layers
+    ])
 
 mix_hidden_lr = 1e-3
 lstm_para_attr = ParameterAttribute(initial_std=0.0, learning_rate=1.0)
-hidden_para_attr = ParameterAttribute(initial_std=default_std, learning_rate=mix_hidden_lr)
-
-lstm_0 = lstmemory(name='lstm0',
-                   input=hidden_0, 
-                   act=ReluActivation(),
-                   gate_act=SigmoidActivation(),
-                   state_act=SigmoidActivation(),
-                   bias_attr=std_0,
-                   param_attr=lstm_para_attr)
+hidden_para_attr = ParameterAttribute(
+    initial_std=default_std, learning_rate=mix_hidden_lr)
+
+lstm_0 = lstmemory(
+    name='lstm0',
+    input=hidden_0,
+    act=ReluActivation(),
+    gate_act=SigmoidActivation(),
+    state_act=SigmoidActivation(),
+    bias_attr=std_0,
+    param_attr=lstm_para_attr)
 
 #stack L-LSTM and R-LSTM with direct edges
 input_tmp = [hidden_0, lstm_0]
 
-
 for i in range(1, depth):
 
-    mix_hidden = mixed_layer(name='hidden'+str(i),
-                             size=hidden_dim, 
-                             bias_attr=std_default,
-                             input=[full_matrix_projection(input=input_tmp[0], param_attr=hidden_para_attr),
-                                    full_matrix_projection(input=input_tmp[1], param_attr=lstm_para_attr)
-                                   ]
-                             )
-
-    lstm = lstmemory(name='lstm'+str(i),
-                     input=mix_hidden,
-                     act=ReluActivation(),
-                     gate_act=SigmoidActivation(),
-                     state_act=SigmoidActivation(),
-                     reverse=((i % 2)==1),
-                     bias_attr=std_0,
-                     param_attr=lstm_para_attr)
+    mix_hidden = mixed_layer(
+        name='hidden' + str(i),
+        size=hidden_dim,
+        bias_attr=std_default,
+        input=[
+            full_matrix_projection(
+                input=input_tmp[0], param_attr=hidden_para_attr),
+            full_matrix_projection(
+                input=input_tmp[1], param_attr=lstm_para_attr)
+        ])
+
+    lstm = lstmemory(
+        name='lstm' + str(i),
+        input=mix_hidden,
+        act=ReluActivation(),
+        gate_act=SigmoidActivation(),
+        state_act=SigmoidActivation(),
+        reverse=((i % 2) == 1),
+        bias_attr=std_0,
+        param_attr=lstm_para_attr)
 
     input_tmp = [mix_hidden, lstm]
 
-feature_out = mixed_layer(name='output',
-                          size=label_dict_len,
-                          bias_attr=std_default, 
-                          input=[full_matrix_projection(input=input_tmp[0], param_attr=hidden_para_attr),
-                                 full_matrix_projection(input=input_tmp[1], param_attr=lstm_para_attr)
-                                ],
-                          )
-
-
+feature_out = mixed_layer(
+    name='output',
+    size=label_dict_len,
+    bias_attr=std_default,
+    input=[
+        full_matrix_projection(
+            input=input_tmp[0], param_attr=hidden_para_attr),
+        full_matrix_projection(
+            input=input_tmp[1], param_attr=lstm_para_attr)
+    ], )
 
 if not is_predict:
-    crf_l = crf_layer( name = 'crf',
-                       size = label_dict_len,
-                       input = feature_out, 
-                       label = target,
-                       param_attr=ParameterAttribute(name='crfw',initial_std=default_std, learning_rate=mix_hidden_lr)
-
-                      )
-
-    
-    crf_dec_l = crf_decoding_layer(name = 'crf_dec_l',
-                                   size = label_dict_len,
-                                   input = feature_out,
-                                   label = target,
-                                   param_attr=ParameterAttribute(name='crfw')
-                                       )
-
+    crf_l = crf_layer(
+        name='crf',
+        size=label_dict_len,
+        input=feature_out,
+        label=target,
+        param_attr=ParameterAttribute(
+            name='crfw', initial_std=default_std, learning_rate=mix_hidden_lr))
+
+    crf_dec_l = crf_decoding_layer(
+        name='crf_dec_l',
+        size=label_dict_len,
+        input=feature_out,
+        label=target,
+        param_attr=ParameterAttribute(name='crfw'))
 
     eval = sum_evaluator(input=crf_dec_l)
-        
+
     outputs(crf_l)
 
 else:
-    crf_dec_l = crf_decoding_layer(name = 'crf_dec_l',
-                                   size = label_dict_len,
-                                   input = feature_out,
-                                   param_attr=ParameterAttribute(name='crfw')
-                                       )
+    crf_dec_l = crf_decoding_layer(
+        name='crf_dec_l',
+        size=label_dict_len,
+        input=feature_out,
+        param_attr=ParameterAttribute(name='crfw'))
 
     outputs(crf_dec_l)
-

demo/semantic_role_labeling/predict.py

@@ -26,7 +26,8 @@ UNK_IDX = 0
 
 
 class Prediction():
-    def __init__(self, train_conf, dict_file, model_dir, label_file, predicate_dict_file):
+    def __init__(self, train_conf, dict_file, model_dir, label_file,
+                 predicate_dict_file):
         """
         train_conf: trainer configure.
         dict_file: word dictionary file name.
@@ -35,7 +36,7 @@ class Prediction():
 
         self.dict = {}
         self.labels = {}
-        self.predicate_dict={}
+        self.predicate_dict = {}
         self.labels_reverse = {}
         self.load_dict_label(dict_file, label_file, predicate_dict_file)
 
@@ -44,25 +45,18 @@ class Prediction():
         len_pred = len(self.predicate_dict)
 
         conf = parse_config(
-            train_conf,
-            'dict_len=' + str(len_dict) + 
-            ',label_len=' + str(len_label) +
-            ',pred_len=' + str(len_pred) +
-            ',is_predict=True')
+            train_conf, 'dict_len=' + str(len_dict) + ',label_len=' +
+            str(len_label) + ',pred_len=' + str(len_pred) + ',is_predict=True')
         self.network = swig_paddle.GradientMachine.createFromConfigProto(
             conf.model_config)
         self.network.loadParameters(model_dir)
 
         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(len_dict), 
-            integer_value_sequence(len_pred),
-            integer_value_sequence(2)
-            ]
+            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(len_dict),
+            integer_value_sequence(len_pred), integer_value_sequence(2)
+        ]
         self.converter = DataProviderConverter(slots)
 
     def load_dict_label(self, dict_file, label_file, predicate_dict_file):
@@ -78,6 +72,7 @@ class Prediction():
 
         for line_count, line in enumerate(open(predicate_dict_file, 'r')):
             self.predicate_dict[line.strip()] = line_count
+
     def get_data(self, data_file):
         """
         Get input data of paddle format.
@@ -88,9 +83,10 @@ class Prediction():
                 ).split('\t')
                 words = sentence.split()
                 sen_len = len(words)
-                 
+
                 word_slot = [self.dict.get(w, UNK_IDX) for w in words]
-                predicate_slot = [self.predicate_dict.get(predicate, UNK_IDX)] * sen_len
+                predicate_slot = [self.predicate_dict.get(predicate, UNK_IDX)
+                                  ] * sen_len
                 ctx_n2_slot = [self.dict.get(ctx_n2, UNK_IDX)] * sen_len
                 ctx_n1_slot = [self.dict.get(ctx_n1, UNK_IDX)] * sen_len
                 ctx_0_slot = [self.dict.get(ctx_0, UNK_IDX)] * sen_len
@@ -99,7 +95,7 @@ class Prediction():
 
                 marks = mark.split()
                 mark_slot = [int(w) for w in marks]
-                
+
                 yield word_slot, ctx_n2_slot, ctx_n1_slot, \
                       ctx_0_slot, ctx_p1_slot, ctx_p2_slot, predicate_slot, mark_slot
 
@@ -123,8 +119,9 @@ class Prediction():
 
 
 def option_parser():
-    usage = ("python predict.py -c config -w model_dir " 
-             "-d word dictionary -l label_file -i input_file  -p pred_dict_file")
+    usage = (
+        "python predict.py -c config -w model_dir "
+        "-d word dictionary -l label_file -i input_file  -p pred_dict_file")
     parser = OptionParser(usage="usage: %s [options]" % usage)
     parser.add_option(
         "-c",
@@ -187,8 +184,9 @@ def main():
     output_file = options.output_file
 
     swig_paddle.initPaddle("--use_gpu=0")
-    predict = Prediction(train_conf, dict_file, model_path, label_file, predict_dict_file)
-    predict.predict(data_file,output_file)
+    predict = Prediction(train_conf, dict_file, model_path, label_file,
+                         predict_dict_file)
+    predict.predict(data_file, output_file)
 
 
 if __name__ == '__main__':

demo/sentiment/predict.py

@@ -71,9 +71,7 @@ class SentimentPrediction():
         transform word into integer index according to the dictionary.
         """
         words = data.strip().split()
-        word_slot = [
-            self.word_dict[w] for w in words if w in self.word_dict
-        ]
+        word_slot = [self.word_dict[w] for w in words if w in self.word_dict]
         return word_slot
 
     def batch_predict(self, data_batch):
@@ -85,8 +83,8 @@ class SentimentPrediction():
             if self.label is None:
                 print("predicting label is %d" % (lab[0]))
             else:
-                print("predicting label is %s" %
-                      (self.label[lab[0]]))
+                print("predicting label is %s" % (self.label[lab[0]]))
+
 
 def option_parser():
     usage = "python predict.py -n config -w model_dir -d dictionary -i input_file "
@@ -143,9 +141,10 @@ def main():
         batch.append([predict.get_index(line)])
         if len(batch) == batch_size:
             predict.batch_predict(batch)
-            batch=[]
+            batch = []
     if len(batch) > 0:
         predict.batch_predict(batch)
 
+
 if __name__ == '__main__':
     main()

doc/CMakeLists.txt

@@ -7,25 +7,50 @@ if(NOT DEFINED SPHINX_THEME_DIR)
 endif()
 
 # configured documentation tools and intermediate build results
-set(BINARY_BUILD_DIR "${CMAKE_CURRENT_BINARY_DIR}/_build")
+set(BINARY_BUILD_DIR_EN "${CMAKE_CURRENT_BINARY_DIR}/en/_build")
 
 # Sphinx cache with pickled ReST documents
-set(SPHINX_CACHE_DIR "${CMAKE_CURRENT_BINARY_DIR}/_doctrees")
+set(SPHINX_CACHE_DIR_EN "${CMAKE_CURRENT_BINARY_DIR}/en/_doctrees")
 
-# HTML output directory
-set(SPHINX_HTML_DIR "${CMAKE_CURRENT_BINARY_DIR}/html")
+# HTML output director
+set(SPHINX_HTML_DIR_EN "${CMAKE_CURRENT_BINARY_DIR}/en/html")
 
 configure_file(
-    "${CMAKE_CURRENT_SOURCE_DIR}/conf.py.in"
-    "${BINARY_BUILD_DIR}/conf.py"
+    "${CMAKE_CURRENT_SOURCE_DIR}/conf.py.en.in"
+    "${BINARY_BUILD_DIR_EN}/conf.py"
     @ONLY)
 
 sphinx_add_target(paddle_docs
                   html
-                  ${BINARY_BUILD_DIR}
-                  ${SPHINX_CACHE_DIR}
+                  ${BINARY_BUILD_DIR_EN}
+                  ${SPHINX_CACHE_DIR_EN}
                   ${CMAKE_CURRENT_SOURCE_DIR}
-                  ${SPHINX_HTML_DIR})
+                  ${SPHINX_HTML_DIR_EN})
 
 add_dependencies(paddle_docs
   gen_proto_py)
+
+
+# configured documentation tools and intermediate build results
+set(BINARY_BUILD_DIR_CN "${CMAKE_CURRENT_BINARY_DIR}/cn/_build")
+
+# Sphinx cache with pickled ReST documents
+set(SPHINX_CACHE_DIR_CN "${CMAKE_CURRENT_BINARY_DIR}/cn/_doctrees")
+
+# HTML output directory
+set(SPHINX_HTML_DIR_CN "${CMAKE_CURRENT_BINARY_DIR}/cn/html")
+
+configure_file(
+    "${CMAKE_CURRENT_SOURCE_DIR}/conf.py.cn.in"
+    "${BINARY_BUILD_DIR_CN}/conf.py"
+    @ONLY)
+
+sphinx_add_target(paddle_docs_cn
+                  html
+                  ${BINARY_BUILD_DIR_CN}
+                  ${SPHINX_CACHE_DIR_CN}
+                  ${CMAKE_CURRENT_SOURCE_DIR}
+                  ${SPHINX_HTML_DIR_CN})
+
+add_dependencies(paddle_docs_cn
+  gen_proto_py)

doc_cn/ui/data_provider/dataprovider.rst → doc/api/data_provider/dataprovider_cn.rst

