Skip to content

P
Paddle

机器未来 / Paddle
与 Fork 源项目一致

Fork自 PaddlePaddle / Paddle

通知 1
Star 1
Fork 0
P
Paddle
提交 1ffdecf1 编写于 作者: G guosheng
浏览文件
操作

Merge branch 'develop' of https://github.com/PaddlePaddle/paddle into add-ROIPooling-dev

上级 3cf01b5d e5d810b9
展开全部 隐藏空白更改
内联 并排
Showing with 74657 addition and 10420 deletion

要显示的变更太多。

To preserve performance only 1000 of 1000+ files are displayed.
文本差异 电子邮件补丁
.clang_format.hook 0 → 100755
浏览文件 @ 1ffdecf1
#!/bin/bash
set -e
readonly VERSION="3.8"
version=$(clang-format -version)
if ! [[ $version == *"$VERSION"* ]]; then
echo "clang-format version check failed."
echo "a version contains '$VERSION' is needed, but get '$version'"
echo "you can install the right version, and make an soft-link to '\$PATH' env"
exit -1
fi
clang-format $@
.gitignore
浏览文件 @ 1ffdecf1
......@@ -22,7 +22,10 @@ cmake-build-*
# generated while compiling
python/paddle/v2/framework/core.so
paddle/pybind/pybind.h
CMakeFiles
cmake_install.cmake
paddle/.timestamp
python/paddlepaddle.egg-info/
paddle/pybind/pybind.h
python/paddle/v2/framework/tests/tmp/*
.pre-commit-config.yaml
浏览文件 @ 1ffdecf1
......@@ -19,10 +19,10 @@
- id: end-of-file-fixer
- repo: local
hooks:
- id: clang-format
- id: clang-format-with-version-check
name: clang-format
description: Format files with ClangFormat.
entry: clang-format -i
entry: bash ./.clang_format.hook -i
language: system
files: \.(c|cc|cxx|cpp|cu|h|hpp|hxx|proto)$
- repo: https://github.com/PaddlePaddle/pre-commit-golang
......@@ -31,6 +31,3 @@
- id: go-fmt
types:
- go
- id: gometalinter
types:
- go
.travis.yml
浏览文件 @ 1ffdecf1
......@@ -4,7 +4,6 @@ cache:
- $HOME/.ccache
- $HOME/.cache/pip
- $TRAVIS_BUILD_DIR/build/third_party
- $TRAVIS_BUILD_DIR/build_android/third_party
sudo: required
dist: trusty
os:
......@@ -12,7 +11,6 @@ os:
env:
- JOB=build_doc
- JOB=check_style
- JOB=build_android
addons:
apt:
packages:
......@@ -23,7 +21,6 @@ addons:
- python
- python-pip
- python2.7-dev
- python-numpy
- python-wheel
- libboost-dev
- curl
......@@ -33,22 +30,27 @@ addons:
- automake
- libtool
- ccache
ssh_known_hosts: 52.76.173.135
before_install:
- if [[ "$JOB" == "check_style" ]]; then sudo ln -s /usr/bin/clang-format-3.8 /usr/bin/clang-format; fi
# Paddle is using protobuf 3.1 currently. Protobuf 3.2 breaks the compatibility. So we specify the python
# protobuf version.
- pip install -r $TRAVIS_BUILD_DIR/python/requirements.txt
- pip install wheel sphinx==1.5.6 recommonmark sphinx-rtd-theme==0.1.9 virtualenv pre-commit LinkChecker
- curl https://glide.sh/get | bash
- eval "$(GIMME_GO_VERSION=1.8.3 gimme)"
- go get -u github.com/alecthomas/gometalinter
- gometalinter --install
- sudo pip install -r $TRAVIS_BUILD_DIR/python/requirements.txt
- sudo pip install wheel sphinx==1.5.6 recommonmark sphinx-rtd-theme==0.1.9 virtualenv pre-commit LinkChecker
- |
function timeout() { perl -e 'alarm shift; exec @ARGV' "$@"; }
script:
- |
timeout 2580 paddle/scripts/travis/${JOB}.sh # 43min timeout
RESULT=$?; if [ $RESULT -eq 0 ] || [ $RESULT -eq 142 ]; then true; else false; fi;
- |
if [[ "$JOB" != "build_doc" ]]; then exit 0; fi;
if [[ "$TRAVIS_PULL_REQUEST" != "false" ]]; then exit 0; fi;
if [[ "$TRAVIS_BRANCH" != "develop" && ! "$TRAVIS_BRANCH" =~ ^v[[:digit:]]+\.[[:digit:]]+(\.[[:digit:]]+)?(-\S*)?$ ]]; then exit 0; fi;
export DEPLOY_DOCS_SH=https://raw.githubusercontent.com/PaddlePaddle/PaddlePaddle.org/master/scripts/deploy/deploy_docs.sh
export DOCS_DIR=`pwd`
cd ..
curl $DEPLOY_DOCS_SH | bash -s $CONTENT_DEC_PASSWD $TRAVIS_BRANCH $DOCS_DIR $DOCS_DIR/build/doc
notifications:
email:
on_success: change
......
CMakeLists.txt
浏览文件 @ 1ffdecf1
......@@ -27,7 +27,7 @@ if(NOT CMAKE_CROSSCOMPILING)
endif(NOT CMAKE_CROSSCOMPILING)
find_package(Git REQUIRED)
find_package(Threads REQUIRED)
if(NOT ANDROID)
if(NOT ANDROID AND NOT IOS)
find_package(Boost QUIET)
endif()
......@@ -55,6 +55,7 @@ option(WITH_C_API "Compile PaddlePaddle with C-API(Prediction)" OFF)
option(WITH_GOLANG "Compile PaddlePaddle with GOLANG" OFF)
option(GLIDE_INSTALL "Download and install go dependencies " ON)
option(USE_NNPACK "Compile PaddlePaddle with NNPACK library" OFF)
option(USE_EIGEN_FOR_BLAS "Use matrix multiplication in Eigen" OFF)
# CMAKE_BUILD_TYPE
if(NOT CMAKE_BUILD_TYPE)
......@@ -63,24 +64,37 @@ if(NOT CMAKE_BUILD_TYPE)
FORCE)
endif()
if(ANDROID)
if(${CMAKE_SYSTEM_VERSION} VERSION_LESS "21")
message(FATAL_ERROR "Unsupport standalone toolchains with Android API level lower than 21")
if(ANDROID OR IOS)
if(ANDROID)
if(${CMAKE_SYSTEM_VERSION} VERSION_LESS "16")
message(FATAL_ERROR "Unsupport standalone toolchains with Android API level lower than 16")
elseif(${CMAKE_SYSTEM_VERSION} VERSION_LESS "21")
# TODO: support glog for Android api 16 ~ 19 in the future
message(WARNING "Using the unofficial git repository <https://github.com/Xreki/glog.git> instead")
endif()
endif()
set(WITH_GPU OFF CACHE STRING
"Disable GPU when cross-compiling for Android" FORCE)
"Disable GPU when cross-compiling for Android and iOS" FORCE)
set(WITH_AVX OFF CACHE STRING
"Disable AVX when cross-compiling for Android" FORCE)
"Disable AVX when cross-compiling for Android and iOS" FORCE)
set(WITH_PYTHON OFF CACHE STRING
"Disable PYTHON when cross-compiling for Android" FORCE)
"Disable PYTHON when cross-compiling for Android and iOS" FORCE)
set(WITH_RDMA OFF CACHE STRING
"Disable RDMA when cross-compiling for Android" FORCE)
"Disable RDMA when cross-compiling for Android and iOS" FORCE)
set(WITH_MKLDNN OFF CACHE STRING
"Disable MKLDNN when cross-compiling for Android" FORCE)
"Disable MKLDNN when cross-compiling for Android and iOS" FORCE)
set(WITH_MKLML OFF CACHE STRING
"Disable MKLML package when cross-compiling for Android" FORCE)
endif(ANDROID)
"Disable MKLML package when cross-compiling for Android and iOS" FORCE)
# Compile PaddlePaddle mobile inference library
if (NOT WITH_C_API)
set(WITH_C_API ON CACHE STRING
"Always compile the C_API when cross-compiling for Android and iOS" FORCE)
endif()
set(MOBILE_INFERENCE ON)
add_definitions(-DPADDLE_MOBILE_INFERENCE)
endif()
set(THIRD_PARTY_PATH "${CMAKE_BINARY_DIR}/third_party" CACHE STRING
"A path setting third party libraries download & build directories.")
......@@ -91,6 +105,12 @@ if (WITH_C_API AND WITH_PYTHON)
"different Python interpreter from compiling.")
endif()
if(MOBILE_INFERENCE)
set(THIRD_PARTY_BUILD_TYPE MinSizeRel)
else()
set(THIRD_PARTY_BUILD_TYPE Release)
endif()
########################################################################################
include(external/mklml) # download mklml package
......@@ -106,7 +126,8 @@ include(external/swig) # download, build, install swig
include(external/warpctc) # download, build, install warpctc
include(external/any) # download libn::any
include(external/eigen) # download eigen3
include(external/pybind11) # download pybind11
include(external/pybind11) # download pybind11
include(external/nccl)
include(cudnn) # set cudnn libraries, must before configure
include(configure) # add paddle env configuration
......@@ -137,9 +158,9 @@ set(EXTERNAL_LIBS
)
if(WITH_GPU)
list(APPEND EXTERNAL_LIB ${CUDA_LIBRARIES} ${CUDA_rt_LIBRARY})
list(APPEND EXTERNAL_LIBS ${CUDA_LIBRARIES} ${CUDA_rt_LIBRARY})
if(NOT WITH_DSO)
list(APPEND EXTERNAL_LIB ${CUDNN_LIBRARY} ${CUDA_CUBLAS_LIBRARIES} ${CUDA_curand_LIBRARY})
list(APPEND EXTERNAL_LIBS ${CUDNN_LIBRARY} ${CUDA_CUBLAS_LIBRARIES} ${CUDA_curand_LIBRARY} ${NCCL_LIBRARY})
endif(NOT WITH_DSO)
endif(WITH_GPU)
......@@ -154,9 +175,11 @@ endif(USE_NNPACK)
add_subdirectory(proto)
# "add_subdirectory(go)" should be placed after the following loine,
# because it depends on paddle/optimizer.
add_subdirectory(paddle/optimizer)
if(NOT MOBILE_INFERENCE)
# "add_subdirectory(go)" should be placed after the following loine,
# because it depends on paddle/optimizer.
add_subdirectory(paddle/optimizer)
endif()
# "add_subdirectory(paddle)" and "add_subdirectory(python)" should be
# placed after this block, because they depends on it.
......
CONTRIBUTING.md
浏览文件 @ 1ffdecf1
./doc/howto/dev/contribute_to_paddle_en.md
# Contribute Code
We sincerely appreciate your contribution. This document explains our workflow and work style.
## Workflow
PaddlePaddle uses this [Git branching model](http://nvie.com/posts/a-successful-git-branching-model/). The following steps guide usual contributions.
1. Fork
Our development community has been growing fastly; it doesn't make sense for everyone to write into the official repo. So, please file Pull Requests from your fork. To make a fork, just head over to the GitHub page and click the ["Fork" button](https://help.github.com/articles/fork-a-repo/).
1. Clone
To make a copy of your fork to your local computers, please run
```bash
git clone https://github.com/your-github-account/paddle
cd paddle
```
1. Create the local feature branch
For daily works like adding a new feature or fixing a bug, please open your feature branch before coding:
```bash
git checkout -b my-cool-stuff
```
1. Commit
Before issuing your first `git commit` command, please install [`pre-commit`](http://pre-commit.com/) by running the following commands:
```bash
pip install pre-commit
pre-commit install
```
Our pre-commit configuration requires clang-format 3.8 for auto-formating C/C++ code and yapf for Python.
Once installed, `pre-commit` checks the style of code and documentation in every commit. We will see something like the following when you run `git commit`:
```
➜ git commit
CRLF end-lines remover...............................(no files to check)Skipped
yapf.................................................(no files to check)Skipped
Check for added large files..............................................Passed
Check for merge conflicts................................................Passed
Check for broken symlinks................................................Passed
Detect Private Key...................................(no files to check)Skipped
Fix End of Files.....................................(no files to check)Skipped
clang-formater.......................................(no files to check)Skipped
[my-cool-stuff c703c041] add test file
1 file changed, 0 insertions(+), 0 deletions(-)
create mode 100644 233
```
1. Build and test
Users can build PaddlePaddle natively on Linux and Mac OS X. But to unify the building environment and to make it easy for debugging, the recommended way is [using Docker](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/howto/dev/build_en.md).