 DataProvider的介绍
 ==================
 
-DataProvider是PaddlePaddle负责提供数据的模块。其作用是将数据传入内存或显存，让神经网络可以进行训练或预测。用户可以通过简单使用Python接口 `PyDataProvider2 <pydataprovider2.html>`_ ，来自定义传数据的过程。如果有更复杂的使用，或者需要更高的效率，用户也可以在C++端自定义一个 ``DataProvider`` 。
+DataProvider是PaddlePaddle负责提供数据的模块。其作用是将数据传入内存或显存，让神经网络可以进行训练或预测。用户可以通过简单使用Python接口 `PyDataProvider2 <pydataprovider2.html>`_ ，来自定义传数据的过程。如果有更复杂的使用，或者需要更高的效率，用户也可以在C++端自定义一个 ``DataProvider`` 。
 
 PaddlePaddle需要用户在网络配置（trainer_config.py）中定义使用哪种DataProvider，并且在DataProvider中实现如何访问训练文件列表（train.list）或测试文件列表（test.list）。
 
-- train.list和test.list存放在本地（推荐直接存放到训练目录，以相对路径引用)。一般情况下，两者均为纯文本文件，其中每一行对应一个数据文件地址：
-  
-  - 如果数据文件存于本地磁盘，这个地址则为它的绝对路径或相对路径(相对于PaddlePaddle程序运行时的路径)。
-  - 地址也可以为hdfs文件路径，或者数据库连接路径等。
-  - 由于这个地址会被DataProvider使用，因此，如何解析该地址也是用户自定义DataProvider时需要考虑的地方。
+- train.list和test.list存放在本地（推荐直接存放到训练目录，以相对路径引用)。一般情况下，两者均为纯文本文件，其中每一行对应一个数据文件地址：
+  
+  - 如果数据文件存于本地磁盘，这个地址则为它的绝对路径或相对路径(相对于PaddlePaddle程序运行时的路径)。
+  - 地址也可以为hdfs文件路径，或者数据库连接路径等。
+  - 由于这个地址会被DataProvider使用，因此，如何解析该地址也是用户自定义DataProvider时需要考虑的地方。
 - 如果没有设置test.list，或设置为None，那么在训练过程中不会执行测试操作；否则，会根据命令行参数指定的测试方式，在训练过程中进行测试，从而防止过拟合。

doc/api/data_provider/pydataprovider2_en.rst

-..  _api_pydataprovider:
+..  _api_pydataprovider2:
 
 PyDataProvider2
 ===============
@@ -24,18 +24,18 @@ of 28 x 28 pixels.
 
 A small part of the original data as an example is shown as below:
 
-.. literalinclude:: ../../../doc_cn/ui/data_provider/mnist_train.txt
+.. literalinclude:: src/mnist_train.txt
 
 Each line of the data contains two parts, separated by :code:`;`. The first part is
 label of an image. The second part contains 28x28 pixel float values.
 
 Just write path of the above data into train.list. It looks like this:
 
-.. literalinclude:: ../../../doc_cn/ui/data_provider/train.list
+.. literalinclude:: src/train.list
 
 The corresponding dataprovider is shown as below:
 
-.. literalinclude:: ../../../doc_cn/ui/data_provider/mnist_provider.py
+.. literalinclude:: src/mnist_provider.dict.py
 
 The first line imports PyDataProvider2 package.
 The main function is the process function, that has two parameters.
@@ -74,7 +74,7 @@ sample by using keywords :code:`yield`.
 Only a few lines of codes need to be added into the training configuration file,
 you can take this as an example.
 
-.. literalinclude:: ../../../doc_cn/ui/data_provider/mnist_config.py
+.. literalinclude:: src/mnist_config.py
 
 Here we specify training data by :code:`train.list`, and no testing data is specified.
 The method which actually provide data is :code:`process`.
@@ -83,7 +83,7 @@ User also can use another style to provide data, which defines the
 :code:`data_layer`'s name explicitly when `yield`. For example,
 the :code:`dataprovider` is shown as below.
 
-.. literalinclude:: ../../../doc_cn/ui/data_provider/mnist_provider.dict.py
+.. literalinclude:: src/mnist_provider.dict.py
    :linenos:
 
 If user did't give the :code:`data_layer`'s name, PaddlePaddle will use
@@ -104,6 +104,8 @@ And PaddlePadle will do all of the rest things\:
 
 Is this cool?
 
+..  _api_pydataprovider2_sequential_model:
+
 DataProvider for the sequential model
 -------------------------------------
 A sequence model takes sequences as its input. A sequence is made up of several
@@ -119,11 +121,11 @@ negative sentiment (marked by 0 and 1 respectively).
 
 A small part of the original data as an example can be found in the path below:
 
-.. literalinclude:: ../../../doc_cn/ui/data_provider/sentimental_train.txt
+.. literalinclude:: src/sentimental_train.txt
 
 The corresponding data provider can be found in the path below:
 
-.. literalinclude:: ../../../doc_cn/ui/data_provider/sentimental_provider.py
+.. literalinclude:: src/sentimental_provider.py
 
 This data provider for sequential model is a little more complex than that
 for MINST dataset.
@@ -141,7 +143,7 @@ initialized. The :code:`on_init` function has the following parameters:
 To pass these parameters into DataProvider, the following lines should be added
 into trainer configuration file.
 
-.. literalinclude:: ../../../doc_cn/ui/data_provider/sentimental_config.py
+.. literalinclude:: src/sentimental_config.py
 
 The definition is basically same as MNIST example, except:
 * Load dictionary in this configuration

doc/api/index_cn.rst

+API
+===
+
+DataProvider API
+----------------
+
+..  toctree::
+    :maxdepth: 1
+
+    data_provider/dataprovider_cn.rst
+    data_provider/pydataprovider2_cn.rst
+
+..  _api_trainer_config:
+
+Model Config API
+----------------
+
+..  toctree::
+    :maxdepth: 1
+
+    trainer_config_helpers/optimizers.rst
+    trainer_config_helpers/data_sources.rst
+    trainer_config_helpers/layers.rst
+    trainer_config_helpers/activations.rst 
+    trainer_config_helpers/poolings.rst
+    trainer_config_helpers/networks.rst
+    trainer_config_helpers/evaluators.rst
+    trainer_config_helpers/attrs.rst
+
+
+Applications API
+----------------
+
+..  toctree::
+    :maxdepth: 1
+
+    predict/swig_py_paddle_cn.rst

doc/api/index_en.rst

@@ -7,7 +7,7 @@ DataProvider API
 ..  toctree::
     :maxdepth: 1
 
-    data_provider/index_en.rst
+    data_provider/dataprovider_en.rst
     data_provider/pydataprovider2_en.rst
 
 ..  _api_trainer_config:

doc_cn/ui/predict/swig_py_paddle.rst → doc/api/predict/swig_py_paddle_cn.rst

@@ -34,7 +34,7 @@ PaddlePaddle使用swig对常用的预测接口进行了封装，通过编译会
 
 如下是一段使用mnist model来实现手写识别的预测代码。完整的代码见 ``src_root/doc/ui/predict/predict_sample.py`` 。mnist model可以通过 ``src_root\demo\mnist`` 目录下的demo训练出来。
 
-..  literalinclude:: ../../../doc/ui/predict/predict_sample.py
+..  literalinclude:: src/predict_sample.py
     :language: python
     :lines: 15-18,121-136
 

doc/api/predict/swig_py_paddle_en.rst

@@ -13,7 +13,7 @@ Here is a sample python script that shows the typical prediction process for the
 MNIST classification problem. A complete sample code could be found at
 :code:`src_root/doc/ui/predict/predict_sample.py`.
 
-..  literalinclude:: ./predict_sample.py
+..  literalinclude:: src/predict_sample.py
     :language: python
     :lines: 15-18,90-100,101-104
 
@@ -23,7 +23,7 @@ python's :code:`help()` function. Let's walk through the above python script:
 
 * At the beginning, use :code:`swig_paddle.initPaddle()` to initialize
   PaddlePaddle with command line arguments, for more about command line arguments
-  see `Command Line Arguments <../cmd_argument/detail_introduction.html>`_.
+  see :ref:`cmd_detail_introduction` .
 * Parse the configuration file that is used in training with :code:`parse_config()`.
   Because data to predict with always have no label, and output of prediction work
   normally is the output layer rather than the cost layer, so you should modify
@@ -36,7 +36,7 @@ python's :code:`help()` function. Let's walk through the above python script:
     - Note: As swig_paddle can only accept C++ matrices, we offer a utility
       class DataProviderConverter that can accept the same input data with
       PyDataProvider2, for more information please refer to document
-      of `PyDataProvider2 <../data_provider/pydataprovider2.html>`_.
+      of :ref:`api_pydataprovider2` .
 * Do the prediction with :code:`forwardTest()`, which takes the converted
   input data and outputs the activations of the output layer.
 

doc/api/trainer_config_helpers/layers.rst

+..  _api_trainer_config_helpers_layers:
+
 ======
 Layers
 ======

doc_cn/conf.py.in → doc/conf.py.cn.in

@@ -62,7 +62,7 @@ source_suffix = ['.rst', '.md', '.Rmd']
 source_encoding = 'utf-8'
 
 # The master toctree document.
-master_doc = 'index'
+master_doc = 'index_cn'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.
@@ -79,7 +79,7 @@ language = 'zh_CN'
 
 # List of patterns, relative to source directory, that match files and
 # directories to ignore when looking for source files.
-exclude_patterns = ['_build']
+exclude_patterns = ['_build', '**/*_en*', '*_en*']
 
 # The reST default role (used for this markup: `text`) to use for all
 # documents.

doc/conf.py.in → doc/conf.py.en.in

@@ -63,7 +63,7 @@ source_suffix = ['.rst', '.md', '.Rmd']
 source_encoding = 'utf-8'
 
 # The master toctree document.
-master_doc = 'index'
+master_doc = 'index_en'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.
@@ -80,7 +80,7 @@ language = None
 
 # List of patterns, relative to source directory, that match files and
 # directories to ignore when looking for source files.
-exclude_patterns = ['_build']
+exclude_patterns = ['_build', '**/*_cn*', '*_cn*']
 
 # The reST default role (used for this markup: `text`) to use for all
 # documents.
@@ -144,6 +144,6 @@ def setup(app):
     # no c++ API for now
     app.add_config_value('recommonmark_config', {
             'url_resolver': lambda url: github_doc_root + url,
-	    'enable_eval_rst': True,
+        'enable_eval_rst': True,
             }, True)
     app.add_transform(AutoStructify)

doc_cn/faq/index.rst → doc/faq/index_cn.rst

 ####################
-PaddlePaddle常见问题
+FAQ
 ####################
 
 ..  contents::
@@ -33,10 +33,9 @@ PyDataProvider使用的是异步加载，同时在内存里直接随即选取数
 个内存池实际上决定了shuffle的粒度。所以，如果将这个内存池减小，又要保证数据是随机的，
 那么最好将数据文件在每次读取之前做一次shuffle。可能的代码为
 
-..  literalinclude:: reduce_min_pool_size.py
+..  literalinclude:: src/reduce_min_pool_size.py
 
-这样做可以极大的减少内存占用，并且可能会加速训练过程，详细文档参考 `这里
-<../ui/data_provider/pydataprovider2.html#provider>`_ 。
+这样做可以极大的减少内存占用，并且可能会加速训练过程，详细文档参考 `这里 <../ui/data_provider/pydataprovider2.html#provider>`_ 。
 
 神经元激活内存
 ++++++++++++++
@@ -76,7 +75,7 @@ PaddlePaddle支持非常多的优化算法(Optimizer)，不同的优化算法需
 使用 :code:`pydataprovider`时，可以减少缓存池的大小，同时设置内存缓存功能，即可以极大的加速数据载入流程。
 :code:`DataProvider` 缓存池的减小，和之前减小通过减小缓存池来减小内存占用的原理一致。
 
-..  literalinclude:: reduce_min_pool_size.py
+..  literalinclude:: src/reduce_min_pool_size.py
 
 同时 :code:`@provider` 接口有一个 :code:`cache` 参数来控制缓存方法，将其设置成 :code:`CacheType.CACHE_PASS_IN_MEM` 的话，会将第一个 :code:`pass` (过完所有训练数据即为一个pass)生成的数据缓存在内存里，在之后的 :code:`pass` 中，不会再从 :code:`python` 端读取数据，而是直接从内存的缓存里读取数据。这也会极大减少数据读入的耗时。
 
@@ -90,11 +89,11 @@ PaddlePaddle支持Sparse的训练，sparse训练需要训练特征是 :code:`spa
 
 使用一个词前两个词和后两个词，来预测这个中间的词。这个任务的DataProvider为\:
 
-..  literalinclude:: word2vec_dataprovider.py
+..  literalinclude:: src/word2vec_dataprovider.py
 
 这个任务的配置为\:
 
-..  literalinclude:: word2vec_config.py
+..  literalinclude:: src/word2vec_config.py
 
 更多关于sparse训练的内容请参考 `sparse训练的文档 <TBD>`_
 
@@ -158,7 +157,7 @@ PaddlePaddle的参数使用名字 :code:`name` 作为参数的ID，相同名字
 这里 :code:`hidden_a` 和 :code:`hidden_b` 使用了同样的parameter和bias。并且softmax层的两个输入也使用了同样的参数 :code:`softmax_param`。
 
 7. *-cp27mu-linux_x86_64.whl is not a supported wheel on this platform.
------------------------------------------------------------------------
+---------------------------------------------------------------------------
 
 出现这个问题的主要原因是，系统编译wheel包的时候，使用的 :code:`wheel` 包是最新的，
 而系统中的 :code:`pip` 包比较老。具体的解决方法是，更新 :code:`pip` 包并重新编译PaddlePaddle。
@@ -220,7 +219,7 @@ PaddlePaddle的参数使用名字 :code:`name` 作为参数的ID，相同名字
 
 
 10. CMake源码编译, 找到的PythonLibs和PythonInterp版本不一致
-----------------------------------------------------------
+----------------------------------------------------------------
 
 这是目前CMake寻找Python的逻辑存在缺陷，如果系统安装了多个Python版本，CMake找到的Python库和Python解释器版本可能有不一致现象，导致编译PaddlePaddle失败。正确的解决方法是，
 用户强制指定特定的Python版本，具体操作如下：
@@ -231,7 +230,7 @@ PaddlePaddle的参数使用名字 :code:`name` 作为参数的ID，相同名字
 
 用户需要指定本机上Python的路径：``<exc_path>``, ``<lib_path>``, ``<inc_path>``
 
-10. A protocol message was rejected because it was too big
+10. A protocol message was rejected because it was too big
 ----------------------------------------------------------
 
 如果在训练NLP相关模型时，出现以下错误：

doc_cn/introduction/index.rst → doc/getstarted/basic_usage/index_cn.rst

@@ -58,6 +58,7 @@ PaddlePaddle是源于百度的一个深度学习平台。这份简短的介绍
     cost = regression_cost(input= ȳ, label=y)
     outputs(cost)
 
+
 这段简短的配置展示了PaddlePaddle的基本用法：
 
 - 第一部分定义了数据输入。一般情况下，PaddlePaddle先从一个文件列表里获得数据文件地址，然后交给用户自定义的函数（例如上面的 `process`函数）进行读入和预处理从而得到真实输入。本文中由于输入数据是随机生成的不需要读输入文件，所以放一个空列表（`empty.list`）即可。
@@ -65,10 +66,10 @@ PaddlePaddle是源于百度的一个深度学习平台。这份简短的介绍
 - 第二部分主要是选择学习算法，它定义了模型参数改变的规则。PaddlePaddle提供了很多优秀的学习算法，这里使用一个基于momentum的随机梯度下降(SGD)算法，该算法每批量(batch)读取12个采样数据进行随机梯度计算来更新更新。
 
 - 最后一部分是神经网络的配置。由于PaddlePaddle已经实现了丰富的网络层，所以很多时候你需要做的只是定义正确的网络层并把它们连接起来。这里使用了三种网络单元：
-	
-	- **数据层**：数据层 `data_layer` 是神经网络的入口，它读入数据并将它们传输到接下来的网络层。这里数据层有两个，分别对应于变量 `x` 和 `y`。
-	- **全连接层**：全连接层 `fc_layer` 是基础的计算单元，这里利用它建模变量之间的线性关系。计算单元是神经网络的核心，PaddlePaddle支持大量的计算单元和任意深度的网络连接，从而可以拟合任意的函数来学习复杂的数据关系。
-	- **回归误差代价层**：回归误差代价层 `regression_cost` 是众多误差代价函数层的一种，它们在训练过程作为网络的出口，用来计算模型的误差，是模型参数优化的目标函数。
+    
+    - **数据层**：数据层 `data_layer` 是神经网络的入口，它读入数据并将它们传输到接下来的网络层。这里数据层有两个，分别对应于变量 `x` 和 `y`。
+    - **全连接层**：全连接层 `fc_layer` 是基础的计算单元，这里利用它建模变量之间的线性关系。计算单元是神经网络的核心，PaddlePaddle支持大量的计算单元和任意深度的网络连接，从而可以拟合任意的函数来学习复杂的数据关系。
+    - **回归误差代价层**：回归误差代价层 `regression_cost` 是众多误差代价函数层的一种，它们在训练过程作为网络的出口，用来计算模型的误差，是模型参数优化的目标函数。
 
 定义了网络结构并保存为 `trainer_config.py` 之后，运行以下训练命令：
 
@@ -99,8 +100,8 @@ PaddlePaddle将每个模型参数作为一个numpy数组单独存为一个文件
     # w=1.999743, b=0.300137
 
 .. image:: ./parameters.png
-	 :align: center
-	 :scale: 80 %
+     :align: center
+     :scale: 80 %
 
 从图中可以看到，虽然 `w` 和 `b` 都使用随机值初始化，但在起初的几轮训练中它们都在快速逼近真实值，并且后续仍在不断改进，使得最终得到的模型几乎与真实模型一致。
 

doc/getstarted/basic_usage/index_en.rst

@@ -99,11 +99,3 @@ In PaddlePaddle, training is just to get a collection of model parameters, which
 Although starts from a random guess, you can see that value of ``w`` changes quickly towards 2 and ``b`` changes quickly towards 0.3. In the end, the predicted line is almost identical with real answer.
 
 There, you have recovered the underlying pattern between ``X`` and ``Y`` only from observed data.
-
-
-5. Where to Go from Here
--------------------------
-
-- `Install and Build <../build_and_install/index.html>`_
-- `Tutorials <../demo/quick_start/index_en.html>`_
-- `Example and Demo <../demo/index.html>`_