1. Keep pulling
An experienced Git user pulls from the official repo often -- daily or even hourly, so they notice conflicts with others work early, and it's easier to resolve smaller conflicts.
```bash
git remote add upstream https://github.com/PaddlePaddle/Paddle
git pull upstream develop
```
1. Push and file a pull request
You can "push" your local work into your forked repo:
```bash
git push origin my-cool-stuff
```
The push allows you to create a pull request, requesting owners of this [official repo](https://github.com/PaddlePaddle/Paddle) to pull your change into the official one.
To create a pull request, please follow [these steps](https://help.github.com/articles/creating-a-pull-request/).
If your change is for fixing an issue, please write ["Fixes <issue-URL>"](https://help.github.com/articles/closing-issues-using-keywords/) in the description section of your pull request. Github would close the issue when the owners merge your pull request.
Please remember to specify some reviewers for your pull request. If you don't know who are the right ones, please follow Github's recommendation.
1. Delete local and remote branches
To keep your local workspace and your fork clean, you might want to remove merged branches:
```bash
git push origin :my-cool-stuff
git checkout develop
git pull upstream develop
git branch -d my-cool-stuff
```
### Code Review
- Please feel free to ping your reviewers by sending them the URL of your pull request via IM or email. Please do this after your pull request passes the CI.
- Please answer reviewers' every comment. If you are to follow the comment, please write "Done"; please give a reason otherwise.
- If you don't want your reviewers to get overwhelmed by email notifications, you might reply their comments by [in a batch](https://help.github.com/articles/reviewing-proposed-changes-in-a-pull-request/).
- Reduce the unnecessary commits. Some developers commit often. It is recommended to append a sequence of small changes into one commit by running `git commit --amend` instead of `git commit`.
## Coding Standard
### Code Style
Our C/C++ code follows the [Google style guide](http://google.github.io/styleguide/cppguide.html).
Our Python code follows the [PEP8 style guide](https://www.python.org/dev/peps/pep-0008/).
Our build process helps to check the code style. In [`build.sh`](https://github.com/PaddlePaddle/Paddle/blob/b84e8226514b8bb4405c3c28e54aa5077193d179/paddle/scripts/docker/build.sh#L42), the entry point of our [builder Docker image](https://github.com/PaddlePaddle/Paddle/blob/b84e8226514b8bb4405c3c28e54aa5077193d179/Dockerfile#L88), the CMake argument `WITH_STYLE_CHECK` is set to `ON` by default. This flag is on
Please install pre-commit, which automatically reformat the changes to C/C++ and Python code whenever we run `git commit`. To check the whole codebase, we can run the command `pre-commit run -a`, as in the [`check_style.sh` file](https://github.com/PaddlePaddle/Paddle/blob/b84e8226514b8bb4405c3c28e54aa5077193d179/paddle/scripts/travis/check_style.sh#L30), which is invoked by [our Travis CI configuration](https://github.com/PaddlePaddle/Paddle/blob/b84e8226514b8bb4405c3c28e54aa5077193d179/.travis.yml#L43).
### Unit Tests
Please remember to add related unit tests.
- For C/C++ code, please follow [`google-test` Primer](https://github.com/google/googletest/blob/master/googletest/docs/Primer.md).
- For Python code, please use [Python's standard `unittest` package](http://pythontesting.net/framework/unittest/unittest-introduction/).
### Writing Logs
We use [glog](https://github.com/google/glog) for logging in our C/C++ code.
For general information, please use `LOG`. For debug information, please use [`VLOG`](http://htmlpreview.github.io/?https://github.com/google/glog/blob/master/doc/glog.html#verbose). The reason is at [here](https://groups.google.com/a/chromium.org/d/msg/chromium-dev/3NDNd1KzXeY/AZKMMx37fdQJ).
`VLOG` requires a *verbose level* parameter. For example:
```c++
VLOG(3) << "Operator FC is taking " << num_inputs << "inputs."
```
When we run a PaddlePaddle application or test, we can specify a verbose threshold. For example:
```bash
GLOG_vmodule=buddy_allocator=2 \
GLOG_v=10 \
python \
../python/paddle/v2/framework/tests/test_recurrent_op.py
```
This will enable VLOG messages generated by `buddy_allocator.{h,cc}` and in the verbose range of 0 to 3, so you will see above example VLOG message, which is in level 3. This suggests that we output overall messages in lower verbose levels, so they display with higher probability. When coding C++, please follow the verbose level convention as follows:
- verbose level 1: [framework](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/framework)
- verbose level 3: [operators](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/operators)
- verbose level 5: [memory](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/memory), [platform](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/platform)
- verbose level 7: [math](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/math)
Dockerfile
浏览文件 @ 1ffdecf1
......@@ -10,13 +10,11 @@ RUN /bin/bash -c 'if [[ -n ${UBUNTU_MIRROR} ]]; then sed -i 's#http://archive.ub
ARG WITH_GPU
ARG WITH_AVX
ARG WITH_DOC
ARG WITH_STYLE_CHECK
ENV WOBOQ OFF
ENV WITH_GPU=${WITH_GPU:-OFF}
ENV WITH_GPU=${WITH_GPU:-ON}
ENV WITH_AVX=${WITH_AVX:-ON}
ENV WITH_DOC=${WITH_DOC:-OFF}
ENV WITH_STYLE_CHECK=${WITH_STYLE_CHECK:-OFF}
ENV HOME /root
# Add bash enhancements
......@@ -24,7 +22,7 @@ COPY ./paddle/scripts/docker/root/ /root/
RUN apt-get update && \
apt-get install -y \
git python-pip python-dev openssh-server bison \
git python-pip python-dev openssh-server bison libnccl-dev \
wget unzip unrar tar xz-utils bzip2 gzip coreutils ntp \
curl sed grep graphviz libjpeg-dev zlib1g-dev \
python-matplotlib gcc-4.8 g++-4.8 \
......@@ -71,20 +69,6 @@ RUN pip install -r /root/requirements.txt
RUN apt-get install -y libssl-dev libffi-dev
RUN pip install certifi urllib3[secure]
# TODO(qijun) The template library Eigen doesn't work well with GCC 5
# coming with the default Docker image, so we switch to use GCC 4.8
# by default. And I will check Eigen library later.
RUN ln -sf gcc-4.8 /usr/bin/gcc && \
ln -sf gcc-ar-4.8 /usr/bin/gcc-ar && \
ln -sf gcc-nm-4.8 /usr/bin/gcc-nm && \
ln -sf gcc-ranlib-4.8 /usr/bin/gcc-ranlib && \
ln -sf gcc-4.8 /usr/bin/x86_64-linux-gnu-gcc && \
ln -sf gcc-ar-4.8 /usr/bin/x86_64-linux-gnu-gcc-ar && \
ln -sf gcc-nm-4.8 /usr/bin/x86_64-linux-gnu-gcc-nm && \
ln -sf gcc-ranlib-4.8 /usr/bin/x86_64-linux-gnu-gcc-ranlib && \
ln -sf g++-4.8 /usr/bin/g++ && \
ln -sf g++-4.8 /usr/bin/x86_64-linux-gnu-g++
# Install woboq_codebrowser to /woboq
RUN git clone https://github.com/woboq/woboq_codebrowser /woboq && \
......
Dockerfile.android
浏览文件 @ 1ffdecf1
......@@ -4,9 +4,16 @@ MAINTAINER PaddlePaddle Authors <paddle-dev@baidu.com>
ARG UBUNTU_MIRROR
RUN /bin/bash -c 'if [[ -n ${UBUNTU_MIRROR} ]]; then sed -i 's#http://archive.ubuntu.com/ubuntu#${UBUNTU_MIRROR}#g' /etc/apt/sources.list; fi'
# ENV variables
ARG ANDROID_ABI
ARG ANDROID_API
ENV ANDROID_ABI=${ANDROID_ABI:-"armeabi-v7a"}
ENV ANDROID_API=${ANDROID_API:-21}
ENV HOME=/root \
ANDROID_NDK_HOME=/opt/android-ndk-linux \
ANDROID_STANDALONE_TOOLCHAIN=/opt/android-toolchain-gcc
ANDROID_TOOLCHAINS_DIR=/opt/toolchains
RUN apt-get update && \
apt-get install -y \
......@@ -15,12 +22,11 @@ RUN apt-get update && \
apt-get clean -y
# Install Go and glide
RUN wget -O go.tgz https://storage.googleapis.com/golang/go1.8.1.linux-amd64.tar.gz && \
tar -C /usr/local -xzf go.tgz && \
RUN wget -qO- go.tgz https://storage.googleapis.com/golang/go1.8.1.linux-amd64.tar.gz | \
tar -xz -C /usr/local && \
mkdir /root/gopath && \
mkdir /root/gopath/bin && \
mkdir /root/gopath/src && \
rm go.tgz
mkdir /root/gopath/src
ENV GOROOT=/usr/local/go GOPATH=/root/gopath
# should not be in the same line with GOROOT definition, otherwise docker build could not find GOROOT.
ENV PATH=${PATH}:${GOROOT}/bin:${GOPATH}/bin
......@@ -37,13 +43,12 @@ RUN pip install --upgrade pip && \
pip install pre-commit
# Android NDK
RUN mkdir /opt/android-ndk-tmp && \
RUN mkdir -p ${ANDROID_TOOLCHAINS_DIR} && \
mkdir -p /opt/android-ndk-tmp && \
cd /opt/android-ndk-tmp && \
wget -q https://dl.google.com/android/repository/android-ndk-r14b-linux-x86_64.zip && \
unzip -q android-ndk-r14b-linux-x86_64.zip && \
mv android-ndk-r14b ${ANDROID_NDK_HOME} && \
${ANDROID_NDK_HOME}/build/tools/make-standalone-toolchain.sh --arch=arm --platform=android-21 --install-dir=${ANDROID_STANDALONE_TOOLCHAIN} && \
rm -rf /opt/android-ndk-tmp && \
rm -rf ${ANDROID_NDK_HOME}
rm -rf /opt/android-ndk-tmp
CMD ["bash", "/paddle/paddle/scripts/docker/build_android.sh"]
README.md
浏览文件 @ 1ffdecf1
......@@ -51,19 +51,19 @@ Please refer to our [release announcement](https://github.com/PaddlePaddle/Paddl
- **Connected to Products**
In addition, PaddlePaddle is also designed to be easily deployable. At Baidu,
PaddlePaddle has been deployed into products or service with a vast number
PaddlePaddle has been deployed into products and services with a vast number
of users, including ad click-through rate (CTR) prediction, large-scale image
classification, optical character recognition(OCR), search ranking, computer
virus detection, recommendation, etc. It is widely utilized in products at
Baidu and it has achieved a significant impact. We hope you can also exploit
the capability of PaddlePaddle to make a huge impact for your product.
Baidu and it has achieved a significant impact. We hope you can also explore
the capability of PaddlePaddle to make an impact on your product.
## Installation
It is recommended to check out the
[Docker installation guide](http://doc.paddlepaddle.org/develop/doc/getstarted/build_and_install/docker_install_en.html)
before looking into the
[build from source guide](http://doc.paddlepaddle.org/develop/doc/getstarted/build_and_install/build_from_source_en.html)
[build from source guide](http://doc.paddlepaddle.org/develop/doc/getstarted/build_and_install/build_from_source_en.html).
## Documentation
......@@ -72,7 +72,7 @@ We provide [English](http://doc.paddlepaddle.org/develop/doc/) and
- [Deep Learning 101](http://book.paddlepaddle.org/index.html)
You might want to start from this online interactive book that can run in Jupyter Notebook.
You might want to start from this online interactive book that can run in a Jupyter Notebook.
- [Distributed Training](http://doc.paddlepaddle.org/develop/doc/howto/usage/cluster/cluster_train_en.html)
......
benchmark/IntelOptimizedPaddle.md 0 → 100644
浏览文件 @ 1ffdecf1
# Benchmark
Machine:
- Server
- Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz, 2 Sockets, 20 Cores per socket
- Laptop
- DELL XPS15-9560-R1745: i7-7700HQ 8G 256GSSD
- i5 MacBook Pro (Retina, 13-inch, Early 2015)
- Desktop
- i7-6700k
System: CentOS release 6.3 (Final), Docker 1.12.1.
PaddlePaddle: paddlepaddle/paddle:latest (TODO: will rerun after 0.11.0)
- MKL-DNN tag v0.10
- MKLML 2018.0.20170720
- OpenBLAS v0.2.20
On each machine, we will test and compare the performance of training on single node using MKL-DNN / MKLML / OpenBLAS respectively.
## Benchmark Model
### Server
Test on batch size 64, 128, 256 on Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz
Input image size - 3 * 224 * 224, Time: images/second
- VGG-19
| BatchSize | 64 | 128 | 256 |
|--------------|-------| -----| --------|
| OpenBLAS | 7.82 | 8.62 | 10.34 |
| MKLML | 11.02 | 12.86 | 15.33 |
| MKL-DNN | 27.69 | 28.8 | 29.27 |
chart on batch size 128
TBD
- ResNet
- GoogLeNet
### Laptop
TBD
### Desktop
TBD
benchmark/paddle/image/provider.py
浏览文件 @ 1ffdecf1
......@@ -22,5 +22,5 @@ def initHook(settings, height, width, color, num_class, **kwargs):
def process(settings, file_list):
for i in xrange(1024):
img = np.random.rand(1, settings.data_size).reshape(-1, 1).flatten()
lab = random.randint(0, settings.num_class)
lab = random.randint(0, settings.num_class - 1)
yield img.astype('float32'), int(lab)
benchmark/paddle/image/resnet.py 0 → 100644
浏览文件 @ 1ffdecf1
#!/usr/bin/env python
from paddle.trainer_config_helpers import *
height = 224
width = 224
num_class = 1000
batch_size = get_config_arg('batch_size', int, 64)
layer_num = get_config_arg("layer_num", int, 50)
is_test = get_config_arg("is_test", bool, False)
args = {'height': height, 'width': width, 'color': True, 'num_class': num_class}
define_py_data_sources2(
"train.list", None, module="provider", obj="process", args=args)
settings(
batch_size=batch_size,
learning_rate=0.01 / batch_size,
learning_method=MomentumOptimizer(0.9),
regularization=L2Regularization(0.0005 * batch_size))
#######################Network Configuration #############
def conv_bn_layer(name,
input,
filter_size,
num_filters,
stride,
padding,
channels=None,
active_type=ReluActivation()):
"""
A wrapper for conv layer with batch normalization layers.
Note:
conv layer has no activation.
"""
tmp = img_conv_layer(
name=name + "_conv",
input=input,
filter_size=filter_size,
num_channels=channels,
num_filters=num_filters,
stride=stride,
padding=padding,
act=LinearActivation(),
bias_attr=False)
return batch_norm_layer(
name=name + "_bn", input=tmp, act=active_type, use_global_stats=is_test)
def bottleneck_block(name, input, num_filters1, num_filters2):
"""
A wrapper for bottlenect building block in ResNet.
Last conv_bn_layer has no activation.
Addto layer has activation of relu.
"""
last_name = conv_bn_layer(
name=name + '_branch2a',
input=input,
filter_size=1,
num_filters=num_filters1,
stride=1,
padding=0)
last_name = conv_bn_layer(
name=name + '_branch2b',
input=last_name,
filter_size=3,
num_filters=num_filters1,
stride=1,
padding=1)
last_name = conv_bn_layer(
name=name + '_branch2c',
input=last_name,
filter_size=1,
num_filters=num_filters2,
stride=1,
padding=0,
active_type=LinearActivation())
return addto_layer(
name=name + "_addto", input=[input, last_name], act=ReluActivation())
def mid_projection(name, input, num_filters1, num_filters2, stride=2):
"""
A wrapper for middile projection in ResNet.
projection shortcuts are used for increasing dimensions,
and other shortcuts are identity
branch1: projection shortcuts are used for increasing
dimensions, has no activation.
branch2x: bottleneck building block, shortcuts are identity.
"""
# stride = 2
branch1 = conv_bn_layer(
name=name + '_branch1',
input=input,
filter_size=1,
num_filters=num_filters2,
stride=stride,
padding=0,
active_type=LinearActivation())
last_name = conv_bn_layer(
name=name + '_branch2a',
input=input,
filter_size=1,
num_filters=num_filters1,
stride=stride,
padding=0)
last_name = conv_bn_layer(
name=name + '_branch2b',
input=last_name,
filter_size=3,
num_filters=num_filters1,
stride=1,
padding=1)
last_name = conv_bn_layer(
name=name + '_branch2c',
input=last_name,
filter_size=1,
num_filters=num_filters2,
stride=1,
padding=0,
active_type=LinearActivation())
return addto_layer(
name=name + "_addto", input=[branch1, last_name], act=ReluActivation())
img = data_layer(name='image', size=height * width * 3)
def deep_res_net(res2_num=3, res3_num=4, res4_num=6, res5_num=3):
"""
A wrapper for 50,101,152 layers of ResNet.
res2_num: number of blocks stacked in conv2_x
res3_num: number of blocks stacked in conv3_x
res4_num: number of blocks stacked in conv4_x
res5_num: number of blocks stacked in conv5_x
"""
# For ImageNet
# conv1: 112x112
tmp = conv_bn_layer(
"conv1",
input=img,
filter_size=7,
channels=3,
num_filters=64,
stride=2,
padding=3)
tmp = img_pool_layer(name="pool1", input=tmp, pool_size=3, stride=2)
# conv2_x: 56x56
tmp = mid_projection(
name="res2_1", input=tmp, num_filters1=64, num_filters2=256, stride=1)
for i in xrange(2, res2_num + 1, 1):
tmp = bottleneck_block(
name="res2_" + str(i), input=tmp, num_filters1=64, num_filters2=256)
# conv3_x: 28x28
tmp = mid_projection(
name="res3_1", input=tmp, num_filters1=128, num_filters2=512)
for i in xrange(2, res3_num + 1, 1):
tmp = bottleneck_block(
name="res3_" + str(i),
input=tmp,
num_filters1=128,
num_filters2=512)
# conv4_x: 14x14
tmp = mid_projection(
name="res4_1", input=tmp, num_filters1=256, num_filters2=1024)
for i in xrange(2, res4_num + 1, 1):
tmp = bottleneck_block(
name="res4_" + str(i),
input=tmp,
num_filters1=256,
num_filters2=1024)
# conv5_x: 7x7
tmp = mid_projection(
name="res5_1", input=tmp, num_filters1=512, num_filters2=2048)
for i in xrange(2, res5_num + 1, 1):
tmp = bottleneck_block(
name="res5_" + str(i),
input=tmp,
num_filters1=512,
num_filters2=2048)
tmp = img_pool_layer(
name='avgpool',
input=tmp,
pool_size=7,
stride=1,
pool_type=AvgPooling())
return fc_layer(input=tmp, size=num_class, act=SoftmaxActivation())
if layer_num == 50:
resnet = deep_res_net(3, 4, 6, 3)
elif layer_num == 101:
resnet = deep_res_net(3, 4, 23, 3)
elif layer_num == 152:
resnet = deep_res_net(3, 8, 36, 3)
else:
print("Wrong layer number.")
lbl = data_layer(name="label", size=num_class)
loss = cross_entropy(name='loss', input=resnet, label=lbl)
inputs(img, lbl)
outputs(loss)
benchmark/paddle/image/run_mkldnn.sh 0 → 100755
浏览文件 @ 1ffdecf1
set -e
function train() {
unset OMP_NUM_THREADS MKL_NUM_THREADS
export OMP_DYNAMIC="FALSE"
export KMP_AFFINITY="granularity=fine,compact,0,0"
topology=$1
layer_num=$2
bs=$3
use_mkldnn=$4
if [ $4 == "True" ]; then
thread=1
log="logs/${topology}-${layer_num}-mkldnn-${bs}.log"
elif [ $4 == "False" ]; then
thread=`nproc`
# each trainer_count use only 1 core to avoid conflict
export OMP_NUM_THREADS=1
export MKL_NUM_THREADS=1
log="logs/${topology}-${layer_num}-${thread}mklml-${bs}.log"
else
echo "Wrong input $3, use True or False."
exit 0
fi
args="batch_size=${bs},layer_num=${layer_num}"
config="${topology}.py"
paddle train --job=time \
--config=$config \
--use_mkldnn=$use_mkldnn \
--use_gpu=False \
--trainer_count=$thread \
--log_period=10 \
--test_period=100 \
--config_args=$args \
2>&1 | tee ${log}
}
if [ ! -d "train.list" ]; then
echo " " > train.list
fi
if [ ! -d "logs" ]; then
mkdir logs
fi
for use_mkldnn in True False; do
for batchsize in 64 128 256; do
train vgg 19 $batchsize $use_mkldnn
train resnet 50 $batchsize $use_mkldnn
done
done
benchmark/paddle/image/vgg.py 0 → 100644
浏览文件 @ 1ffdecf1
#!/usr/bin/env python
from paddle.trainer_config_helpers import *
height = 224
width = 224
num_class = 1000
batch_size = get_config_arg('batch_size', int, 64)
layer_num = get_config_arg('layer_num', int, 19)
args = {'height': height, 'width': width, 'color': True, 'num_class': num_class}
define_py_data_sources2(
"train.list", None, module="provider", obj="process", args=args)
settings(
batch_size=batch_size,
learning_rate=0.001 / batch_size,
learning_method=MomentumOptimizer(0.9),
regularization=L2Regularization(0.0005 * batch_size))
img = data_layer(name='image', size=height * width * 3)
def vgg_network(vgg_num=3):
tmp = img_conv_group(
input=img,
num_channels=3,
conv_padding=1,
conv_num_filter=[64, 64],
conv_filter_size=3,
conv_act=ReluActivation(),
pool_size=2,
pool_stride=2,
pool_type=MaxPooling())
tmp = img_conv_group(
input=tmp,
conv_num_filter=[128, 128],
conv_padding=1,
conv_filter_size=3,
conv_act=ReluActivation(),
pool_stride=2,
pool_type=MaxPooling(),
pool_size=2)
channels = []
for i in range(vgg_num):
channels.append(256)
tmp = img_conv_group(
input=tmp,
conv_num_filter=channels,
conv_padding=1,
conv_filter_size=3,
conv_act=ReluActivation(),
pool_stride=2,
pool_type=MaxPooling(),
pool_size=2)
channels = []
for i in range(vgg_num):
channels.append(512)
tmp = img_conv_group(
input=tmp,
conv_num_filter=channels,
conv_padding=1,
conv_filter_size=3,
conv_act=ReluActivation(),
pool_stride=2,
pool_type=MaxPooling(),
pool_size=2)
tmp = img_conv_group(
input=tmp,
conv_num_filter=channels,
conv_padding=1,
conv_filter_size=3,
conv_act=ReluActivation(),
pool_stride=2,
pool_type=MaxPooling(),
pool_size=2)
tmp = fc_layer(
input=tmp,
size=4096,
act=ReluActivation(),
layer_attr=ExtraAttr(drop_rate=0.5))
tmp = fc_layer(
input=tmp,
size=4096,
act=ReluActivation(),
layer_attr=ExtraAttr(drop_rate=0.5))
return fc_layer(input=tmp, size=num_class, act=SoftmaxActivation())
if layer_num == 16:
vgg = vgg_network(3)
elif layer_num == 19:
vgg = vgg_network(4)
else:
print("Wrong layer number.")