doc/getstarted/build_and_install/build_from_source_en.md

@@ -14,6 +14,13 @@ cd paddle
 git submodule update --init --recursive
 ```
 
+If you already have a local PaddlePaddle repo and have not initialized the submodule, your local submodule folder will be empty. You can simply run the last line of the above codes in your PaddlePaddle home directory to initialize your submodule folder.
+
+If you have already initialized your submodule and you would like to sync with the upstream submodule repo, you can run the following command
+```
+git submodule update --remote
+```
+
 ## <span id="requirements">Requirements</span>
 
 To compile the source code, your computer must be equipped with the following dependencies.
@@ -42,10 +49,8 @@ PaddlePaddle supports some build options. To enable it, first you need to instal
 <tbody>
 <tr><td class="left">WITH_GPU</td><td class="left">Compile with GPU mode.</td></tr>
 <tr><td class="left">WITH_DOUBLE</td><td class="left">Compile with double precision floating-point, default: single precision.</td></tr>
-<tr><td class="left">WITH_GLOG</td><td class="left">Compile with glog. If not found, default: an internal log implementation.</td></tr>
-<tr><td class="left">WITH_GFLAGS</td><td class="left">Compile with gflags. If not found, default: an internal flag implementation.</td></tr>
 <tr><td class="left">WITH_TESTING</td><td class="left">Compile with gtest for PaddlePaddle's unit testing.</td></tr>
-<tr><td class="left">WITH_DOC</td><td class="left">	Compile to generate PaddlePaddle's docs, default: disabled (OFF).</td></tr>
+<tr><td class="left">WITH_DOC</td><td class="left">    Compile to generate PaddlePaddle's docs, default: disabled (OFF).</td></tr>
 <tr><td class="left">WITH_SWIG_PY</td><td class="left">Compile with python predict API, default: disabled (OFF).</td></tr>
 <tr><td class="left">WITH_STYLE_CHECK</td><td class="left">Compile with code style check, default: enabled (ON).</td></tr>
 </tbody>
@@ -79,7 +84,7 @@ As a simple example, consider the following:
 
     ```bash
     pip install 'sphinx>=1.4.0'
-    pip install sphinx_rtd_theme breathe recommonmark
+    pip install sphinx_rtd_theme recommonmark
 
     # install doxygen on Ubuntu
     sudo apt-get install doxygen 

doc_cn/build_and_install/cmake/compile_options.rst → doc/getstarted/build_and_install/cmake/build_from_source_cn.rst

-PaddlePaddle的编译选项
-======================
-
-PaddlePaddle的编译选项，包括生成CPU/GPU二进制文件、链接何种BLAS库等。用户可在调用cmake的时候设置它们，详细的cmake使用方法可以参考 `官方文档 <https://cmake.org/cmake-tutorial>`_ 。
-
-Bool型的编译选项
-----------------
-用户可在cmake的命令行中，通过使用 ``-D`` 命令设置该类编译选项，例如
-
-..  code-block:: bash
-
-    cmake .. -DWITH_GPU=OFF
-
-..  csv-table:: Bool型的编译选项
-    :widths: 1, 7, 2
-    :file: compile_options.csv
-
-BLAS/CUDA/Cudnn的编译选项
---------------------------
-BLAS
-+++++
-
-PaddlePaddle支持以下任意一种BLAS库：`MKL <https://software.intel.com/en-us/intel-mkl>`_ ，`ATLAS <http://math-atlas.sourceforge.net/>`_ ，`OpenBlAS <http://www.openblas.net/>`_ 和 `REFERENCE BLAS <http://www.netlib.org/blas/>`_ 。
-
-..  csv-table:: BLAS路径相关的编译选项
-    :widths: 1, 2, 7
-    :file: cblas_settings.csv
-
-CUDA/Cudnn
-+++++++++++
-
-PaddlePaddle可以使用cudnn v2之后的任何一个版本来编译运行，但尽量请保持编译和运行使用的cudnn是同一个版本。 我们推荐使用最新版本的cudnn v5.1。
-
-编译选项的设置
-++++++++++++++
-
-PaddePaddle通过编译时指定路径来实现引用各种BLAS/CUDA/Cudnn库。cmake编译时，首先在系统路径(/usr/lib\:/usr/local/lib)中搜索这几个库，同时也会读取相关路径变量来进行搜索。 通过使用 ``-D`` 命令可以设置，例如 
-
-..  code-block:: bash
-
-    cmake .. -DMKL_ROOT=/opt/mkl/ -DCUDNN_ROOT=/opt/cudnnv5
-
+PaddlePaddle的编译选项
+======================
+
+PaddlePaddle的编译选项，包括生成CPU/GPU二进制文件、链接何种BLAS库等。用户可在调用cmake的时候设置它们，详细的cmake使用方法可以参考 `官方文档 <https://cmake.org/cmake-tutorial>`_ 。
+
+Bool型的编译选项
+----------------
+用户可在cmake的命令行中，通过使用 ``-D`` 命令设置该类编译选项，例如
+
+..  code-block:: bash
+
+    cmake .. -DWITH_GPU=OFF
+
+..  csv-table:: Bool型的编译选项
+    :widths: 1, 7, 2
+    :file: compile_options.csv
+
+BLAS/CUDA/Cudnn的编译选项
+--------------------------
+BLAS
++++++
+
+PaddlePaddle支持以下任意一种BLAS库：`MKL <https://software.intel.com/en-us/intel-mkl>`_ ，`ATLAS <http://math-atlas.sourceforge.net/>`_ ，`OpenBlAS <http://www.openblas.net/>`_ 和 `REFERENCE BLAS <http://www.netlib.org/blas/>`_ 。
+
+..  csv-table:: BLAS路径相关的编译选项
+    :widths: 1, 2, 7
+    :file: cblas_settings.csv
+
+CUDA/Cudnn
++++++++++++
+
+PaddlePaddle可以使用cudnn v2之后的任何一个版本来编译运行，但尽量请保持编译和运行使用的cudnn是同一个版本。 我们推荐使用最新版本的cudnn v5.1。
+
+编译选项的设置
+++++++++++++++
+
+PaddePaddle通过编译时指定路径来实现引用各种BLAS/CUDA/Cudnn库。cmake编译时，首先在系统路径(/usr/lib\:/usr/local/lib)中搜索这几个库，同时也会读取相关路径变量来进行搜索。 通过使用 ``-D`` 命令可以设置，例如 
+
+..  code-block:: bash
+
+    cmake .. -DMKL_ROOT=/opt/mkl/ -DCUDNN_ROOT=/opt/cudnnv5
+
 注意：这几个编译选项的设置，只在第一次cmake的时候有效。如果之后想要重新设置，推荐清理整个编译目录（``rm -rf``）后，再指定。
\ No newline at end of file

doc_cn/build_and_install/cmake/compile_options.csv → doc/getstarted/build_and_install/cmake/compile_options.csv

-选项,说明,默认值
-WITH_GPU,是否支持GPU。,取决于是否寻找到CUDA工具链
-WITH_DOUBLE,是否使用双精度浮点数。,否
-WITH_DSO,是否运行时动态加载CUDA动态库，而非静态加载CUDA动态库。,是
-WITH_AVX,是否编译含有AVX指令集的PaddlePaddle二进制文件,是
-WITH_PYTHON,是否内嵌PYTHON解释器。方便今后的嵌入式移植工作。,是
-WITH_STYLE_CHECK,是否编译时进行代码风格检查,是
-WITH_RDMA,是否开启RDMA,否
-WITH_GLOG,是否开启GLOG。如果不开启，则会使用一个简化版的日志，同时方便今后的嵌入式移植工作。,取决于是否寻找到GLOG
-WITH_GFLAGS,是否使用GFLAGS。如果不开启，则会使用一个简化版的命令行参数解析器，同时方便今后的嵌入式移植工作。,取决于是否寻找到GFLAGS
-WITH_TIMER,是否开启计时功能。如果开启会导致运行略慢，打印的日志变多，但是方便调试和测Benchmark,否
-WITH_TESTING,是否开启单元测试,取决于是否寻找到GTEST
-WITH_DOC,是否编译中英文文档,否
+选项,说明,默认值
+WITH_GPU,是否支持GPU。,取决于是否寻找到CUDA工具链
+WITH_DOUBLE,是否使用双精度浮点数。,否
+WITH_DSO,是否运行时动态加载CUDA动态库，而非静态加载CUDA动态库。,是
+WITH_AVX,是否编译含有AVX指令集的PaddlePaddle二进制文件,是
+WITH_PYTHON,是否内嵌PYTHON解释器。方便今后的嵌入式移植工作。,是
+WITH_STYLE_CHECK,是否编译时进行代码风格检查,是
+WITH_RDMA,是否开启RDMA,否
+WITH_TIMER,是否开启计时功能。如果开启会导致运行略慢，打印的日志变多，但是方便调试和测Benchmark,否
+WITH_TESTING,是否开启单元测试,取决于是否寻找到GTEST
+WITH_DOC,是否编译中英文文档,否
 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,取决于是否寻找到SWIG
\ No newline at end of file

doc_cn/build_and_install/install/docker_install.rst → doc/getstarted/build_and_install/docker_install_cn.rst

@@ -111,7 +111,24 @@ cuda相关的Driver和设备映射进container中，脚本类似于
 
 简单的含有ssh的Dockerfile如下：
 
-..  literalinclude:: paddle_ssh.Dockerfile
+..  code-block:: bash
+
+    FROM paddledev/paddle:cpu-latest
+
+    MAINTAINER PaddlePaddle dev team <paddle-dev@baidu.com>
+
+    RUN apt-get update
+    RUN apt-get install -y openssh-server
+    RUN mkdir /var/run/sshd
+    RUN echo 'root:root' | chpasswd
+
+    RUN sed -ri 's/^PermitRootLogin\s+.*/PermitRootLogin yes/' /etc/ssh/sshd_config
+    RUN sed -ri 's/UsePAM yes/#UsePAM yes/g' /etc/ssh/sshd_config
+
+    EXPOSE 22
+
+    CMD    ["/usr/sbin/sshd", "-D"]
+
 
 使用该Dockerfile构建出镜像，然后运行这个container即可。相关命令为\:
 

doc/getstarted/build_and_install/docker_install_en.rst

@@ -17,7 +17,7 @@ CPU-only one and a CUDA GPU one.  We do so by configuring
 `dockerhub.com <https://hub.docker.com/r/paddledev/paddle/>`_
 automatically runs the following commands:
 
-.. code-block:: base
+.. code-block:: bash
 
    docker build -t paddle:cpu -f paddle/scripts/docker/Dockerfile .
    docker build -t paddle:gpu -f paddle/scripts/docker/Dockerfile.gpu .
@@ -104,3 +104,78 @@ container:
 
 Then we can direct our Web browser to the HTML version of source code
 at http://localhost:8088/paddle/
+
+
+Development Using Docker
+------------------------
+
+Develpers can work on PaddlePaddle using Docker.  This allows
+developers to work on different platforms -- Linux, Mac OS X, and
+Windows -- in a consistent way.
+
+The general development workflow with Docker and Bazel is as follows:
+
+1. Get the source code of Paddle:
+
+   .. code-block:: bash
+
+      git clone --recursive https://github.com/paddlepaddle/paddle
+
+
+2. Build a development Docker image :code:`paddle:dev` from the source
+   code.  This image contains all the development tools and
+   dependencies of PaddlePaddle.
+
+
+   .. code-block:: bash
+
+      cd paddle
+      docker build -t paddle:dev -f paddle/scripts/docker/Dockerfile .
+
+
+3. Run the image as a container and mounting local source code
+   directory into the container.  This allows us to change the code on
+   the host and build it within the container.
+
+   .. code-block:: bash
+
+      docker run       \
+       -d              \
+       --name paddle   \
+       -p 2022:22      \
+       -v $PWD:/paddle \
+       -v $HOME/.cache/bazel:/root/.cache/bazel \
+       paddle:dev
+
+   where :code:`-d` makes the container running in background,
+   :code:`--name paddle` allows us to run a nginx container to serve
+   documents in this container, :code:`-p 2022:22` allows us to SSH
+   into this container, :code:`-v $PWD:/paddle` shares the source code
+   on the host with the container, :code:`-v
+   $HOME/.cache/bazel:/root/.cache/bazel` shares Bazel cache on the
+   host with the container.
+
+4. SSH into the container:
+
+   .. code-block:: bash
+
+      ssh root@localhost -p 2022
+
+5. We can edit the source code in the container or on this host.  Then
+   we can build using cmake
+
+   .. code-block:: bash
+
+      cd /paddle # where paddle source code has been mounted into the container
+      mkdir -p build
+      cd build
+      cmake -DWITH_TESTING=ON ..
+      make -j `nproc`
+      CTEST_OUTPUT_ON_FAILURE=1 ctest
+
+   or Bazel in the container:
+
+   .. code-block:: bash
+
+      cd /paddle
+      bazel test ...

doc_cn/build_and_install/index.rst → doc/getstarted/build_and_install/index_cn.rst

@@ -9,8 +9,8 @@ PaddlePaddle提供数个预编译的二进制来进行安装，包括Docker镜
 .. toctree::
    :maxdepth: 1
    
-   install/docker_install.rst 
-   install/ubuntu_install.rst
+   docker_install_cn.rst 
+   ubuntu_install_cn.rst
 
 
 
@@ -19,9 +19,9 @@ PaddlePaddle提供数个预编译的二进制来进行安装，包括Docker镜
 
 ..  warning::
 
-	编译选项主要推荐高级用户查看，普通用户请走安装流程。
+    编译选项主要推荐高级用户查看，普通用户请走安装流程。
 
-.. toctree::
-   :maxdepth: 1
+..  toctree::
+    :maxdepth: 1
 
-   cmake/index.rst
+    cmake/build_from_source_cn.rst
\ No newline at end of file

doc_cn/build_and_install/install/ubuntu_install.rst → doc/getstarted/build_and_install/ubuntu_install_cn.rst

@@ -38,7 +38,18 @@ PaddlePaddle提供了ubuntu 14.04 deb安装包。
 
 安装完成后，可以使用命令 :code:`paddle version` 查看安装后的paddle 版本:
 
-..  literalinclude:: paddle_version.txt
+..  code-block:: shell
+
+    PaddlePaddle 0.8.0b1, compiled with
+        with_avx: ON
+        with_gpu: OFF
+        with_double: OFF
+        with_python: ON
+        with_rdma: OFF
+        with_metric_learning:
+        with_timer: OFF
+        with_predict_sdk:
+
 
 可能遇到的问题
 --------------
@@ -48,9 +59,9 @@ libcudart.so/libcudnn.so找不到
 
 安装完成后，运行 :code:`paddle train` 报错\:
 
-.. 	code-block:: shell
+..  code-block:: shell
 
-	  0831 12:36:04.151525  1085 hl_dso_loader.cc:70] Check failed: nullptr != *dso_handle For Gpu version of PaddlePaddle, it couldn't find CUDA library: libcudart.so Please make sure you already specify its path.Note: for training data on Cpu using Gpu version of PaddlePaddle,you must specify libcudart.so via LD_LIBRARY_PATH.
+      0831 12:36:04.151525  1085 hl_dso_loader.cc:70] Check failed: nullptr != *dso_handle For Gpu version of PaddlePaddle, it couldn't find CUDA library: libcudart.so Please make sure you already specify its path.Note: for training data on Cpu using Gpu version of PaddlePaddle,you must specify libcudart.so via LD_LIBRARY_PATH.
 
 原因是未设置cuda运行时环境变量。 如果使用GPU版本的PaddlePaddle，请安装CUDA 7.5 和CUDNN 5到本地环境中，并设置：
 

doc/getstarted/index_cn.rst

+GET STARTED
+============
+
+..  toctree::
+  :maxdepth: 2
+
+  build_and_install/index_cn.rst
+  basic_usage/index_cn.rst

doc_cn/cluster/k8s/distributed_training_on_kubernetes.md → doc/howto/cluster/k8s/distributed_training_on_k8s_cn.md

@@ -306,4 +306,4 @@ I1116 09:10:18.019069    50 ParameterClient2.cpp:122] pserver 2 192.168.223.143:
 I1116 09:10:18.019492    50 ParameterClient2.cpp:122] pserver 3 192.168.223.143:7165
 I1116 09:10:18.019716    50 ParameterClient2.cpp:122] pserver 4 192.168.129.71:7164
 I1116 09:10:18.019836    50 ParameterClient2.cpp:122] pserver 5 192.168.129.71:7165
-```
\ No newline at end of file
+```

doc_cn/cluster/k8s/job.yaml → doc/howto/cluster/k8s/job.yaml

@@ -40,4 +40,4 @@ spec:
         - name: jobpath
           mountPath: /home/jobpath       
       restartPolicy: Never
-    
\ No newline at end of file
+    

doc_cn/cluster/k8s/start_paddle.py → doc/howto/cluster/k8s/start_paddle.py

@@ -19,7 +19,6 @@ import socket
 import os
 import argparse
 
-
 # configuration for cluster
 API = "/api/v1/namespaces/"
 JOBSELECTOR = "labelSelector=job-name="
@@ -145,8 +144,8 @@ def startPaddle(idMap={}, train_args_dict=None):
 
 
 if __name__ == '__main__':
-    parser = argparse.ArgumentParser(prog="start_paddle.py",
-                                     description='simple tool for k8s')
+    parser = argparse.ArgumentParser(
+        prog="start_paddle.py", description='simple tool for k8s')
     args, train_args_list = parser.parse_known_args()
     train_args = refine_unknown_args(train_args_list)
     train_args_dict = dict(zip(train_args[:-1:2], train_args[1::2]))

doc/howto/cmd_parameter/detail_introduction_en.md

+```eval_rst
+..  _cmd_detail_introduction:
+```
+
 # Detail Description
 
 ## Common

doc/howto/cmd_parameter/index_en.md

 ```eval_rst
-..  _cmd_line_index_en:
+..  _cmd_line_index:
 ```
 # How to Set Command-line Parameters
 

doc_cn/concepts/use_concepts.rst → doc/howto/concepts/use_concepts_cn.rst

@@ -8,29 +8,29 @@ PaddlePaddle是一个深度学习框架，支持单机模式和多机模式。
 
 本文首先介绍trainer进程中的一些使用概念，然后介绍pserver进程中概念。
 
-..	contents::
+..    contents::
 
 系统框图
 ========
 
 下图描述了用户使用框图，PaddlePaddle的trainer进程里内嵌了Python解释器，trainer进程可以利用这个解释器执行Python脚本，Python脚本里定义了模型配置、训练算法、以及数据读取函数。其中，数据读取程序往往定义在一个单独Python脚本文件里，被称为数据提供器（DataProvider），通常是一个Python函数。模型配置、训练算法通常定义在另一单独Python文件中, 称为训练配置文件。下面将分别介绍这两部分。
 
-..	graphviz:: 
-
-	digraph pp_process {
-		rankdir=LR;
-		config_file [label="用户神经网络配置"];
-		subgraph cluster_pp {
-			style=filled;
-			color=lightgrey;
-			node [style=filled, color=white, shape=box];
-			label = "PaddlePaddle C++";
-			py [label="Python解释器"];
-		}
-		data_provider [label="用户数据解析"];
-		config_file -> py;
-		py -> data_provider [dir="back"];
-	}
+..    graphviz:: 
+
+    digraph pp_process {
+        rankdir=LR;
+        config_file [label="用户神经网络配置"];
+        subgraph cluster_pp {
+            style=filled;
+            color=lightgrey;
+            node [style=filled, color=white, shape=box];
+            label = "PaddlePaddle C++";
+            py [label="Python解释器"];
+        }
+        data_provider [label="用户数据解析"];
+        config_file -> py;
+        py -> data_provider [dir="back"];
+    }
 
 数据提供器
 ==========
@@ -47,7 +47,7 @@ DataProvider是PaddlePaddle系统的数据提供器，将用户的原始数据
 
 一个简单的训练配置文件为：
 
-..  literalinclude:: trainer_config.py
+..  literalinclude:: src/trainer_config.py
     :linenos:
 
 文件开头 ``from paddle.trainer_config_helpers import *`` ，是因为PaddlePaddle配置文件与C++模块通信的最基础协议是protobuf，为了避免用户直接写复杂的protobuf string，我们为用户定以Python接口来配置网络，该Python代码可以生成protobuf包，这就是`trainer_config_helpers`_的作用。因此，在文件的开始，需要import这些函数。 这个包里面包含了模型配置需要的各个模块。
@@ -100,11 +100,11 @@ DataProvider是PaddlePaddle系统的数据提供器，将用户的原始数据
 
 例如，和 ``fc_layer`` 同样功能的 ``mixed_layer`` 是:
 
-..	code-block:: python
+..    code-block:: python
    
-   	data = data_layer(name='data', size=200)
-   	with mixed_layer(size=200) as out:
-   		out += full_matrix_projection(input=data)
+       data = data_layer(name='data', size=200)
+       with mixed_layer(size=200) as out:
+           out += full_matrix_projection(input=data)
 
 PaddlePaddle 可以使用 ``mixed layer`` 配置出非常复杂的网络，甚至可以直接配置一个完整的LSTM。用户可以参考 `mixed_layer`_ 的相关文档进行配置。
 
@@ -114,13 +114,13 @@ PaddlePaddle 可以使用 ``mixed layer`` 配置出非常复杂的网络，甚
 
 PaddlePaddle多机采用经典的 Parameter Server 架构对多个节点的 trainer 进行同步。多机训练的经典拓扑结构如下\:
 
-..	graphviz:: pserver_topology.dot
+..    graphviz:: src/pserver_topology.dot
 
 图中每个灰色方块是一台机器，在每个机器中，先使用命令 ``paddle pserver`` 启动一个pserver进程，并指定端口号，可能的参数是\:
 
-..	code-block:: bash
+..    code-block:: bash
 
-	paddle pserver --port=5000 --num_gradient_servers=4 --tcp_rdma='tcp' --nics='eth0'
+    paddle pserver --port=5000 --num_gradient_servers=4 --tcp_rdma='tcp' --nics='eth0'
 
 * ``--port=5000`` : 指定 pserver 进程端口是 5000 。
 * ``--gradient_servers=4`` : 有四个训练进程(PaddlePaddle 将 trainer 也称作 GradientServer ，因为其为负责提供Gradient) 。
@@ -128,9 +128,9 @@ PaddlePaddle多机采用经典的 Parameter Server 架构对多个节点的 trai
 
 启动之后 pserver 进程之后，需要启动 trainer 训练进程，在各个机器上运行如下命令\:
 
-..	code-block:: bash
+..    code-block:: bash
 
-	paddle train --port=5000 --pservers=192.168.100.101,192.168.100.102,192.168.100.103,192.168.100.104 --config=...
+    paddle train --port=5000 --pservers=192.168.100.101,192.168.100.102,192.168.100.103,192.168.100.104 --config=...
 
 对于简单的多机协同训练使用上述方式即可。另外，pserver/train 通常在高级情况下，还需要设置下面两个参数\：
 

doc/howto/contribute_to_paddle_en.md

 # How to Contribute Code
 
 We sincerely appreciate your contributions. You can use fork and pull request
-workflow to merge your code. 
- 
+workflow to merge your code.
+
 ## Code Requirements
 - Your code must be fully documented by
   [doxygen](http://www.stack.nl/~dimitri/doxygen/) style.
@@ -12,11 +12,11 @@ workflow to merge your code.
 - Pass all unit tests.
 
 The following tutorial guides you into submitting your contibution.
- 
+
 ## [Creating a Fork](https://help.github.com/articles/fork-a-repo/)
- 
+
 Just head over to the GitHub page and click the "Fork" button.
-It's just that simple. 
+It's just that simple.
 
 ## Clone
 
@@ -25,7 +25,7 @@ The **develop** is the main branch, and other user's branches are feature branch
 
 Once you've created a fork, you can use your favorite git client to clone your
 repo or just head straight to the command line:
- 
+
 ```shell
 # Clone your fork to your local machine
 git clone --branch develop https://github.com/USERNAME/Paddle.git
@@ -47,6 +47,22 @@ Then you can start to develop by making a local developement branch
 git checkout -b MY_COOL_STUFF_BRANCH
 ```
 
+## Using `pre-commit` hook
+
+Paddle developers use [pre-commit](http://pre-commit.com/) tool to manage git
+pre-commit hooks. It can help us format source codes (cpp, python), check some
+basic thing before commit (only one EOL for each file, do not add a huge file
+in git). `pre-commit` tests is a part of unit tests in Travis-CI now, every
+PR doesn't fit hook can not be merged into Paddle.
+
+To use [pre-commit](http://pre-commit.com/), you should install it by
+`pip install pre-commit`, and currently, Paddle uses `clang-format` to format
+c/cpp sources. Please make sure clang-format 3.8+ installed.
+
+Then just run `pre-commit install` in your Paddle clone directory. When you
+commit your code, the pre-commit hook will check the local code if there is
+anything not suitable to commit, and so on.
+
 ## Commit
 
 Commit your changes by following command lines:
@@ -83,7 +99,7 @@ git pull --rebase upstream develop
 
 If there are no unique commits locally, git will simply perform a fast-forward.
 However, if you have been making changes (in the vast majority of cases you
-probably shouldn't be), you may have to deal with conflicts. 
+probably shouldn't be), you may have to deal with conflicts.
 
 Now, your local master branch is up-to-date with everything modified upstream.
 

doc/howto/deep_model/index_cn.rst

+How to Configure Deep Models
+============================
+
+..  toctree::
+  :maxdepth: 1
+
+  rnn/recurrent_group_cn.md
+  rnn/hierarchical_layer_cn.rst
+  rnn/hrnn_rnn_api_compare_cn.rst
+  rnn/hrnn_demo_cn.rst

doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst → doc/howto/deep_model/rnn/hrnn_rnn_api_compare_cn.rst

@@ -24,18 +24,18 @@
 
 - 本例中的原始数据一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。这个数据也被单层RNN网络直接使用。
 
-..  literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg
+..  literalinclude:: ../../../../paddle/gserver/tests/Sequence/tour_train_wdseg
     :language: text
 
 
 - 双层序列数据一共有4个样本。 每个样本间用空行分开，整体数据和原始数据完全一样。但于双层序列的LSTM来说，第一个样本同时encode两条数据成两个向量。这四条数据同时处理的句子数量为\ :code:`[2, 3, 2, 3]`\ 。
 
-..  literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg.nest
+..  literalinclude:: ../../../../paddle/gserver/tests/Sequence/tour_train_wdseg.nest
     :language: text
 
 其次，对于两种不同的输入数据类型，不同DataProvider对比如下(`sequenceGen.py <https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/gserver/tests/sequenceGen.py>`_)\：
 
-..  literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py
+..  literalinclude:: ../../../../paddle/gserver/tests/sequenceGen.py
     :language: python
     :lines: 21-39
     :linenos:
@@ -43,10 +43,11 @@
 - 这是普通的单层时间序列的DataProvider代码，其说明如下：
   
   * DataProvider共返回两个数据，分别是words和label。即上述代码中的第19行。
-  - words是原始数据中的每一句话，所对应的词表index数组。它是integer_value_sequence类型的，即整数数组。words即为这个数据中的单层时间序列。
-  - label是原始数据中对于每一句话的分类标签，它是integer_value类型的。
 
-..  literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py
+    - words是原始数据中的每一句话，所对应的词表index数组。它是integer_value_sequence类型的，即整数数组。words即为这个数据中的单层时间序列。
+    - label是原始数据中对于每一句话的分类标签，它是integer_value类型的。
+
+..  literalinclude:: ../../../../paddle/gserver/tests/sequenceGen.py
     :language: python
     :lines: 42-71
     :linenos:
@@ -63,7 +64,7 @@
 
 首先，我们看一下单层RNN的配置。代码中9-15行(高亮部分)即为单层RNN序列的使用代码。这里使用了PaddlePaddle预定义好的RNN处理函数。在这个函数中，RNN对于每一个时间步通过了一个LSTM网络。
 
-..  literalinclude:: ../../../paddle/gserver/tests/sequence_layer_group.conf
+..  literalinclude:: ../../../../paddle/gserver/tests/sequence_layer_group.conf
     :language: python
     :lines: 38-63
     :linenos:
@@ -84,7 +85,7 @@
 
 * 至此，\ :code:`lstm_last`\ 便和单层RNN配置中的\ :code:`lstm_last`\ 具有相同的结果了。
 
-..  literalinclude:: ../../../paddle/gserver/tests/sequence_nest_layer_group.conf
+..  literalinclude:: ../../../../paddle/gserver/tests/sequence_nest_layer_group.conf
     :language: python
     :lines: 38-64
     :linenos:
@@ -106,7 +107,7 @@
 
 - 单层RNN：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。
 
-..  literalinclude:: ../../../paddle/gserver/tests/sequence_rnn.conf
+..  literalinclude:: ../../../../paddle/gserver/tests/sequence_rnn.conf
     :language: python
     :lines: 36-48
 
@@ -115,7 +116,7 @@
   - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。
   - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每个时间步都用了上一个时间步的输出结果”一致。
 
-..  literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn.conf
+..  literalinclude:: ../../../../paddle/gserver/tests/sequence_nest_rnn.conf
     :language: python
     :lines: 39-66
 
@@ -151,14 +152,14 @@
 
 * 单层RNN\:
 
-..  literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py
+..  literalinclude:: ../../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py
     :language: python
     :lines: 42-59
     :linenos:
 
 * 双层RNN\ \:
 
-..  literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py
+..  literalinclude:: ../../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py
     :language: python
     :lines: 41-80
     :linenos:
@@ -181,11 +182,11 @@ Memory
 
 Memory是PaddlePaddle实现RNN时候使用的一个概念。RNN即时间递归神经网络，通常要求时间步之间具有一些依赖性，即当前时间步下的神经网络依赖前一个时间步神经网络中某一个神经元输出。如下图所示。
 
-..  graphviz:: glossary_rnn.dot
+..  graphviz:: src/glossary_rnn.dot
 
 上图中虚线的连接，即是跨越时间步的网络连接。PaddlePaddle在实现RNN的时候，将这种跨越时间步的连接用一个特殊的神经网络单元实现。这个神经网络单元就叫Memory。Memory可以缓存上一个时刻某一个神经元的输出，然后在下一个时间步输入给另一个神经元。使用Memory的RNN实现便如下图所示。
 
-..  graphviz:: glossary_rnn_with_memory.dot
+..  graphviz:: src/glossary_rnn_with_memory.dot
 
 使用这种方式，PaddlePaddle可以比较简单的判断哪些输出是应该跨越时间步的，哪些不是。
 

doc_cn/algorithm/rnn/rnn-tutorial.md → doc/howto/deep_model/rnn/recurrent_group_cn.md

-# Recurrent Group教程
-
-## 概述
-
-序列数据是自然语言处理任务面对的一种主要输入数据类型。
-
-一句话是由词语构成的序列，多句话进一步构成了段落。因此，段落可以看作是一个嵌套的双层的序列，这个序列的每个元素又是一个序列。
-
-双层序列是PaddlePaddle支持的一种非常灵活的数据组织方式，帮助我们更好地描述段落、多轮对话等更为复杂的语言数据。基于双层序列输入，我们可以设计搭建一个灵活的、层次化的RNN，分别从词语和句子级别编码输入数据，同时也能够引入更加复杂的记忆机制，更好地完成一些复杂的语言理解任务。
-
-在PaddlePaddle中，`recurrent_group`是一种任意复杂的RNN单元，用户只需定义RNN在一个时间步内完成的计算，PaddlePaddle负责完成信息和误差在时间序列上的传播。