lab = data_layer('label', num_class)
loss = cross_entropy(input=vgg, label=lab)
outputs(loss)
cmake/cblas.cmake
浏览文件 @ 1ffdecf1
# Find the CBlas and lapack libraries
#
# It will search MKL, atlas, OpenBlas, reference-cblas in order.
# It will search MKLML, atlas, OpenBlas, reference-cblas in order.
#
# If any cblas implementation found, the following variable will be set.
# CBLAS_PROVIDER # one of MKL, ATLAS, OPENBLAS, REFERENCE
# CBLAS_PROVIDER # one of MKLML, ATLAS, OPENBLAS, REFERENCE
# CBLAS_INC_DIR # the include directory for cblas.
# CBLAS_LIBS # a list of libraries should be linked by paddle.
# # Each library should be full path to object file.
#
# User should set one of MKL_ROOT, ATLAS_ROOT, OPENBLAS_ROOT, REFERENCE_CBLAS_ROOT
# during cmake. If none of them set, it will try to find cblas implementation in
# system paths.
#
set(CBLAS_FOUND OFF)
......@@ -30,44 +25,6 @@ if(WITH_MKLML AND MKLML_INC_DIR AND MKLML_LIB)
return()
endif()
## Then find MKL.
set(INTEL_MKL_ROOT "/opt/intel/mkl" CACHE PATH "Folder contains intel mkl libs")
set(MKL_ROOT $ENV{MKL_ROOT} CACHE PATH "Folder contains env MKL")
set(MKL_INCLUDE_SEARCH_PATHS
${MKL_ROOT}/include
${INTEL_MKL_ROOT}/include)
set(MKL_LIB_SEARCH_PATHS
${MKL_ROOT}/lib
${MKL_ROOT}/lib/intel64
${INTEL_MKL_ROOT}/lib
${INTEL_MKL_ROOT}/lib/intel64)
find_path(MKL_INC_DIR mkl.h PATHS
${MKL_INCLUDE_SEARCH_PATHS})
find_path(MKL_LAPACK_INC_DIR mkl_lapacke.h PATHS
${MKL_INCLUDE_SEARCH_PATHS})
find_library(MKL_CORE_LIB NAMES mkl_core PATHS
${MKL_LIB_SEARCH_PATHS})
find_library(MKL_SEQUENTIAL_LIB NAMES mkl_sequential PATHS
${MKL_LIB_SEARCH_PATHS})
find_library(MKL_INTEL_LP64 NAMES mkl_intel_lp64 PATHS
${MKL_LIB_SEARCH_PATHS})
if(MKL_LAPACK_INC_DIR AND MKL_INC_DIR AND MKL_CORE_LIB AND MKL_SEQUENTIAL_LIB AND MKL_INTEL_LP64)
set(CBLAS_FOUND ON)
set(CBLAS_PROVIDER MKL)
set(CBLAS_INC_DIR ${MKL_INC_DIR} ${MKL_LAPACK_INC_DIR})
set(CBLAS_LIBRARIES ${MKL_INTEL_LP64} ${MKL_SEQUENTIAL_LIB} ${MKL_CORE_LIB})
add_definitions(-DPADDLE_USE_MKL)
add_definitions(-DLAPACK_FOUND)
message(STATUS "Found MKL (include: ${MKL_INC_DIR}, library: ${CBLAS_LIBRARIES})")
message(STATUS "Found lapack in MKL (include: ${MKL_LAPACK_INC_DIR})")
return()
endif()
## Then find atlas.
set(ATLAS_ROOT $ENV{ATLAS_ROOT} CACHE PATH "Folder contains Atlas")
set(ATLAS_INCLUDE_SEARCH_PATHS
......@@ -171,3 +128,10 @@ if (REFERENCE_CBLAS_INCLUDE_DIR AND REFERENCE_CBLAS_LIBRARY)
add_definitions(-DPADDLE_USE_REFERENCE_CBLAS)
message(STATUS "Found reference-cblas (include: ${CBLAS_INC_DIR}, library: ${CBLAS_LIBRARIES})")
endif()
if(IOS_USE_VECLIB_FOR_BLAS AND VECLIB_FOUND)
set(CBLAS_FOUND ON)
set(CBLAS_PROVIDER vecLib)
set(CBLAS_INC_DIR ${VECLIB_INC_DIR})
add_definitions(-DPADDLE_USE_VECLIB)
endif()
cmake/configure.cmake
浏览文件 @ 1ffdecf1
......@@ -24,10 +24,18 @@ if(WITH_DOUBLE)
add_definitions(-DPADDLE_TYPE_DOUBLE)
endif(WITH_DOUBLE)
if(WITH_TESTING)
add_definitions(-DPADDLE_WITH_TESTING)
endif(WITH_TESTING)
if(NOT WITH_TIMER)
add_definitions(-DPADDLE_DISABLE_TIMER)
endif(NOT WITH_TIMER)
if(USE_EIGEN_FOR_BLAS)
add_definitions(-DPADDLE_USE_EIGEN_FOR_BLAS)
endif(USE_EIGEN_FOR_BLAS)
if(NOT WITH_PROFILER)
add_definitions(-DPADDLE_DISABLE_PROFILER)
endif(NOT WITH_PROFILER)
......@@ -45,19 +53,20 @@ if(NOT WITH_GOLANG)
endif(NOT WITH_GOLANG)
if(NOT WITH_GPU)
add_definitions(-DPADDLE_ONLY_CPU)
add_definitions(-DHPPL_STUB_FUNC)
list(APPEND CMAKE_CXX_SOURCE_FILE_EXTENSIONS cu)
else()
add_definitions(-DPADDLE_WITH_CUDA)
FIND_PACKAGE(CUDA REQUIRED)
if(${CUDA_VERSION_MAJOR} VERSION_LESS 7)
message(FATAL_ERROR "Paddle need CUDA >= 7.0 to compile")
message(FATAL_ERROR "Paddle needs CUDA >= 7.0 to compile")
endif()
if(NOT CUDNN_FOUND)
message(FATAL_ERROR "Paddle need cudnn to compile")
message(FATAL_ERROR "Paddle needs cudnn to compile")
endif()
set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} "-Xcompiler ${SIMD_FLAG}")
......
cmake/cpplint.cmake
浏览文件 @ 1ffdecf1
......@@ -26,9 +26,9 @@ set(IGNORE_PATTERN
.*ImportanceSampler.*
.*cblas\\.h.*
.*\\.pb\\.txt
.*LtrDataProvider.*
.*MultiDataProvider.*
.*pb.*)
.*pb.*
.*pybind.h)
# add_style_check_target
#
......
cmake/cross_compiling/android.cmake
浏览文件 @ 1ffdecf1
......@@ -20,6 +20,7 @@
# The supported variables are listed belows:
#
# ANDROID_STANDALONE_TOOLCHAIN
# ANDROID_TOOLCHAIN
# ANDROID_ABI
# ANDROID_NATIVE_API_LEVEL
# ANDROID_ARM_MODE
......@@ -57,6 +58,10 @@ IF(NOT DEFINED CMAKE_SYSTEM_VERSION AND ANDROID_NATIVE_API_LEVEL)
ENDIF()
ENDIF()
IF(NOT DEFINED ANDROID_TOOLCHAIN)
SET(ANDROID_TOOLCHAIN clang)
ENDIF()
IF(NOT DEFINED ANDROID_ABI)
SET(ANDROID_ABI "armeabi-v7a")
ENDIF()
......@@ -82,6 +87,7 @@ IF("${CMAKE_VERSION}" VERSION_LESS "3.7.0")
"${CMAKE_VERSION}), when cross-compiling for Android.")
IF(ANDROID_STANDALONE_TOOLCHAIN)
# Use standalone toolchain
SET(CMAKE_SYSROOT "${ANDROID_STANDALONE_TOOLCHAIN}/sysroot")
IF(NOT CMAKE_SYSTEM_VERSION)
......@@ -96,26 +102,44 @@ IF("${CMAKE_VERSION}" VERSION_LESS "3.7.0")
ENDIF()
# Toolchain
SET(ANDROID_TOOLCHAIN "gcc")
SET(ANDROID_TOOLCHAIN_ROOT ${ANDROID_STANDALONE_TOOLCHAIN})
IF(ANDROID_ABI MATCHES "^armeabi(-v7a)?$")
SET(ANDROID_TOOLCHAIN_NAME arm-linux-androideabi)
IF(ANDROID_ABI STREQUAL "armeabi")
SET(CMAKE_SYSTEM_PROCESSOR armv5te)
ELSEIF(ANDROID_ABI STREQUAL "armeabi-v7a")
SET(CMAKE_SYSTEM_PROCESSOR armv7-a)
ENDIF()
ENDIF()
IF(ANDROID_ABI STREQUAL "arm64-v8a")
SET(ANDROID_TOOLCHAIN_NAME aarch64-linux-android)
SET(CMAKE_SYSTEM_PROCESSOR aarch64)
ELSE(ANDROID_NDK)
# TODO: use android ndk
ENDIF()
IF(ANDROID_ABI MATCHES "^armeabi(-v7a)?$")
SET(ANDROID_TOOLCHAIN_NAME arm-linux-androideabi)
IF(ANDROID_ABI STREQUAL "armeabi")
SET(CMAKE_SYSTEM_PROCESSOR armv5te)
SET(ANDROID_CLANG_TRIPLE armv5te-none-linux-androideabi)
ELSEIF(ANDROID_ABI STREQUAL "armeabi-v7a")
SET(CMAKE_SYSTEM_PROCESSOR armv7-a)
SET(ANDROID_CLANG_TRIPLE armv7-none-linux-androideabi)
ENDIF()
SET(ANDROID_TOOLCHAIN_PREFIX "${ANDROID_TOOLCHAIN_ROOT}/bin/${ANDROID_TOOLCHAIN_NAME}-")
ELSEIF(ANDROID_ABI STREQUAL "arm64-v8a")
SET(ANDROID_TOOLCHAIN_NAME aarch64-linux-android)
SET(CMAKE_SYSTEM_PROCESSOR aarch64)
SET(ANDROID_CLANG_TRIPLE aarch64-none-linux-android)
ELSE()
MESSAGE(FATAL_ERROR "Invalid Android ABI: ${ANDROID_ABI}.")
ENDIF()
SET(ANDROID_TOOLCHAIN_PREFIX "${ANDROID_TOOLCHAIN_ROOT}/bin/${ANDROID_TOOLCHAIN_NAME}-")
IF(ANDROID_TOOLCHAIN STREQUAL clang)
SET(ANDROID_C_COMPILER_NAME clang)
SET(ANDROID_CXX_COMPILER_NAME clang++)
SET(CMAKE_C_COMPILER_TARGET ${ANDROID_CLANG_TRIPLE})
SET(CMAKE_CXX_COMPILER_TARGET ${ANDROID_CLANG_TRIPLE})
ELSEIF(ANDROID_TOOLCHAIN STREQUAL gcc)
SET(ANDROID_C_COMPILER_NAME gcc)
SET(ANDROID_CXX_COMPILER_NAME g++)
ELSE()
MESSAGE(FATAL_ERROR "Invalid Android toolchain: ${ANDROID_TOOLCHAIN}")
ENDIF()
# C compiler
IF(NOT CMAKE_C_COMPILER)
SET(ANDROID_C_COMPILER "${ANDROID_TOOLCHAIN_PREFIX}gcc")
SET(ANDROID_C_COMPILER "${ANDROID_TOOLCHAIN_PREFIX}${ANDROID_C_COMPILER_NAME}")
ELSE()
GET_FILENAME_COMPONENT(ANDROID_C_COMPILER ${CMAKE_C_COMPILER} PROGRAM)
ENDIF()
......@@ -125,7 +149,7 @@ IF("${CMAKE_VERSION}" VERSION_LESS "3.7.0")
# CXX compiler
IF(NOT CMAKE_CXX_COMPILER)
SET(ANDROID_CXX_COMPILER "${ANDROID_TOOLCHAIN_PREFIX}g++")
SET(ANDROID_CXX_COMPILER "${ANDROID_TOOLCHAIN_PREFIX}${ANDROID_CXX_COMPILER_NAME}")
ELSE()
GET_FILENAME_COMPONENT(ANDROID_CXX_COMPILER ${CMAKE_CXX_COMPILER} PROGRAM)
ENDIF()
......@@ -137,7 +161,7 @@ IF("${CMAKE_VERSION}" VERSION_LESS "3.7.0")
SET(CMAKE_CXX_COMPILER ${ANDROID_CXX_COMPILER} CACHE PATH "CXX compiler" FORCE)