-
-更进一步，`recurrent_group`同样可以扩展到双层序列的处理上。通过两个嵌套的`recurrent_group`分别定义子句级别和词语级别上需要完成的运算，最终实现一个层次化的复杂RNN。
-
-目前，在PaddlePaddle中，能够对双向序列进行处理的有`recurrent_group`和部分Layer，具体可参考文档：<a href = "hierarchical-layer.html">支持双层序列作为输入的Layer</a>。
- 
-## 相关概念
-
-### 基本原理
-`recurrent_group` 是PaddlePaddle支持的一种任意复杂的RNN单元。使用者只需要关注于设计RNN在一个时间步之内完成的计算，PaddlePaddle负责完成信息和梯度在时间序列上的传播。
-
-PaddlePaddle中，`recurrent_group`的一个简单调用如下：
-
-``` python
-recurrent_group(step, input, reverse)
-```
-- step：一个可调用的函数，定义一个时间步之内RNN单元完成的计算
-- input：输入，必须是一个单层序列，或者一个双层序列
-- reverse：是否以逆序处理输入序列
- 
-使用`recurrent_group`的核心是设计step函数的计算逻辑。step函数内部可以自由组合PaddlePaddle支持的各种layer，完成任意的运算逻辑。`recurrent_group` 的输入（即input）会成为step函数的输入，由于step 函数只关注于RNN一个时间步之内的计算，在这里`recurrent_group`替我们完成了原始输入数据的拆分。
-
-### 输入
-`recurrent_group`处理的输入序列主要分为以下三种类型：
- 
-- **数据输入**：一个双层序列进入`recurrent_group`会被拆解为一个单层序列，一个单层序列进入`recurrent_group`会被拆解为非序列，然后交给step函数，这一过程对用户是完全透明的。可以有以下两种：1）通过data_layer拿到的用户输入；2）其它layer的输出。
-		
-- **只读Memory输入**：`StaticInput` 定义了一个只读的Memory，由`StaticInput`指定的输入不会被`recurrent_group`拆解，`recurrent_group` 循环展开的每个时间步总是能够引用所有输入，可以是一个非序列，或者一个单层序列。
-	  
-- **序列生成任务的输入**：`GeneratedInput`只用于在序列生成任务中指定输入数据。
-
-### 输入示例
-
-序列生成任务大多遵循encoder-decoer架构，encoder和decoder可以是能够处理序列的任意神经网络单元，而RNN是最流行的选择。
-
-给定encoder输出和当前词，decoder每次预测产生下一个最可能的词语。在这种结构中，decoder接受两个输入：
-    
-- 要生成的目标序列：是decoder的数据输入，也是decoder循环展开的依据，`recurrent_group`会对这类输入进行拆解。
-
-- encoder输出，可以是一个非序列，或者一个单层序列：是一个unbounded memory，decoder循环展开的每一个时间步会引用全部结果，不应该被拆解，这种类型的输入必须通过`StaticInput`指定。关于Unbounded Memory的更多讨论请参考论文 [Neural Turning Machine](https://arxiv.org/abs/1410.5401)。
-		
-在序列生成任务中，decoder RNN总是引用上一时刻预测出的词的词向量，作为当前时刻输入。`GeneratedInput`自动完成这一过程。
-		 
-### 输出
-`step`函数必须返回一个或多个Layer的输出，这个Layer的输出会作为整个`recurrent_group` 最终的输出结果。在输出的过程中，`recurrent_group` 会将每个时间步的输出拼接，这个过程对用户也是透明的。
-
-### memory
-memory只能在`recurrent_group`中定义和使用。memory不能独立存在，必须指向一个PaddlePaddle定义的Layer。引用memory得到这layer上一时刻输出，因此，可以将memory理解为一个时延操作。
-
-可以显示地指定一个layer的输出用于初始化memory。不指定时，memory默认初始化为0。
-
-## 双层RNN介绍
-`recurrent_group`帮助我们完成对输入序列的拆分，对输出的合并，以及计算逻辑在序列上的循环展开。
-
-利用这种特性，两个嵌套的`recurrent_group`能够处理双层序列，实现词语和句子两个级别的双层RNN结构。
-
-- 单层（word-level）RNN：每个状态（state）对应一个词（word）。
-- 双层（sequence-level）RNN：一个双层RNN由多个单层RNN组成，每个单层RNN（即双层RNN的每个状态）对应一个子句（subseq）。
-
-为了描述方便，下文以NLP任务为例，将含有子句（subseq）的段落定义为一个双层序列，将含有词语的句子定义为一个单层序列，那么0层序列即为一个词语。
-
-## 双层RNN的使用
-
-### 训练流程的使用方法
-使用 `recurrent_group`需要遵循以下约定：
- 
-- **单进单出**：输入和输出都是单层序列。
-  - 如果有多个输入，不同输入序列含有的词语数必须严格相等。
-  - 输出一个单层序列，输出序列的词语数和输入序列一致。
-  - memory：在step函数中定义 memory指向一个layer，通过引用memory得到这个layer上一个时刻输出，形成recurrent 连接。memory的is_seq参数必须为false。如果没有定义memory，每个时间步之内的运算是独立的。
-  - boot_layer：memory的初始状态，默认初始状为0，memory的is_seq参数必须为false。
- 
-- **双进双出**：输入和输出都是双层序列。
-  - 如果有多个输入序列，不同输入含有的子句（subseq）数必须严格相等，但子句含有的词语数可以不相等。
-  - 输出一个双层序列，子句（subseq）数、子句的单词数和指定的一个输入序列一致，默认为第一个输入。
-  - memory：在step函数中定义memory，指向一个layer，通过引用memory得到这个layer上一个时刻的输出，形成recurrent连接。定义在外层`recurrent_group` step函数中的memory，能够记录上一个subseq 的状态，可以是一个单层序列（只作为read-only memory），也可以是一个词语。如果没有定义memory，那么 subseq 之间的运算是独立的。
-  - boot_layer：memory 初始状态，可以是一个单层序列（只作为read-only memory）或一个向量。默认不设置，即初始状态为0。
-
-- **双进单出**：目前还未支持，会报错"In hierachical RNN, all out links should be from sequences now"。
- 
-
-### 生成流程的使用方法
-使用`beam_search`需要遵循以下约定：
-
-- 单层RNN：从一个word生成下一个word。
+# Recurrent Group教程
+
+## 概述
+
+序列数据是自然语言处理任务面对的一种主要输入数据类型。
+
+一句话是由词语构成的序列，多句话进一步构成了段落。因此，段落可以看作是一个嵌套的双层的序列，这个序列的每个元素又是一个序列。
+
+双层序列是PaddlePaddle支持的一种非常灵活的数据组织方式，帮助我们更好地描述段落、多轮对话等更为复杂的语言数据。基于双层序列输入，我们可以设计搭建一个灵活的、层次化的RNN，分别从词语和句子级别编码输入数据，同时也能够引入更加复杂的记忆机制，更好地完成一些复杂的语言理解任务。
+
+在PaddlePaddle中，`recurrent_group`是一种任意复杂的RNN单元，用户只需定义RNN在一个时间步内完成的计算，PaddlePaddle负责完成信息和误差在时间序列上的传播。
+
+更进一步，`recurrent_group`同样可以扩展到双层序列的处理上。通过两个嵌套的`recurrent_group`分别定义子句级别和词语级别上需要完成的运算，最终实现一个层次化的复杂RNN。
+
+目前，在PaddlePaddle中，能够对双向序列进行处理的有`recurrent_group`和部分Layer，具体可参考文档：<a href = "hierarchical-layer.html">支持双层序列作为输入的Layer</a>。
+ 
+## 相关概念
+
+### 基本原理
+`recurrent_group` 是PaddlePaddle支持的一种任意复杂的RNN单元。使用者只需要关注于设计RNN在一个时间步之内完成的计算，PaddlePaddle负责完成信息和梯度在时间序列上的传播。
+
+PaddlePaddle中，`recurrent_group`的一个简单调用如下：
+
+``` python
+recurrent_group(step, input, reverse)
+```
+- step：一个可调用的函数，定义一个时间步之内RNN单元完成的计算
+- input：输入，必须是一个单层序列，或者一个双层序列
+- reverse：是否以逆序处理输入序列
+ 
+使用`recurrent_group`的核心是设计step函数的计算逻辑。step函数内部可以自由组合PaddlePaddle支持的各种layer，完成任意的运算逻辑。`recurrent_group` 的输入（即input）会成为step函数的输入，由于step 函数只关注于RNN一个时间步之内的计算，在这里`recurrent_group`替我们完成了原始输入数据的拆分。
+
+### 输入
+`recurrent_group`处理的输入序列主要分为以下三种类型：
+ 
+- **数据输入**：一个双层序列进入`recurrent_group`会被拆解为一个单层序列，一个单层序列进入`recurrent_group`会被拆解为非序列，然后交给step函数，这一过程对用户是完全透明的。可以有以下两种：1）通过data_layer拿到的用户输入；2）其它layer的输出。
+		
+- **只读Memory输入**：`StaticInput` 定义了一个只读的Memory，由`StaticInput`指定的输入不会被`recurrent_group`拆解，`recurrent_group` 循环展开的每个时间步总是能够引用所有输入，可以是一个非序列，或者一个单层序列。
+	  
+- **序列生成任务的输入**：`GeneratedInput`只用于在序列生成任务中指定输入数据。
+
+### 输入示例
+
+序列生成任务大多遵循encoder-decoer架构，encoder和decoder可以是能够处理序列的任意神经网络单元，而RNN是最流行的选择。
+
+给定encoder输出和当前词，decoder每次预测产生下一个最可能的词语。在这种结构中，decoder接受两个输入：
+    
+- 要生成的目标序列：是decoder的数据输入，也是decoder循环展开的依据，`recurrent_group`会对这类输入进行拆解。
+
+- encoder输出，可以是一个非序列，或者一个单层序列：是一个unbounded memory，decoder循环展开的每一个时间步会引用全部结果，不应该被拆解，这种类型的输入必须通过`StaticInput`指定。关于Unbounded Memory的更多讨论请参考论文 [Neural Turning Machine](https://arxiv.org/abs/1410.5401)。
+		
+在序列生成任务中，decoder RNN总是引用上一时刻预测出的词的词向量，作为当前时刻输入。`GeneratedInput`自动完成这一过程。
+		 
+### 输出
+`step`函数必须返回一个或多个Layer的输出，这个Layer的输出会作为整个`recurrent_group` 最终的输出结果。在输出的过程中，`recurrent_group` 会将每个时间步的输出拼接，这个过程对用户也是透明的。
+
+### memory
+memory只能在`recurrent_group`中定义和使用。memory不能独立存在，必须指向一个PaddlePaddle定义的Layer。引用memory得到这layer上一时刻输出，因此，可以将memory理解为一个时延操作。
+
+可以显示地指定一个layer的输出用于初始化memory。不指定时，memory默认初始化为0。
+
+## 双层RNN介绍
+`recurrent_group`帮助我们完成对输入序列的拆分，对输出的合并，以及计算逻辑在序列上的循环展开。
+
+利用这种特性，两个嵌套的`recurrent_group`能够处理双层序列，实现词语和句子两个级别的双层RNN结构。
+
+- 单层（word-level）RNN：每个状态（state）对应一个词（word）。
+- 双层（sequence-level）RNN：一个双层RNN由多个单层RNN组成，每个单层RNN（即双层RNN的每个状态）对应一个子句（subseq）。
+
+为了描述方便，下文以NLP任务为例，将含有子句（subseq）的段落定义为一个双层序列，将含有词语的句子定义为一个单层序列，那么0层序列即为一个词语。
+
+## 双层RNN的使用
+
+### 训练流程的使用方法
+使用 `recurrent_group`需要遵循以下约定：
+ 
+- **单进单出**：输入和输出都是单层序列。
+  - 如果有多个输入，不同输入序列含有的词语数必须严格相等。
+  - 输出一个单层序列，输出序列的词语数和输入序列一致。
+  - memory：在step函数中定义 memory指向一个layer，通过引用memory得到这个layer上一个时刻输出，形成recurrent 连接。memory的is_seq参数必须为false。如果没有定义memory，每个时间步之内的运算是独立的。
+  - boot_layer：memory的初始状态，默认初始状为0，memory的is_seq参数必须为false。
+ 
+- **双进双出**：输入和输出都是双层序列。
+  - 如果有多个输入序列，不同输入含有的子句（subseq）数必须严格相等，但子句含有的词语数可以不相等。
+  - 输出一个双层序列，子句（subseq）数、子句的单词数和指定的一个输入序列一致，默认为第一个输入。
+  - memory：在step函数中定义memory，指向一个layer，通过引用memory得到这个layer上一个时刻的输出，形成recurrent连接。定义在外层`recurrent_group` step函数中的memory，能够记录上一个subseq 的状态，可以是一个单层序列（只作为read-only memory），也可以是一个词语。如果没有定义memory，那么 subseq 之间的运算是独立的。
+  - boot_layer：memory 初始状态，可以是一个单层序列（只作为read-only memory）或一个向量。默认不设置，即初始状态为0。
+
+- **双进单出**：目前还未支持，会报错"In hierachical RNN, all out links should be from sequences now"。
+ 
+
+### 生成流程的使用方法
+使用`beam_search`需要遵循以下约定：
+
+- 单层RNN：从一个word生成下一个word。
 - 双层RNN：即把单层RNN生成后的subseq给拼接成一个新的双层seq。从语义上看，也不存在一个subseq直接生成下一个subseq的情况。

doc/howto/deep_model/rnn/rnn_en.rst

@@ -30,7 +30,7 @@ Then at the :code:`process` function, each :code:`yield` function will return th
     yield src_ids, trg_ids, trg_ids_next
 
 
-For more details description of how to write a data provider, please refer to `PyDataProvider2 <../../ui/data_provider/index.html>`_. The full data provider file is located at :code:`demo/seqToseq/dataprovider.py`.
+For more details description of how to write a data provider, please refer to :ref:`api_pydataprovider2` . The full data provider file is located at :code:`demo/seqToseq/dataprovider.py`.
 
 ===============================================
 Configure Recurrent Neural Network Architecture
@@ -42,8 +42,8 @@ Simple Gated Recurrent Neural Network
 
 Recurrent neural network process a sequence at each time step sequentially. An example of the architecture of LSTM is listed below.
 
-.. image:: ../../../tutorials/sentiment_analysis/bi_lstm.jpg
-	 :align: center
+.. image:: ../../../tutorials/sentiment_analysis/src/bi_lstm.jpg
+     :align: center
 
 Generally speaking, a recurrent network perform the following operations from :math:`t=1` to :math:`t=T`, or reversely from :math:`t=T` to :math:`t=1`.
 
@@ -102,11 +102,11 @@ Sequence to Sequence Model with Attention
 We will use the sequence to sequence model with attention as an example to demonstrate how you can configure complex recurrent neural network models. An illustration of the sequence to sequence model with attention is shown in the following figure.
 
 .. image:: ../../../tutorials/text_generation/encoder-decoder-attention-model.png
- 	 :align: center
+      :align: center
 
 In this model, the source sequence :math:`S = \{s_1, \dots, s_T\}` is encoded with a bidirectional gated recurrent neural networks. The hidden states of the bidirectional gated recurrent neural network :math:`H_S = \{H_1, \dots, H_T\}` is called *encoder vector* The decoder is a gated recurrent neural network. When decoding each token :math:`y_t`, the gated recurrent neural network generates a set of weights :math:`W_S^t = \{W_1^t, \dots, W_T^t\}`, which are used to compute a weighted sum of the encoder vector. The weighted sum of the encoder vector is utilized to condition the generation of the token :math:`y_t`.
 
-The encoder part of the model is listed below. It calls :code:`grumemory` to represent gated recurrent neural network. It is the recommended way of using recurrent neural network if the network architecture is simple, because it is faster than :code:`recurrent_group`. We have implemented most of the commonly used recurrent neural network architectures, you can refer to `Layers <../../ui/api/trainer_config_helpers/layers_index.html>`_  for more details.
+The encoder part of the model is listed below. It calls :code:`grumemory` to represent gated recurrent neural network. It is the recommended way of using recurrent neural network if the network architecture is simple, because it is faster than :code:`recurrent_group`. We have implemented most of the commonly used recurrent neural network architectures, you can refer to :ref:`api_trainer_config_helpers_layers` for more details.
 
 We also project the encoder vector to :code:`decoder_size` dimensional space, get the first instance of the backward recurrent network, and project it to :code:`decoder_size` dimensional space:
 
@@ -246,6 +246,6 @@ The code is listed below:
     outputs(beam_gen)
 
 
-Notice that this generation technique is only useful for decoder like generation process. If you are working on sequence tagging tasks, please refer to `Semantic Role Labeling Demo <../../demo/semantic_role_labeling/index.html>`_ for more details.
+Notice that this generation technique is only useful for decoder like generation process. If you are working on sequence tagging tasks, please refer to :ref:`semantic_role_labeling` for more details.
 
 The full configuration file is located at :code:`demo/seqToseq/seqToseq_net.py`.

doc/howto/index_cn.rst

+HOW TO
+=======
+
+Usage
+-------
+
+..  toctree::
+  :maxdepth: 1
+
+  concepts/use_concepts_cn.rst
+  cluster/k8s/paddle_on_k8s_cn.md
+  cluster/k8s/distributed_training_on_k8s_cn.md
+
+Development
+------------
+
+..  toctree::
+  :maxdepth: 1
+
+  write_docs/index_cn.rst
+  deep_model/index_cn.rst
+
+Optimization
+-------------
+
+..  toctree::
+  :maxdepth: 1

doc/howto/optimization/gpu_profiling_en.rst

@@ -51,7 +51,7 @@ In this tutorial, we will focus on nvprof and nvvp.
 :code:`test_GpuProfiler` from :code:`paddle/math/tests` directory will be used to evaluate
 above profilers. 
 
-.. literalinclude:: ../../paddle/math/tests/test_GpuProfiler.cpp
+.. literalinclude:: ../../../paddle/math/tests/test_GpuProfiler.cpp
    :language: c++
    :lines: 111-124
    :linenos:
@@ -77,7 +77,7 @@ As a simple example, consider the following:
 
 1. Add :code:`REGISTER_TIMER_INFO` and :code:`printAllStatus` functions (see the emphasize-lines).
 
-    .. literalinclude:: ../../paddle/math/tests/test_GpuProfiler.cpp
+    .. literalinclude:: ../../../paddle/math/tests/test_GpuProfiler.cpp
         :language: c++
         :lines: 111-124
         :emphasize-lines: 8-10,13
@@ -124,7 +124,7 @@ To use this command line profiler **nvprof**, you can simply issue the following
 
 1. Add :code:`REGISTER_GPU_PROFILER` function (see the emphasize-lines).
 
-    .. literalinclude:: ../../paddle/math/tests/test_GpuProfiler.cpp
+    .. literalinclude:: ../../../paddle/math/tests/test_GpuProfiler.cpp
         :language: c++
         :lines: 111-124
         :emphasize-lines: 6-7

doc/howto/source/api.rst

-API
-===
-
-.. doxygenfile:: paddle/api/PaddleAPI.h
-.. doxygenfile:: paddle/api/Internal.h

doc/howto/source/cuda/index.rst

-CUDA
-====
-
-.. toctree::
-  :maxdepth: 2
-
-  matrix.rst
-  nn.rst
-  utils.rst

doc/howto/source/cuda/matrix.rst

-Matrix
-======
-
-Base
-----
-
-hl_matrix.h
-```````````
-.. doxygenfile:: paddle/cuda/include/hl_matrix.h
-
-hl_matrix_base.h
-````````````````
-.. doxygenfile:: paddle/cuda/include/hl_matrix_base.cuh
-
-hl_matrix_apply.cuh
-```````````````````
-.. doxygenfile:: paddle/cuda/include/hl_matrix_apply.cuh
-
-hl_matrix_ops.cuh
-`````````````````
-.. doxygenfile:: paddle/cuda/include/hl_matrix_ops.cuh
-
-hl_matrix_type.cuh
-``````````````````
-.. doxygenfile:: paddle/cuda/include/hl_matrix_type.cuh
-
-hl_sse_matrix_kernel.cuh
-````````````````````````
-.. doxygenfile:: paddle/cuda/include/hl_sse_matrix_kernel.cuh
-
-Matrix Function 
----------------
-
-hl_batch_transpose.h
-````````````````````
-.. doxygenfile:: paddle/cuda/include/hl_batch_transpose.h
-
-hl_aggregate.h
-``````````````
-.. doxygenfile:: paddle/cuda/include/hl_aggregate.h
-
-hl_top_k.h
-``````````
-.. doxygenfile:: paddle/cuda/include/hl_top_k.h
-
-hl_table_apply.h
-````````````````
-.. doxygenfile:: paddle/cuda/include/hl_table_apply.h
-
-Sparse Matrix
--------------
-
-hl_sparse.h
-```````````
-.. doxygenfile:: paddle/cuda/include/hl_sparse.h
-
-hl_sparse.ph
-````````````
-.. doxygenfile:: paddle/cuda/include/hl_sparse.ph

doc/howto/source/cuda/nn.rst

-Neural Network
-==============
-
-Base
-----
-
-.. doxygenfile:: paddle/cuda/include/hl_gpu.h
-.. doxygenfile:: paddle/cuda/include/hl_functions.h
-.. doxygenfile:: paddle/cuda/include/hl_avx_functions.h
-.. doxygenfile:: paddle/cuda/include/hl_gpu_functions.cuh
-.. doxygenfile:: paddle/cuda/include/hl_activation_functions.h
-
-
-CNN Related APIs
-----------------
-.. doxygenfile:: paddle/cuda/include/hl_cnn.h
-.. doxygenfile:: paddle/cuda/include/hl_cuda_cudnn.h
-.. doxygenfile:: paddle/cuda/include/hl_cuda_cudnn.ph
-
-RNN Related APIs
-----------------
-
-.. doxygenfile:: paddle/cuda/include/hl_recurrent_apply.cuh
-.. doxygenfile:: paddle/cuda/include/hl_sequence.h
-
-LSTM Model
-``````````
-
-.. doxygenfile:: paddle/cuda/include/hl_lstm.h
-.. dpxygenfile:: paddle/cuda/include/hl_cpu_lstm.cuh
-.. doxygenfile:: paddle/cuda/include/hl_gpu_lstm.cuh
-.. doxygenfile:: paddle/cuda/include/hl_lstm_ops.cuh
-
-GRU Model
-`````````
-
-.. doxygenfile:: paddle/cuda/include/hl_gru_ops.cuh
-.. doxygenfile:: paddle/cuda/include/hl_cpu_gru.cuh
-.. doxygenfile:: paddle/cuda/include/hl_gpu_gru.cuh

doc/howto/source/cuda/utils.rst

-Utils
-=====
-
-Dynamic Link Libs
------------------
-.. doxygenfile:: paddle/cuda/include/hl_dso_loader.h
-
-GPU Resources
--------------
-
-hl_cuda.ph
-``````````
-.. doxygenfile:: paddle/cuda/include/hl_cuda.ph
-
-hl_cuda.h
-`````````
-.. doxygenfile:: paddle/cuda/include/hl_cuda.h
-
-HPPL Base
----------
-.. doxygenfile:: paddle/cuda/include/hl_base.h
-
-CUBLAS Wrapper
---------------
-.. doxygenfile:: paddle/cuda/include/hl_cuda_cublas.h
-
-Timer
------
-.. doxygenfile:: paddle/cuda/include/hl_time.h
-
-Thread Resource
----------------
-.. doxygenfile:: paddle/cuda/include/hl_thread.ph
-
-Device Function
----------------
-.. doxygenfile:: paddle/cuda/include/hl_device_functions.cuh

doc/howto/source/gserver/activations.rst

-Activations
-===========
-
-..  doxygenclass:: paddle::ActivationFunction
-    :members:

doc/howto/source/gserver/dataproviders.rst

-==============
-Data Providers
-==============
-
-DataProviders
-=============
-
-Base
-----
-..  doxygenclass:: paddle::DataProvider
-    :members:
-
-DataProviderGroup
------------------
-..  doxygenclass:: paddle::DataProviderGroup
-    :members:
-
-MultiDataProvider
------------------
-..  doxygenclass:: paddle::MultiDataProvider
-    :members:
-
-PyDataProvider
-==============
-
-IFieldScanner
--------------
-..  doxygenclass:: paddle::IFieldScanner
-    :members:
-
-DenseScanner
--------------
-..  doxygenclass:: paddle::DenseScanner
-    :members:
-
-IndexScanner
--------------
-..  doxygenclass:: paddle::IndexScanner
-    :members:
-
-SparseNonValueScanner
----------------------
-..  doxygenclass:: paddle::SparseNonValueScanner
-    :members:
-
-SparseValueScanner
-------------------
-..  doxygenclass:: paddle::SparseValueScanner
-    :members:
-
-SequenceScanner
----------------
-..  doxygenclass:: paddle::SparseValueScanner
-    :members:
-
-IPyDataProviderCache
---------------------
-..  doxygenclass:: paddle::IPyDataProviderCache
-    :members:
-
-NoCacheStrategy
----------------
-..  doxygenclass:: paddle::NoCacheStrategy
-    :members:
-
-CacheOnePassInMemory
---------------------
-..  doxygenclass:: paddle::CacheOnePassInMemory
-    :members:
-
-IPyDataProvider
----------------
-..  doxygenclass:: paddle::PyDataProvider2
-    :members:
-
-ProtoDataProvider
-=================
-
-ProtoDataProvider
-----------------
-..  doxygenclass:: paddle::ProtoDataProvider
-    :members:
-
-ProtoSequenceDataProvider
--------------------------
-..  doxygenclass:: paddle::ProtoSequenceDataProvider
-    :members:

doc/howto/source/gserver/evaluators.rst

-==========
-Evaluators
-==========
-
-Base
-====
-
-..  doxygenclass:: paddle::Evaluator
-    :members:
-
-Sum
-===
-
-SumEvaluator
-------------
-..  doxygenclass:: paddle::SumEvaluator
-    :members:
-
-ColumnSumEvaluator
-------------------
-..  doxygenclass:: paddle::ColumnSumEvaluator
-    :members:
-
-Classification
-==============
-
-ClassificationErrorEvaluator
----------------------------
-..  doxygenclass:: paddle::ClassificationErrorEvaluator
-    :members:
-
-SequenceClassificationErrorEvaluator
-------------------------------------
-..  doxygenclass:: paddle::SequenceClassificationErrorEvaluator
-    :members:
-
-AucEvaluator
--------------
-..  doxygenclass:: paddle::AucEvaluator
-    :members:
-
-PrecisionRecallEvaluator
-------------------------
-..  doxygenclass:: paddle::PrecisionRecallEvaluator
-    :members:
-
-ChunkEvaluator
---------------
-..  doxygenclass:: paddle::ChunkEvaluator
-    :members:
-
-CTCEvaluator
-------------
-..  doxygenclass:: paddle::CTCErrorEvaluator
-    :members:
-
-
-Rank
-====
-
-PnpairEvaluator
--------------
-..  doxygenclass:: paddle::PnpairEvaluator
-    :members:
-
-AucEvaluator
--------------
-..  doxygenclass:: paddle::RankAucEvaluator
-    :members:
-
-
-Printer
-=======
-
-ValuePrinter
--------------
-..  doxygenclass:: paddle::ValuePrinter
-    :members:
-
-GradientPrinter
----------------
-..  doxygenclass:: paddle::GradientPrinter
-    :members:
-
-MaxIdPrinter
-------------
-..  doxygenclass:: paddle::MaxIdPrinter
-    :members:
-
-MaxFramePrinter
----------------
-..  doxygenclass:: paddle::MaxFramePrinter
-    :members:
-
-SequenceTextPrinter
-------------------
-..  doxygenclass:: paddle::SequenceTextPrinter
-    :members:
-
-ClassificationErrorPrinter
---------------------------
-..  doxygenclass:: paddle::ClassificationErrorPrinter
-    :members:

doc/howto/source/gserver/gradientmachines.rst

-Gradient Machines
-=================
-
-GradientMachine
----------------
-..  doxygenclass:: paddle::GradientMachine
-    :members:
-
-GradientMachineMode
--------------------
-..  doxygenclass:: paddle::IGradientMachineMode
-    :members:
-
-MultiGradientMachine
---------------------
-..  doxygenclass:: paddle::MultiGradientMachine
-    :members:
-
-TrainerThread
-`````````````
-..  doxygenclass:: paddle::TrainerThread
-    :members:
-
-RecurrentGradientMachine
-------------------------
-..  doxygenclass:: paddle::RecurrentGradientMachine
-    :members:

doc/howto/source/gserver/index.rst

-GServer
-=======
-
-.. toctree::
-  :maxdepth: 2
-
-  activations.rst
-  dataproviders.rst
-  evaluators.rst
-  gradientmachines.rst
-  layers.rst
-  neworks.rst

doc/howto/source/gserver/layers.rst

-======
-Layers
-======
-
-Base
-====
-
-Layer 
------
-..  doxygenclass:: paddle::Layer
-    :members:
-
-Projection
-----------
-..  doxygenclass:: paddle::Projection
-    :members:
-
-Operator
---------
-..  doxygenclass:: paddle::Operator
-    :members:
-    
-Data Layer
-==========
-
-..  doxygenclass:: paddle::DataLayer
-    :members:
-
-Fully Connected Layers
-======================
-
-FullyConnectedLayer
--------------------
-..  doxygenclass:: paddle::FullyConnectedLayer
-    :members:
-
-SelectiveFullyConnectedLayer
-----------------------------
-..  doxygenclass:: paddle::SelectiveFullyConnectedLayer
-    :members:
-
-Conv Layers
-===========
-
-ConvBaseLayer
--------------