# Toolchain and ABI specific flags.
SET(ANDROID_COMPILER_FLAGS "-ffunction-sections -fdata-sections -finline-limit=64")
SET(ANDROID_COMPILER_FLAGS "-ffunction-sections -fdata-sections")
SET(ANDROID_LINKER_FLAGS "-Wl,--gc-sections")
IF(ANDROID_ABI STREQUAL "armeabi")
......@@ -145,8 +169,7 @@ IF("${CMAKE_VERSION}" VERSION_LESS "3.7.0")
-march=armv5te
-mtune=xscale
-msoft-float)
ENDIF()
IF(ANDROID_ABI STREQUAL "armeabi-v7a")
ELSEIF(ANDROID_ABI STREQUAL "armeabi-v7a")
LIST(APPEND ANDROID_COMPILER_FLAGS
-march=armv7-a
-mfloat-abi=softfp)
......@@ -156,6 +179,8 @@ IF("${CMAKE_VERSION}" VERSION_LESS "3.7.0")
LIST(APPEND ANDROID_COMPILER_FLAGS -mfpu=vfpv3-d16)
ENDIF()
LIST(APPEND ANDROID_LINKER_FLAGS -Wl,--fix-cortex-a8)
ELSEIF(ANDROID_ABI STREQUAL "arm64-v8a")
LIST(APPEND ANDROID_COMPILER_FLAGS -march=armv8-a)
ENDIF()
IF(ANDROID_ABI MATCHES "^armeabi(-v7a)?$")
......@@ -164,10 +189,18 @@ IF("${CMAKE_VERSION}" VERSION_LESS "3.7.0")
ELSE()
LIST(APPEND ANDROID_COMPILER_FLAGS -mthumb)
ENDIF()
IF(ANDROID_TOOLCHAIN STREQUAL clang)
# Disable integrated-as for better compatibility.
LIST(APPEND ANDROID_COMPILER_FLAGS -fno-integrated-as)
ENDIF()
ENDIF()
IF(ANDROID_ABI STREQUAL "arm64-v8a")
LIST(APPEND ANDROID_COMPILER_FLAGS -march=armv8-a)
IF(ANDROID_TOOLCHAIN STREQUAL clang)
# CMake automatically forwards all compiler flags to the linker,
# and clang doesn't like having -Wa flags being used for linking.
# To prevent CMake from doing this would require meddling with
# the CMAKE_<LANG>_COMPILE_OBJECT rules, which would get quite messy.
LIST(APPEND ANDROID_LINKER_FLAGS -Qunused-arguments)
ENDIF()
STRING(REPLACE ";" " " ANDROID_COMPILER_FLAGS "${ANDROID_COMPILER_FLAGS}")
......
cmake/cross_compiling/ios.cmake 0 → 100644
浏览文件 @ 1ffdecf1
# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# This is a toolchain file for cross-compiling for iOS, and the
# configuration largely refers to public toolchain file:
# https://raw.githubusercontent.com/leetal/ios-cmake/master/ios.toolchain.cmake
# and
# https://github.com/cristeab/ios-cmake
#
# Supports options:
# IOS_PLATFORM = OS (default) or SIMULATOR
# This decides if SDKS will be selected from the iPhoneOS.platform or iPhoneSimulator.platform folders
# OS - the default, used to build for iPhone and iPad physical devices, which have an arm arch.
# SIMULATOR - used to build for the Simulator platforms, which have an x86 arch.
# IOS_ARCH
# The archectures wanted to support, such "arm64", "armv7;arm64"
# IOS_DEPLOYMENT_TARGET
# The minimum iOS deployment version, such as "7.0"
# IOS_ENABLE_BITCODE = ON (default) or OFF
# IOS_USE_VECLIB_FOR_BLAS = OFF (default) or ON
# IOS_DEVELOPER_ROOT = automatic(default) or /path/to/platform/Developer folder
# By default this location is automatcially chosen based on the IOS_PLATFORM value above.
# If set manually, it will override the default location and force the user of a particular Developer Platform
# IOS_SDK_ROOT = automatic(default) or /path/to/platform/Developer/SDKs/SDK folder
# By default this location is automatcially chosen based on the IOS_DEVELOPER_ROOT value.
# In this case it will always be the most up-to-date SDK found in the IOS_DEVELOPER_ROOT path.
# If set manually, this will force the use of a specific SDK version
# Macros:
# set_xcode_property (TARGET XCODE_PROPERTY XCODE_VALUE)
# A convenience macro for setting xcode specific properties on targets
# example: set_xcode_property (myioslib IPHONEOS_DEPLOYMENT_TARGET "3.1")
# find_host_package (PROGRAM ARGS)
# A macro used to find executable programs on the host system, not within the iOS environment.
# Thanks to the android-cmake project for providing the command
if(NOT IOS)
return()
endif()
set(CMAKE_SYSTEM_NAME Darwin)
# Get the Xcode version being used.
execute_process(COMMAND xcodebuild -version
OUTPUT_VARIABLE XCODE_VERSION
RESULT_VARIABLE XCODE_VERSION_RESULT
ERROR_QUIET OUTPUT_STRIP_TRAILING_WHITESPACE)
if(NOT ${XCODE_VERSION_RESULT})
string(REGEX MATCH "Xcode [0-9\\.]+" XCODE_VERSION "${XCODE_VERSION}")
string(REGEX REPLACE "Xcode ([0-9\\.]+)" "\\1" XCODE_VERSION "${XCODE_VERSION}")
message(STATUS "Building with Xcode version: ${XCODE_VERSION}")
else()
message(FATAL_ERROR "Cannot execute xcodebuild, please check whether xcode is installed.")
endif()
# Required as of cmake 2.8.10
set(CMAKE_OSX_DEPLOYMENT_TARGET "" CACHE STRING "Force unset of the deployment target for iOS" FORCE)
# Setup iOS platform unless specified manually with IOS_PLATFORM
if(NOT DEFINED IOS_PLATFORM)
set(IOS_PLATFORM "OS")
endif()
set(IOS_PLATFORM ${IOS_PLATFORM} CACHE STRING "Type of iOS Platform")
# Set the architecture for iOS
if(NOT DEFINED IOS_ARCH)
if(IOS_PLATFORM STREQUAL "OS")
# FIXME(liuyiqun): support "armv7;armv7s;arm64" future
set(IOS_ARCH "arm64")
elseif(IOS_PLATFORM STREQUAL "SIMULATOR")
# FIXME(liuyiqun): support "i386;x86_64" future
set(IOS_ARCH "x86_64")
endif()
endif()
set(CMAKE_OSX_ARCHITECTURES ${IOS_ARCH} CACHE string "Build architecture for iOS")
# Specify minimum iOS deployment version
if(NOT DEFINED IOS_DEPLOYMENT_TARGET)
set(IOS_DEPLOYMENT_TARGET "7.0")
endif()
set(IOS_DEPLOYMENT_TARGET ${IOS_DEPLOYMENT_TARGET} CACHE STRING "Minimum iOS version")
# Whether to enable bitcode
if(NOT DEFINED IOS_ENABLE_BITCODE)
set(IOS_ENABLE_BITCODE ON)
endif()
set(IOS_ENABLE_BITCODE ${IOS_ENABLE_BITCODE} CACHE BOOL "Whether to enable bitcode")
if(NOT DEFINED IOS_USE_VECLIB_FOR_BLAS)
set(IOS_USE_VECLIB_FOR_BLAS OFF)
endif()
set(IOS_USE_VECLIB_FOR_BLAS ${IOS_UES_VECLIB_FOR_BLAS} CACHE BOOL "Whether to use veclib")
# Check the platform selection and setup for developer root
if(${IOS_PLATFORM} STREQUAL "OS")
set(IOS_PLATFORM_LOCATION "iPhoneOS.platform")
set(XCODE_IOS_PLATFORM iphoneos)
# This causes the installers to properly locate the output libraries
set(CMAKE_XCODE_EFFECTIVE_PLATFORMS "-iphoneos")
elseif(${IOS_PLATFORM} STREQUAL "SIMULATOR")
set(IOS_PLATFORM_LOCATION "iPhoneSimulator.platform")
set(XCODE_IOS_PLATFORM iphonesimulator)
# This causes the installers to properly locate the output libraries
set(CMAKE_XCODE_EFFECTIVE_PLATFORMS "-iphonesimulator")
elseif(${IOS_PLATFORM} STREQUAL "WATCHOS")
set(IOS_PLATFORM_LOCATION "WatchOS.platform")
set(XCODE_IOS_PLATFORM watchos)
# This causes the installers to properly locate the output libraries
set(CMAKE_XCODE_EFFECTIVE_PLATFORMS "-watchos")
else(${IOS_PLATFORM} STREQUAL "OS")
message(FATAL_ERROR "Unsupported IOS_PLATFORM value selected. Please set to\n"
"\t OS, SIMULATOR, or WATCHOS.")
endif()
# Check iOS developer toolchain
if(NOT DEFINED IOS_DEVELOPER_ROOT)
# Setup iOS developer location
execute_process(COMMAND xcode-select -print-path
OUTPUT_VARIABLE XCODE_DEVELOPER_DIR
RESULT_VARIABLE XCODE_DEVELOPER_DIR_RESULT
ERROR_QUIET OUTPUT_STRIP_TRAILING_WHITESPACE)
# Xcode 4.3 changed the installation location, choose the most recent one available
if(${XCODE_VERSION} VERSION_LESS "4.3.0")
set(IOS_DEVELOPER_ROOT "/Developer/Platforms/${IOS_PLATFORM_LOCATION}/Developer")
else()
set(IOS_DEVELOPER_ROOT "${XCODE_DEVELOPER_DIR}/Platforms/${IOS_PLATFORM_LOCATION}/Developer")
endif()
endif()
if(EXISTS ${IOS_DEVELOPER_ROOT})
set(IOS_DEVELOPER_ROOT ${IOS_DEVELOPER_ROOT} CACHE PATH "Location of iOS Platform")
else()
message(FATAL_ERROR "Invalid IOS_DEVELOPER_ROOT: ${IOS_DEVELOPER_ROOT} does not exist.")
endif()
# Check iOS SDK
if(NOT DEFINED IOS_SDK_ROOT)
# Find and use the most recent iOS sdk
file(GLOB IOS_SDK_LISTS "${IOS_DEVELOPER_ROOT}/SDKs/*")
if(IOS_SDK_LISTS)
list(SORT IOS_SDK_LISTS)
list(REVERSE IOS_SDK_LISTS)
list(GET IOS_SDK_LISTS 0 IOS_SDK_ROOT)
else(IOS_SDK_LISTS)
message(FATAL_ERROR "No iOS SDK's found in default search path ${IOS_DEVELOPER_ROOT}."
" Please manually set IOS_SDK_ROOT or install the iOS SDK.")
endif(IOS_SDK_LISTS)
endif()
if(EXISTS ${IOS_SDK_ROOT})
set(IOS_SDK_ROOT ${IOS_SDK_ROOT} CACHE PATH "Location of the selected iOS SDK")
message(STATUS "iOS toolchain: ${IOS_SDK_ROOT}")
else()
message(FATAL_ERROR "Invalid IOS_SDK_ROOT: ${IOS_SDK_ROOT} does not exist.")
endif()
# Set the sysroot default to the most recent SDK
set(CMAKE_OSX_SYSROOT ${IOS_SDK_ROOT} CACHE PATH "Sysroot used for iOS support")
# Get version of iOS SDK
execute_process(COMMAND xcodebuild -sdk ${CMAKE_OSX_SYSROOT} -version SDKVersion
OUTPUT_VARIABLE IOS_SDK_VERSION
RESULT_VARIABLE IOS_SDK_VERSION_RESULT
ERROR_QUIET OUTPUT_STRIP_TRAILING_WHITESPACE)
if(${IOS_SDK_VERSION_RESULT})
string(REGEX MATCH "(([0-9]+)\\.)+([0-9]+)" IOS_SDK_VERSION "${IOS_SDK_ROOT}")
endif()
if(NOT IOS_SDK_VERSION)
message(WARNING "Cannot get SDK's version.")
set(IOS_SDK_VERSION 1)
endif()
set(CMAKE_SYSTEM_VERSION ${IOS_SDK_VERSION})
# Find the C & C++ compilers for the specified SDK.
if(NOT CMAKE_C_COMPILER)
# Default to use clang
execute_process(COMMAND xcrun -sdk ${CMAKE_OSX_SYSROOT} -find clang
OUTPUT_VARIABLE IOS_C_COMPILER
RESULT_VARIABLE IOS_C_COMPILER_RESULT
ERROR_QUIET OUTPUT_STRIP_TRAILING_WHITESPACE)
if(${IOS_C_COMPILER_RESULT})
get_filename_component(IOS_C_COMPILER clang PROGRAM)
endif()
else(NOT CMAKE_C_COMPILER)
# User can set it in cmake command
get_filename_component(IOS_C_COMPILER ${CMAKE_C_COMPILER} PROGRAM)
endif(NOT CMAKE_C_COMPILER)
if(NOT EXISTS ${IOS_C_COMPILER})
message(FATAL_ERROR "Cannot find C compiler: ${IOS_C_COMPILER}")
endif()
if(NOT CMAKE_CXX_COMPILER)
# Default to use clang++
execute_process(COMMAND xcrun -sdk ${CMAKE_OSX_SYSROOT} -find clang++
OUTPUT_VARIABLE IOS_CXX_COMPILER
RESULT_VARIABLE IOS_CXX_COMPILER_RESULT
ERROR_QUIET OUTPUT_STRIP_TRAILING_WHITESPACE)
if(${IOS_CXX_COMPILER_RESULT})
get_filename_component(IOS_CXX_COMPILER clang++ PROGRAM)
endif()
else(NOT CMAKE_CXX_COMPILER)
# User can set it in cmake command
get_filename_component(IOS_CXX_COMPILER ${CMAKE_CXX_COMPILER} PROGRAM)
endif(NOT CMAKE_CXX_COMPILER)
if(NOT EXISTS ${IOS_CXX_COMPILER})
message(FATAL_ERROR "Cannot find CXX compiler: ${IOS_CXX_COMPILER}")
endif()
set(CMAKE_C_COMPILER ${IOS_C_COMPILER} CACHE PATH "C compiler" FORCE)
set(CMAKE_CXX_COMPILER ${IOS_CXX_COMPILER} CACHE PATH "CXX compiler" FORCE)
set(CMAKE_C_OSX_COMPATIBILITY_VERSION_FLAG "-compatibility_version ")
set(CMAKE_C_OSX_CURRENT_VERSION_FLAG "-current_version ")
set(CMAKE_CXX_OSX_COMPATIBILITY_VERSION_FLAG "${CMAKE_C_OSX_COMPATIBILITY_VERSION_FLAG}")
set(CMAKE_CXX_OSX_CURRENT_VERSION_FLAG "${CMAKE_C_OSX_CURRENT_VERSION_FLAG}")
# Set iOS specific C/C++ flags
if(IOS_PLATFORM STREQUAL "OS")
if(XCODE_VERSION VERSION_LESS "7.0")
set(XCODE_IOS_PLATFORM_VERSION_FLAGS "-mios-version-min=${IOS_DEPLOYMENT_TARGET}")
else()
# Xcode 7.0+ uses flags we can build directly from XCODE_IOS_PLATFORM.
set(XCODE_IOS_PLATFORM_VERSION_FLAGS "-m${XCODE_IOS_PLATFORM}-version-min=${IOS_DEPLOYMENT_TARGET}")
endif()
else()
set(XCODE_IOS_FLATFORM_VERSION_FLAGS "-mios-simulator-version-min=${IOS_DEPLOYMENT_TARGET}")
endif()
if(IOS_ENABLE_BITCODE)
set(XCODE_IOS_BITCODE_FLAGS "${IOS_COMPILER_FLAGS} -fembed-bitcode")
else()
set(XCODE_IOS_BITCODE_FLAGS "")
endif()
set(IOS_COMPILER_FLAGS "${XCODE_IOS_PLATFORM_VERSION_FLAGS} ${XCODE_IOS_BITCODE_FLAGS}")
# Hidden visibilty is required for cxx on iOS
set(CMAKE_C_FLAGS "${IOS_COMPILER_FLAGS} ${CMAKE_C_FLAGS}" CACHE STRING "C flags")
set(CMAKE_CXX_FLAGS "${IOS_COMPILER_FLAGS} -fvisibility-inlines-hidden ${CMAKE_CXX_FLAGS}" CACHE STRING "CXX flags")
set(IOS_LINK_FLAGS "${XCODE_IOS_PLATFORM_VERSION_FLAGS} -Wl,-search_paths_first")
if(IOS_USE_VECLIB_FOR_BLAS)
# Find vecLib for iOS
set(VECLIB_SEARCH_DIRS
${IOS_SDK_ROOT}/System/Library/Frameworks/Accelerate.framework/Versions/Current/Frameworks
${IOS_SDK_ROOT}/System/Library/Frameworks/Accelerate.framework/Frameworks
)
find_path(VECLIB_INC_DIR vecLib.h PATHS ${VECLIB_SEARCH_DIRS}/vecLib.framework/Headers)
include(FindPackageHandleStandardArgs)
find_package_handle_standard_args(vecLib DEFAULT_MSG VECLIB_INC_DIR)
if(VECLIB_FOUND)
if(VECLIB_INC_DIR MATCHES "^/System/Library/Frameworks/vecLib.framework.*")
set(IOS_LINK_FLAGS ${IOS_LINK_FLAGS} -lcblas "-framework vecLib")
message(STATUS "Found standalone vecLib.framework")
else()
set(IOS_LINK_FLAGS ${IOS_LINK_FLAGS} -lcblas "-framework Accelerate")
message(STATUS "Found vecLib as part of Accelerate.framework")
endif()
endif()
endif()
set(CMAKE_C_LINK_FLAGS "${IOS_LINK_FLAGS} ${CMAKE_C_LINK_FLAGS}")
set(CMAKE_CXX_LINK_FLAGS "${IOS_LINK_FLAGS} ${CMAKE_CXX_LINK_FLAGS}")
set(CMAKE_PLATFORM_HAS_INSTALLNAME 1)
if(NOT IOS_ENABLE_BITCODE)
set(CMAKE_SHARED_LIBRARY_CREATE_C_FLAGS "-dynamiclib -headerpad_max_install_names")
set(CMAKE_SHARED_MODULE_CREATE_C_FLAGS "-bundle -headerpad_max_install_names")
else()
set(CMAKE_SHARED_LIBRARY_CREATE_C_FLAGS "-dynamiclib")
set(CMAKE_SHARED_MODULE_CREATE_C_FLAGS "-bundle")
endif()
set(CMAKE_SHARED_MODULE_LOADER_C_FLAG "-Wl,-bundle_loader,")
set(CMAKE_SHARED_MODULE_LOADER_CXX_FLAG "-Wl,-bundle_loader,")
set(CMAKE_FIND_LIBRARY_SUFFIXES ".dylib" ".so" ".a")
# hack: if a new cmake (which uses CMAKE_INSTALL_NAME_TOOL) runs on an old build tree
# (where install_name_tool was hardcoded) and where CMAKE_INSTALL_NAME_TOOL isn't in the cache
# and still cmake didn't fail in CMakeFindBinUtils.cmake (because it isn't rerun)
# hardcode CMAKE_INSTALL_NAME_TOOL here to install_name_tool, so it behaves as it did before, Alex
if(NOT DEFINED CMAKE_INSTALL_NAME_TOOL)
find_program(CMAKE_INSTALL_NAME_TOOL install_name_tool)
endif()
# Set the find root to the iOS developer roots and to user defined paths
set(CMAKE_FIND_ROOT_PATH ${IOS_DEVELOPER_ROOT} ${IOS_SDK_ROOT} ${CMAKE_PREFIX_PATH}
CACHE string "iOS find search path root")
# default to searching for frameworks first
set(CMAKE_FIND_FRAMEWORK FIRST)
# set up the default search directories for frameworks
set(CMAKE_SYSTEM_FRAMEWORK_PATH
${IOS_SDK_ROOT}/System/Library/Frameworks
${IOS_SDK_ROOT}/System/Library/PrivateFrameworks
${IOS_SDK_ROOT}/Developer/Library/Frameworks
)
# only search the iOS sdks, not the remainder of the host filesystem
set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER)
set(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY ONLY)
set(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE ONLY)
message(STATUS "iOS: Targeting iOS '${CMAKE_SYSTEM_VERSION}', "
"building for '${IOS_PLATFORM}' platform, with architecture '${CMAKE_OSX_ARCHITECTURES}'")
message(STATUS "System CMAKE_C_FLAGS: ${CMAKE_C_FLAGS}")
message(STATUS "System CMAKE_CXX_FLAGS: ${CMAKE_CXX_FLAGS}")
# Used in ExternalProject command
string(REPLACE ";" "\\$<SEMICOLON>" EXTERNAL_IOS_ARCHITECTURES "${CMAKE_OSX_ARCHITECTURES}")
set(EXTERNAL_OPTIONAL_ARGS
-DCMAKE_OSX_SYSROOT=${CMAKE_OSX_SYSROOT}
-DCMAKE_OSX_ARCHITECTURES=${EXTERNAL_IOS_ARCHITECTURES})
# This little macro lets you set any XCode specific property
macro(set_xcode_property TARGET XCODE_PROPERTY XCODE_VALUE)
set_property (TARGET ${TARGET} PROPERTY XCODE_ATTRIBUTE_${XCODE_PROPERTY} ${XCODE_VALUE})
endmacro(set_xcode_property)
# This macro lets you find executable programs on the host system
macro(find_host_package)
set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER)
set(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY NEVER)
set(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE NEVER)
set(IOS FALSE)
find_package(${ARGN})
set(IOS TRUE)
set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM ONLY)
set(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY ONLY)
set(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE ONLY)
endmacro(find_host_package)
cmake/cudnn.cmake
浏览文件 @ 1ffdecf1
......@@ -2,7 +2,7 @@ if(NOT WITH_GPU)
return()
endif()
set(CUDNN_ROOT "" CACHE PATH "CUDNN ROOT")
set(CUDNN_ROOT "/usr" CACHE PATH "CUDNN ROOT")
find_path(CUDNN_INCLUDE_DIR cudnn.h
PATHS ${CUDNN_ROOT} ${CUDNN_ROOT}/include
$ENV{CUDNN_ROOT} $ENV{CUDNN_ROOT}/include ${CUDA_TOOLKIT_INCLUDE}
......
cmake/external/eigen.cmake
浏览文件 @ 1ffdecf1
......@@ -8,7 +8,7 @@ ExternalProject_Add(
extern_eigen3
${EXTERNAL_PROJECT_LOG_ARGS}
GIT_REPOSITORY "https://github.com/RLovelett/eigen.git"
GIT_TAG "master"
GIT_TAG 70661066beef694cadf6c304d0d07e0758825c10
PREFIX ${EIGEN_SOURCE_DIR}
UPDATE_COMMAND ""
CONFIGURE_COMMAND ""
......