-..  doxygenclass:: paddle::ConvBaseLayer
-    :members:
-
-ConvOperator
-------------
-..  doxygenclass:: paddle::ConvOperator
-    :members:
-
-ConvShiftLayer
---------------
-..  doxygenclass:: paddle::ConvShiftLayer
-    :members:
-
-CudnnConvLayer
---------------
-..  doxygenclass:: paddle::CudnnConvLayer
-    :members:
-
-ExpandConvBaseLayer
--------------------
-..  doxygenclass:: paddle::ExpandConvBaseLayer
-    :members:
-
-ExpandConvLayer
----------------
-..  doxygenclass:: paddle::ExpandConvLayer
-    :members:
-
-ContextProjection
------------------
-..  doxygenclass:: paddle::ContextProjection
-    :members:
-
-Pooling Layers
-==============
-
-PoolLayer
----------
-..  doxygenclass:: paddle::PoolLayer
-    :members:
-
-PoolProjectionLayer
--------------------
-..  doxygenclass:: paddle::PoolProjectionLayer
-    :members:
-
-CudnnPoolLayer
---------------
-..  doxygenclass:: paddle::CudnnPoolLayer
-    :members:
-
-SpatialPyramidPoolLayer
------------------------
-..  doxygenclass:: paddle::SpatialPyramidPoolLayer
-    :members:
-
-MaxOutLayer
------------
-..  doxygenclass:: paddle::MaxOutLayer
-    :members:
-
-Norm Layers
-===========
-
-NormLayer
----------
-..  doxygenclass:: paddle::NormLayer
-    :members:
-
-CMRProjectionNormLayer
-----------------------
-..  doxygenclass:: paddle::CMRProjectionNormLayer
-    :members:
-
-DataNormLayer
--------------
-..  doxygenclass:: paddle::DataNormLayer
-    :members:
-
-ResponseNormLayer
------------------
-..  doxygenclass:: paddle::ResponseNormLayer
-    :members:
-
-BatchNormBaseLayer
-------------------
-..  doxygenclass:: paddle::BatchNormBaseLayer
-    :members:
-
-BatchNormalizationLayer
------------------------
-..  doxygenclass:: paddle::BatchNormalizationLayer
-    :members:
-
-CudnnBatchNormLayer
------------------------
-..  doxygenclass:: paddle::CudnnBatchNormLayer
-    :members:
-
-SumToOneNormLayer
------------------
-..  doxygenclass:: paddle::SumToOneNormLayer
-    :members:
-
-Activation Layer
-================
-
-ParameterReluLayer
-------------------
-..  doxygenclass:: paddle::ParameterReluLayer
-    :members:
-
-Recurrent Layers
-================
-
-RecurrentLayer
---------------
-..  doxygenclass:: paddle::RecurrentLayer
-    :members:
-
-SequenceToBatch
----------------
-..  doxygenclass:: paddle::SequenceToBatch
-    :members:
-
-LSTM
-----
-LstmLayer
-`````````
-..  doxygenclass:: paddle::LstmLayer
-    :members:
-
-LstmStepLayer
-`````````````
-..  doxygenclass:: paddle::LstmStepLayer
-    :members:
-
-LstmCompute
-```````````
-..  doxygenclass:: paddle::LstmCompute
-    :members:
-
-MDLSTM
-------
-MDLstmLayer
-```````````
-..  doxygenclass:: paddle::MDLstmLayer
-    :members:
-
-CoordIterator
-`````````````
-..  doxygenclass:: paddle::CoordIterator
-    :members:
-
-GRU
----
-GatedRecurrentLayer
-```````````````````
-..  doxygenclass:: paddle::GatedRecurrentLayer
-    :members:
-
-GruStepLayer
-````````````
-..  doxygenclass:: paddle::GruStepLayer
-    :members:
-
-GruCompute
-``````````
-..  doxygenclass:: paddle::GruCompute
-    :members:
-
-Recurrent Layer Group
-=====================
-
-AgentLayer
-----------
-..  doxygenclass:: paddle::AgentLayer
-    :members:
-
-SequenceAgentLayer
-------------------
-..  doxygenclass:: paddle::SequenceAgentLayer
-    :members:
-
-GatherAgentLayer
-----------------
-..  doxygenclass:: paddle::GatherAgentLayer
-    :members:
-
-SequenceGatherAgentLayer
-------------------------
-..  doxygenclass:: paddle::SequenceGatherAgentLayer
-    :members:
-
-ScatterAgentLayer
------------------
-..  doxygenclass:: paddle::ScatterAgentLayer
-    :members:
-
-SequenceScatterAgentLayer
--------------------------
-..  doxygenclass:: paddle::SequenceScatterAgentLayer
-    :members:
-
-GetOutputLayer
---------------
-..  doxygenclass:: paddle::GetOutputLayer
-    :members:
-
-Mixed Layer
-===========
-..  doxygenclass:: paddle::MixedLayer
-    :members:
-
-DotMulProjection
-----------------
-..  doxygenclass:: paddle::DotMulProjection
-    :members:
-
-DotMulOperator
---------------
-..  doxygenclass:: paddle::DotMulOperator
-    :members:
-
-FullMatrixProjection
---------------------
-..  doxygenclass:: paddle::FullMatrixProjection
-    :members:
-
-IdentityProjection
-------------------
-..  doxygenclass:: paddle::IdentityProjection
-    :members:
-
-IdentityOffsetProjection
-------------------------
-..  doxygenclass:: paddle::IdentityOffsetProjection
-    :members:
-
-TableProjection
----------------
-..  doxygenclass:: paddle::TableProjection
-    :members:
-
-TransposedFullMatrixProjection
-------------------------------
-..  doxygenclass:: paddle::TransposedFullMatrixProjection
-    :members:
-
-Aggregate Layers
-================
-
-Aggregate
----------
-AverageLayer
-````````````
-..  doxygenclass:: paddle::AverageLayer
-    :members:
-
-MaxLayer
-````````
-..  doxygenclass:: paddle::MaxLayer
-    :members:
-
-SequenceLastInstanceLayer
-`````````````````````````
-..  doxygenclass:: paddle::SequenceLastInstanceLayer
-    :members:
-
-Concat
-------
-ConcatenateLayer
-````````````````
-..  doxygenclass:: paddle::ConcatenateLayer
-    :members:
-
-ConcatenateLayer2
-`````````````````
-..  doxygenclass:: paddle::ConcatenateLayer2
-    :members:
-
-SequenceConcatLayer
-```````````````````
-..  doxygenclass:: paddle::SequenceConcatLayer
-    :members:
-
-Subset
-------
-SubSequenceLayer
-````````````````
-..  doxygenclass:: paddle::SubSequenceLayer
-    :members:
-
-Reshaping Layers
-================
-
-BlockExpandLayer
-----------------
-..  doxygenclass:: paddle::BlockExpandLayer
-    :members:
-
-ExpandLayer
------------
-..  doxygenclass:: paddle::ExpandLayer
-    :members:
-
-FeatureMapExpandLayer
----------------------
-..  doxygenclass:: paddle::FeatureMapExpandLayer
-    :members:
-
-ResizeLayer
------------
-..  doxygenclass:: paddle::ResizeLayer
-    :members:
-
-SequenceReshapeLayer
---------------------
-..  doxygenclass:: paddle::SequenceReshapeLayer
-    :members:
-
-Math Layers
-===========
-
-AddtoLayer
-----------
-..  doxygenclass:: paddle::AddtoLayer
-    :members:
-
-ConvexCombinationLayer
-----------------------
-..  doxygenclass:: paddle::ConvexCombinationLayer
-    :members:
-
-InterpolationLayer
-------------------
-..  doxygenclass:: paddle::InterpolationLayer
-    :members:
-
-MultiplexLayer
---------------
-..  doxygenclass:: paddle::MultiplexLayer
-    :members:
-
-OuterProdLayer
---------------
-..  doxygenclass:: paddle::OuterProdLayer
-    :members:
-
-PowerLayer
-----------
-..  doxygenclass:: paddle::PowerLayer
-    :members:
-
-ScalingLayer
-------------
-..  doxygenclass:: paddle::ScalingLayer
-    :members:
-
-SlopeInterceptLayer
--------------------
-..  doxygenclass:: paddle::SlopeInterceptLayer
-    :members:
-
-TensorLayer
-------------
-..  doxygenclass:: paddle::TensorLayer
-    :members:
-
-TransLayer
-----------
-..  doxygenclass:: paddle::TransLayer
-    :members:
-
-Sampling Layers
-===============
-
-BilinearInterpLayer
--------------------
-..  doxygenclass:: paddle::BilinearInterpLayer
-    :members:
-
-MultinomialSampler
-------------------
-..  doxygenclass:: paddle::MultinomialSampler
-    :members:
-
-MaxIdLayer
-----------
-..  doxygenclass:: paddle::MaxIdLayer
-    :members:
-
-SamplingIdLayer
----------------
-..  doxygenclass:: paddle::SamplingIdLayer
-    :members:
-
-Cost Layers
-===========
-
-CostLayer
------------
-..  doxygenclass:: paddle::CostLayer
-    :members:
-
-HuberTwoClass
-`````````````
-..  doxygenclass:: paddle::HuberTwoClass
-    :members:
-
-LambdaCost
-```````````
-..  doxygenclass:: paddle::LambdaCost
-    :members:
-
-MultiBinaryLabelCrossEntropy
-````````````````````````````
-..  doxygenclass:: paddle::MultiBinaryLabelCrossEntropy
-    :members:
-
-MultiClassCrossEntropy
-```````````````````````
-..  doxygenclass:: paddle::MultiClassCrossEntropy
-    :members:
-
-MultiClassCrossEntropyWithSelfNorm
-``````````````````````````````````
-..  doxygenclass:: paddle::MultiClassCrossEntropyWithSelfNorm
-    :members:
-
-RankingCost
-```````````
-..  doxygenclass:: paddle::RankingCost
-    :members:
-
-SoftBinaryClassCrossEntropy
-```````````````````````````
-..  doxygenclass:: paddle::SoftBinaryClassCrossEntropy
-    :members:
-
-SumOfSquaresCostLayer
-`````````````````````
-..  doxygenclass:: paddle::SumOfSquaresCostLayer
-    :members:
-
-SumCostLayer
-`````````````````````
-..  doxygenclass:: paddle::SumCostLayer
-    :members:
-
-CosSimLayer
------------
-..  doxygenclass:: paddle::CosSimLayer
-    :members:
-
-CosSimVecMatLayer
------------------
-..  doxygenclass:: paddle::CosSimVecMatLayer
-    :members:
-
-CRFDecodingLayer
-----------------
-..  doxygenclass:: paddle::CRFDecodingLayer
-    :members:
-
-CRFLayer
---------
-..  doxygenclass:: paddle::CRFLayer
-    :members:
-
-CTCLayer
---------
-..  doxygenclass:: paddle::CTCLayer
-    :members:
-
-HierarchicalSigmoidLayer
-------------------------
-..  doxygenclass:: paddle::HierarchicalSigmoidLayer
-    :members:
-
-LinearChainCRF
---------------
-..  doxygenclass:: paddle::LinearChainCRF
-    :members:
-
-LinearChainCTC
---------------
-..  doxygenclass:: paddle::LinearChainCTC
-    :members:
-
-NCELayer
---------
-..  doxygenclass:: paddle::NCELayer
-    :members:
-
-Validation Layers
------------------
-
-ValidationLayer
-```````````````
-..  doxygenclass:: paddle::ValidationLayer
-    :members:
-
-AucValidation
-`````````````
-..  doxygenclass:: paddle::AucValidation
-    :members:
-
-PnpairValidation
-````````````````
-..  doxygenclass:: paddle::PnpairValidation
-    :members:
-
-Check Layers
-============
-
-EosIdCheckLayer
----------------
-..  doxygenclass:: paddle::EosIdCheckLayer
-    :members:

doc/howto/source/gserver/neworks.rst

-Networks
-========
-
-NeuralNetwork
--------------
-..  doxygenclass:: paddle::NeuralNetwork
-    :members:
-
-ParallelNeuralNetwork
----------------------
-..  doxygenclass:: paddle::ParallelNeuralNetwork
-    :members:

doc/howto/source/index.rst

-Source Code Documents
-=====================
-
-.. toctree::
-  :maxdepth: 1
-
-  gserver/index.rst
-  trainer.rst
-  parameter/index.rst
-  pserver/index.rst
-  api.rst
-  cuda/index.rst
-  math/index.rst
-  utils/index.rst

doc/howto/source/math/functions.rst

-Functions
-=========
-
-MathFunctions
--------------
-.. doxygenfile:: paddle/math/MathFunctions.h
-
-SIMDFunctions
--------------
-.. doxygenfile:: paddle/math/SIMDFunctions.h

doc/howto/source/math/index.rst

-Math
-====
-
-.. toctree::
-  :maxdepth: 2
-
-  vector.rst
-  matrix.rst
-  functions.rst
-  utils.rst