cmake/external/gflags.cmake
浏览文件 @ 1ffdecf1
......@@ -18,9 +18,9 @@ SET(GFLAGS_SOURCES_DIR ${THIRD_PARTY_PATH}/gflags)
SET(GFLAGS_INSTALL_DIR ${THIRD_PARTY_PATH}/install/gflags)
SET(GFLAGS_INCLUDE_DIR "${GFLAGS_INSTALL_DIR}/include" CACHE PATH "gflags include directory." FORCE)
IF(WIN32)
set(GFLAGS_LIBRARIES "${GFLAGS_INSTALL_DIR}/lib/gflags.lib" CACHE FILEPATH "GFLAGS_LIBRARIES" FORCE)
set(GFLAGS_LIBRARIES "${GFLAGS_INSTALL_DIR}/lib/gflags.lib" CACHE FILEPATH "GFLAGS_LIBRARIES" FORCE)
ELSE(WIN32)
set(GFLAGS_LIBRARIES "${GFLAGS_INSTALL_DIR}/lib/libgflags.a" CACHE FILEPATH "GFLAGS_LIBRARIES" FORCE)
set(GFLAGS_LIBRARIES "${GFLAGS_INSTALL_DIR}/lib/libgflags.a" CACHE FILEPATH "GFLAGS_LIBRARIES" FORCE)
ENDIF(WIN32)
INCLUDE_DIRECTORIES(${GFLAGS_INCLUDE_DIR})
......@@ -36,19 +36,21 @@ ExternalProject_Add(
# change this back to the official Github repo once my PR is
# merged.
GIT_REPOSITORY "https://github.com/wangkuiyi/gflags.git"
GIT_TAG 986964c07427ecb9cdb5bd73f73ebbd40e54dadb
PREFIX ${GFLAGS_SOURCES_DIR}
UPDATE_COMMAND ""
CMAKE_ARGS -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
CMAKE_ARGS -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
CMAKE_ARGS -DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
CMAKE_ARGS -DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${GFLAGS_INSTALL_DIR}
CMAKE_ARGS -DCMAKE_POSITION_INDEPENDENT_CODE=ON
CMAKE_ARGS -DBUILD_TESTING=OFF
CMAKE_ARGS -DCMAKE_BUILD_TYPE=Release
-DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
-DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
-DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
-DCMAKE_INSTALL_PREFIX=${GFLAGS_INSTALL_DIR}
-DCMAKE_POSITION_INDEPENDENT_CODE=ON
-DBUILD_TESTING=OFF
-DCMAKE_BUILD_TYPE=${THIRD_PARTY_BUILD_TYPE}
${EXTERNAL_OPTIONAL_ARGS}
CMAKE_CACHE_ARGS -DCMAKE_INSTALL_PREFIX:PATH=${GFLAGS_INSTALL_DIR}
-DCMAKE_POSITION_INDEPENDENT_CODE:BOOL=ON
-DCMAKE_BUILD_TYPE:STRING=Release
-DCMAKE_BUILD_TYPE:STRING=${THIRD_PARTY_BUILD_TYPE}
)
ADD_LIBRARY(gflags STATIC IMPORTED GLOBAL)
......@@ -56,3 +58,12 @@ SET_PROPERTY(TARGET gflags PROPERTY IMPORTED_LOCATION ${GFLAGS_LIBRARIES})
ADD_DEPENDENCIES(gflags extern_gflags)
LIST(APPEND external_project_dependencies gflags)
IF(WITH_C_API)
INSTALL(DIRECTORY ${GFLAGS_INCLUDE_DIR} DESTINATION third_party/gflags)
IF(ANDROID)
INSTALL(FILES ${GFLAGS_LIBRARIES} DESTINATION third_party/gflags/lib/${ANDROID_ABI})
ELSE()
INSTALL(FILES ${GFLAGS_LIBRARIES} DESTINATION third_party/gflags/lib)
ENDIF()
ENDIF()
cmake/external/glog.cmake
浏览文件 @ 1ffdecf1
......@@ -19,9 +19,9 @@ SET(GLOG_INSTALL_DIR ${THIRD_PARTY_PATH}/install/glog)
SET(GLOG_INCLUDE_DIR "${GLOG_INSTALL_DIR}/include" CACHE PATH "glog include directory." FORCE)
IF(WIN32)
SET(GLOG_LIBRARIES "${GLOG_INSTALL_DIR}/lib/libglog.lib" CACHE FILEPATH "glog library." FORCE)
SET(GLOG_LIBRARIES "${GLOG_INSTALL_DIR}/lib/libglog.lib" CACHE FILEPATH "glog library." FORCE)
ELSE(WIN32)
SET(GLOG_LIBRARIES "${GLOG_INSTALL_DIR}/lib/libglog.a" CACHE FILEPATH "glog library." FORCE)
SET(GLOG_LIBRARIES "${GLOG_INSTALL_DIR}/lib/libglog.a" CACHE FILEPATH "glog library." FORCE)
ENDIF(WIN32)
INCLUDE_DIRECTORIES(${GLOG_INCLUDE_DIR})
......@@ -31,23 +31,25 @@ ExternalProject_Add(
${EXTERNAL_PROJECT_LOG_ARGS}
DEPENDS gflags
GIT_REPOSITORY "https://github.com/google/glog.git"
GIT_TAG v0.3.5
PREFIX ${GLOG_SOURCES_DIR}
UPDATE_COMMAND ""
CMAKE_ARGS -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
CMAKE_ARGS -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
CMAKE_ARGS -DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
CMAKE_ARGS -DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${GLOG_INSTALL_DIR}
CMAKE_ARGS -DCMAKE_INSTALL_LIBDIR=${GLOG_INSTALL_DIR}/lib
CMAKE_ARGS -DCMAKE_POSITION_INDEPENDENT_CODE=ON
CMAKE_ARGS -DWITH_GFLAGS=ON
CMAKE_ARGS -Dgflags_DIR=${GFLAGS_INSTALL_DIR}/lib/cmake/gflags
CMAKE_ARGS -DBUILD_TESTING=OFF
CMAKE_ARGS -DCMAKE_BUILD_TYPE=Release
-DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
-DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
-DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
-DCMAKE_INSTALL_PREFIX=${GLOG_INSTALL_DIR}
-DCMAKE_INSTALL_LIBDIR=${GLOG_INSTALL_DIR}/lib
-DCMAKE_POSITION_INDEPENDENT_CODE=ON
-DWITH_GFLAGS=ON
-Dgflags_DIR=${GFLAGS_INSTALL_DIR}/lib/cmake/gflags
-DBUILD_TESTING=OFF
-DCMAKE_BUILD_TYPE=${THIRD_PARTY_BUILD_TYPE}
${EXTERNAL_OPTIONAL_ARGS}
CMAKE_CACHE_ARGS -DCMAKE_INSTALL_PREFIX:PATH=${GLOG_INSTALL_DIR}
-DCMAKE_INSTALL_LIBDIR:PATH=${GLOG_INSTALL_DIR}/lib
-DCMAKE_POSITION_INDEPENDENT_CODE:BOOL=ON
-DCMAKE_BUILD_TYPE:STRING=Release
-DCMAKE_BUILD_TYPE:STRING=${THIRD_PARTY_BUILD_TYPE}
)
ADD_LIBRARY(glog STATIC IMPORTED GLOBAL)
......@@ -56,3 +58,12 @@ ADD_DEPENDENCIES(glog extern_glog gflags)
LINK_LIBRARIES(glog gflags)
LIST(APPEND external_project_dependencies glog)
IF(WITH_C_API)
INSTALL(DIRECTORY ${GLOG_INCLUDE_DIR} DESTINATION third_party/glog)
IF(ANDROID)
INSTALL(FILES ${GLOG_LIBRARIES} DESTINATION third_party/glog/lib/${ANDROID_ABI})
ELSE()
INSTALL(FILES ${GLOG_LIBRARIES} DESTINATION third_party/glog/lib)
ENDIF()
ENDIF()
cmake/external/gtest.cmake
浏览文件 @ 1ffdecf1
......@@ -48,18 +48,19 @@ IF(WITH_TESTING)
PREFIX ${GTEST_SOURCES_DIR}
UPDATE_COMMAND ""
CMAKE_ARGS -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
CMAKE_ARGS -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
CMAKE_ARGS -DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
CMAKE_ARGS -DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${GTEST_INSTALL_DIR}
CMAKE_ARGS -DCMAKE_POSITION_INDEPENDENT_CODE=ON
CMAKE_ARGS -DBUILD_GMOCK=ON
CMAKE_ARGS -Dgtest_disable_pthreads=ON
CMAKE_ARGS -Dgtest_force_shared_crt=ON
CMAKE_ARGS -DCMAKE_BUILD_TYPE=Release
-DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
-DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
-DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
-DCMAKE_INSTALL_PREFIX=${GTEST_INSTALL_DIR}
-DCMAKE_POSITION_INDEPENDENT_CODE=ON
-DBUILD_GMOCK=ON
-Dgtest_disable_pthreads=ON
-Dgtest_force_shared_crt=ON
-DCMAKE_BUILD_TYPE=${THIRD_PARTY_BUILD_TYPE}
${EXTERNAL_OPTIONAL_ARGS}
CMAKE_CACHE_ARGS -DCMAKE_INSTALL_PREFIX:PATH=${GTEST_INSTALL_DIR}
-DCMAKE_POSITION_INDEPENDENT_CODE:BOOL=ON
-DCMAKE_BUILD_TYPE:STRING=Release
-DCMAKE_BUILD_TYPE:STRING=${THIRD_PARTY_BUILD_TYPE}
)
ADD_LIBRARY(gtest STATIC IMPORTED GLOBAL)
......
cmake/external/mkldnn.cmake
浏览文件 @ 1ffdecf1
......@@ -46,16 +46,20 @@ IF(${CBLAS_PROVIDER} STREQUAL "MKLML")
MESSAGE(STATUS "Build MKLDNN with ${MKLDNN_MKLROOT}")
ENDIF()
SET(MKLDNN_CFLAG "${CMAKE_C_FLAGS} -Wno-error=strict-overflow")
SET(MKLDNN_CXXFLAG "${CMAKE_CXX_FLAGS} -Wno-error=strict-overflow")
ExternalProject_Add(
${MKLDNN_PROJECT}
${EXTERNAL_PROJECT_LOG_ARGS}
DEPENDS ${MKLDNN_DEPENDS}
GIT_REPOSITORY "https://github.com/01org/mkl-dnn.git"
GIT_TAG "v0.9"
GIT_TAG "v0.11"
PREFIX ${MKLDNN_SOURCES_DIR}
UPDATE_COMMAND ""
CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${MKLDNN_INSTALL_DIR}
CMAKE_ARGS -DMKLROOT=${MKLDNN_MKLROOT}
CMAKE_ARGS -DCMAKE_C_FLAGS=${MKLDNN_CFLAG}
CMAKE_ARGS -DCMAKE_CXX_FLAGS=${MKLDNN_CXXFLAG}
CMAKE_CACHE_ARGS -DCMAKE_INSTALL_PREFIX:PATH=${MKLDNN_INSTALL_DIR}
-DMKLROOT:PATH=${MKLDNN_MKLROOT}
)
......
cmake/external/mklml.cmake
浏览文件 @ 1ffdecf1
......@@ -27,8 +27,8 @@ ENDIF()
INCLUDE(ExternalProject)
SET(MKLML_PROJECT "extern_mklml")
SET(MKLML_VER "mklml_lnx_2018.0.20170720")
SET(MKLML_URL "https://github.com/01org/mkl-dnn/releases/download/v0.9/${MKLML_VER}.tgz")
SET(MKLML_VER "mklml_lnx_2018.0.1.20171007")
SET(MKLML_URL "https://github.com/01org/mkl-dnn/releases/download/v0.11/${MKLML_VER}.tgz")
SET(MKLML_SOURCE_DIR "${THIRD_PARTY_PATH}/mklml")
SET(MKLML_DOWNLOAD_DIR "${MKLML_SOURCE_DIR}/src/${MKLML_PROJECT}")
SET(MKLML_DST_DIR "mklml")
......@@ -54,7 +54,8 @@ ExternalProject_Add(
${EXTERNAL_PROJECT_LOG_ARGS}
PREFIX ${MKLML_SOURCE_DIR}
DOWNLOAD_DIR ${MKLML_DOWNLOAD_DIR}
DOWNLOAD_COMMAND wget --no-check-certificate -qO- ${MKLML_URL} | tar xz -C ${MKLML_DOWNLOAD_DIR}
DOWNLOAD_COMMAND wget --no-check-certificate ${MKLML_URL} -c -q -O ${MKLML_VER}.tgz
&& tar zxf ${MKLML_VER}.tgz
DOWNLOAD_NO_PROGRESS 1
UPDATE_COMMAND ""
CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${MKLML_INSTALL_ROOT}
......
cmake/external/nccl.cmake 0 → 100644
浏览文件 @ 1ffdecf1
# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
if(NOT WITH_GPU)
return()
endif()
include(ExternalProject)
set(NCCL_SOURCE_DIR ${THIRD_PARTY_PATH}/nccl)
include_directories(${NCCL_SOURCE_DIR}/src/extern_nccl/src)
if(WITH_DSO)
# If we use DSO, we do not build nccl, just download the dependencies
set(NCCL_BUILD_COMMAND "")
set(NCCL_INSTALL_COMMAND "")
set(NCCL_INSTALL_DIR "")
else()
# otherwise, we build nccl and link it.
set(NCCL_INSTALL_DIR ${THIRD_PARTY_PATH}/install/nccl)
# Note: cuda 8.0 is needed to make nccl
# When cuda is not installed on the system directory, need to set CUDA_HOME to your cuda root
set(NCCL_BUILD_COMMAND "make -j 8")
set(NCCL_INSTALL_COMMAND "make install PREFIX=${NCCL_INSTALL_DIR}")
endif()
ExternalProject_Add(
extern_nccl
${EXTERNAL_PROJECT_LOG_ARGS}
GIT_REPOSITORY "https://github.com/NVIDIA/nccl.git"
GIT_TAG "v1.3.4-1"
PREFIX "${NCCL_SOURCE_DIR}"
UPDATE_COMMAND ""
CONFIGURE_COMMAND ""
BUILD_COMMAND "${NCCL_BUILD_COMMAND}"
INSTALL_COMMAND "${NCCL_INSTALL_COMMAND}"
INSTALL_DIR "${NCCL_INSTALL_DIR}"
TEST_COMMAND ""
)
if(WITH_DSO)
if(${CMAKE_VERSION} VERSION_LESS "3.3.0")
set(dummyfile ${CMAKE_CURRENT_BINARY_DIR}/lib_nccl_dummy.c)
file(WRITE ${dummyfile} "const char * dummy_nccl = \"${dummyfile}\";")
add_library(nccl STATIC ${dummyfile})
else()
add_library(nccl INTERFACE)
endif()
else()
add_library(nccl STATIC IMPORTED GLOBAL)
set_property(TARGET nccl PROPERTY IMPORTED_LOCATION
${NCCL_INSTALL_DIR}/lib/libnccl_static.a)
endif()
add_dependencies(nccl extern_nccl)
cmake/external/openblas.cmake
浏览文件 @ 1ffdecf1
# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.
#
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#
# http://www.apache.org/licenses/LICENSE-2.0
#
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
IF(USE_EIGEN_FOR_BLAS)
return()
ENDIF(USE_EIGEN_FOR_BLAS)
INCLUDE(cblas)
IF(NOT ${CBLAS_FOUND})
......@@ -25,32 +29,50 @@ IF(NOT ${CBLAS_FOUND})
"${CBLAS_INSTALL_DIR}/lib/${CMAKE_STATIC_LIBRARY_PREFIX}openblas${CMAKE_STATIC_LIBRARY_SUFFIX}"
CACHE FILEPATH "openblas library." FORCE)
SET(COMMON_ARGS CC=${CMAKE_C_COMPILER} NO_SHARED=1 NO_LAPACK=1 libs)
SET(OPENBLAS_CC "${CMAKE_C_COMPILER}")
IF(CMAKE_CROSSCOMPILING)
SET(OPTIONAL_ARGS HOSTCC=${HOST_C_COMPILER})
GET_FILENAME_COMPONENT(CROSS_SUFFIX ${CMAKE_C_COMPILER} DIRECTORY)
SET(CROSS_SUFFIX ${CROSS_SUFFIX}/)
IF(ANDROID)
# arm_soft_fp_abi branch of OpenBLAS to support softfp
# https://github.com/xianyi/OpenBLAS/tree/arm_soft_fp_abi
SET(OPENBLAS_COMMIT "b5c96fcfcdc82945502a2303116a64d89985daf5")
IF(ANDROID_ABI MATCHES "^armeabi(-v7a)?$")
SET(TARGET "ARMV7")
SET(OPTIONAL_ARGS ${OPTIONAL_ARGS} TARGET=ARMV7 ARM_SOFTFP_ABI=1 USE_THREAD=0)
ELSEIF(ANDROID_ABI STREQUAL "arm64-v8a")
SET(TARGET "ARMV8")
SET(OPTIONAL_ARGS ${OPTIONAL_ARGS} TARGET=ARMV8 BINARY=64 USE_THREAD=0)
ENDIF()
ELSEIF(IOS)
# FIXME(liuyiqun): support multiple architectures
SET(OPENBLAS_COMMIT "b5c96fcfcdc82945502a2303116a64d89985daf5")
SET(OPENBLAS_CC "${OPENBLAS_CC} ${CMAKE_C_FLAGS} -isysroot ${CMAKE_OSX_SYSROOT}")
IF(CMAKE_OSX_ARCHITECTURES MATCHES "armv7")
SET(OPENBLAS_CC "${OPENBLAS_CC} -arch armv7")
SET(OPTIONAL_ARGS ${OPTIONAL_ARGS} TARGET=ARMV7 ARM_SOFTFP_ABI=1 USE_THREAD=0)
ELSEIF(CMAKE_OSX_ARCHITECTURES MATCHES "arm64")
SET(OPENBLAS_CC "${OPENBLAS_CC} -arch arm64")
SET(OPTIONAL_ARGS ${OPTIONAL_ARGS} TARGET=ARMV8 BINARY=64 USE_THREAD=0 CROSS_SUFFIX=${CROSS_SUFFIX})
ENDIF()
SET(OPTIONAL_ARGS HOSTCC=${HOST_C_COMPILER} TARGET=${TARGET} ARM_SOFTFP_ABI=1 USE_THREAD=0)
ELSEIF(RPI)
# use hardfp
SET(OPENBLAS_COMMIT "v0.2.19")
SET(OPTIONAL_ARGS HOSTCC=${HOST_C_COMPILER} TARGET=ARMV7 USE_THREAD=0)
SET(OPENBLAS_COMMIT "v0.2.20")
SET(OPTIONAL_ARGS ${OPTIONAL_ARGS} TARGET=ARMV7 USE_THREAD=0)
ENDIF()
ELSE()
SET(OPENBLAS_COMMIT "v0.2.19")
IF(APPLE)
SET(OPENBLAS_CC "${CMAKE_C_COMPILER} -isysroot ${CMAKE_OSX_SYSROOT}")
ENDIF()
SET(OPENBLAS_COMMIT "v0.2.20")
SET(OPTIONAL_ARGS "")
IF(CMAKE_SYSTEM_PROCESSOR MATCHES "^x86(_64)?$")
SET(OPTIONAL_ARGS DYNAMIC_ARCH=1 NUM_THREADS=64)
ENDIF()
ENDIF()
SET(COMMON_ARGS CC=${OPENBLAS_CC} NO_SHARED=1 NO_LAPACK=1 libs)
ExternalProject_Add(
extern_openblas
${EXTERNAL_PROJECT_LOG_ARGS}
......@@ -64,6 +86,26 @@ IF(NOT ${CBLAS_FOUND})
UPDATE_COMMAND ""
CONFIGURE_COMMAND ""
)
SET(CBLAS_PROVIDER openblas)
IF(WITH_C_API)
INSTALL(DIRECTORY ${CBLAS_INC_DIR} DESTINATION third_party/openblas)
# Because libopenblas.a is a symbolic link of another library, thus need to
# install the whole directory.
IF(ANDROID)
SET(TMP_INSTALL_DIR third_party/openblas/lib/${ANDROID_ABI})
ELSE()
SET(TMP_INSTALL_DIR third_party/openblas/lib)
ENDIF()
INSTALL(CODE "execute_process(
COMMAND ${CMAKE_COMMAND} -E copy_directory ${CBLAS_INSTALL_DIR}/lib
destination ${CMAKE_INSTALL_PREFIX}/${TMP_INSTALL_DIR}
)"
)
INSTALL(CODE "MESSAGE(STATUS \"Installing: \"
\"${CBLAS_INSTALL_DIR}/lib -> ${CMAKE_INSTALL_PREFIX}/${TMP_INSTALL_DIR}\"
)"
)
ENDIF()
ENDIF(NOT ${CBLAS_FOUND})
MESSAGE(STATUS "BLAS library: ${CBLAS_LIBRARIES}")
......@@ -73,7 +115,7 @@ INCLUDE_DIRECTORIES(${CBLAS_INC_DIR})
# linear algebra libraries for cc_library(xxx SRCS xxx.c DEPS cblas)
SET(dummyfile ${CMAKE_CURRENT_BINARY_DIR}/cblas_dummy.c)
FILE(WRITE ${dummyfile} "const char * dummy = \"${dummyfile}\";")
IF(${CBLAS_PROVIDER} MATCHES MKL)
IF("${CBLAS_PROVIDER}" STREQUAL "MKLML")
ADD_LIBRARY(cblas SHARED ${dummyfile})
ELSE()
ADD_LIBRARY(cblas STATIC ${dummyfile})
......
cmake/external/protobuf.cmake
浏览文件 @ 1ffdecf1
......@@ -173,7 +173,8 @@ FUNCTION(build_protobuf TARGET_NAME BUILD_FOR_HOST)
"-DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}"
"-DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}"
"-Dprotobuf_WITH_ZLIB=ON"
"-DZLIB_ROOT:FILEPATH=${ZLIB_ROOT}")
"-DZLIB_ROOT:FILEPATH=${ZLIB_ROOT}"
${EXTERNAL_OPTIONAL_ARGS})
SET(OPTIONAL_CACHE_ARGS "-DZLIB_ROOT:STRING=${ZLIB_ROOT}")
ENDIF()
......@@ -190,12 +191,12 @@ FUNCTION(build_protobuf TARGET_NAME BUILD_FOR_HOST)
${OPTIONAL_ARGS}
-Dprotobuf_BUILD_TESTS=OFF
-DCMAKE_POSITION_INDEPENDENT_CODE=ON
-DCMAKE_BUILD_TYPE=Release
-DCMAKE_BUILD_TYPE=${THIRD_PARTY_BUILD_TYPE}
-DCMAKE_INSTALL_PREFIX=${PROTOBUF_INSTALL_DIR}
-DCMAKE_INSTALL_LIBDIR=lib
CMAKE_CACHE_ARGS
-DCMAKE_INSTALL_PREFIX:PATH=${PROTOBUF_INSTALL_DIR}
-DCMAKE_BUILD_TYPE:STRING=Release
-DCMAKE_BUILD_TYPE:STRING=${THIRD_PARTY_BUILD_TYPE}
-DCMAKE_VERBOSE_MAKEFILE:BOOL=OFF
-DCMAKE_POSITION_INDEPENDENT_CODE:BOOL=ON
${OPTIONAL_CACHE_ARGS}
......@@ -223,6 +224,15 @@ IF(NOT PROTOBUF_FOUND)
SET(PROTOBUF_PROTOC_LIBRARY ${extern_protobuf_PROTOC_LIBRARY}
CACHE FILEPATH "protoc library." FORCE)
IF(WITH_C_API)
INSTALL(DIRECTORY ${PROTOBUF_INCLUDE_DIR} DESTINATION third_party/protobuf)
IF(ANDROID)
INSTALL(FILES ${PROTOBUF_LIBRARY} DESTINATION third_party/protobuf/lib/${ANDROID_ABI})
ELSE()
INSTALL(FILES ${PROTOBUF_LIBRARY} DESTINATION third_party/protobuf/lib)
ENDIF()
ENDIF()
IF(CMAKE_CROSSCOMPILING)
PROMPT_PROTOBUF_LIB(protobuf_host extern_protobuf)
ELSE()
......
cmake/external/pybind11.cmake
浏览文件 @ 1ffdecf1
INCLUDE(ExternalProject)
# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
SET(PYBIND_SOURCE_DIR ${THIRD_PARTY_PATH}/pybind)
if(NOT WITH_PYTHON)
return()
endif()
include(ExternalProject)
INCLUDE_DIRECTORIES(${PYBIND_SOURCE_DIR}/src/extern_pybind/include)
set(PYBIND_SOURCE_DIR ${THIRD_PARTY_PATH}/pybind)
include_directories(${PYBIND_SOURCE_DIR}/src/extern_pybind/include)
ExternalProject_Add(
extern_pybind
......@@ -17,14 +35,12 @@ ExternalProject_Add(
TEST_COMMAND ""
)
if (${CMAKE_VERSION} VERSION_LESS "3.3.0")
if(${CMAKE_VERSION} VERSION_LESS "3.3.0")
set(dummyfile ${CMAKE_CURRENT_BINARY_DIR}/pybind_dummy.c)
file(WRITE ${dummyfile} "const char * dummy_any = \"${dummyfile}\";")
file(WRITE ${dummyfile} "const char * dummy_pybind = \"${dummyfile}\";")
add_library(pybind STATIC ${dummyfile})
else()
add_library(pybind INTERFACE)
endif()
add_dependencies(pybind extern_pybind)
LIST(APPEND external_project_dependencies pybind)
cmake/external/python.cmake
浏览文件 @ 1ffdecf1
......@@ -12,16 +12,17 @@
# See the License for the specific language governing permissions and
# limitations under the License.
INCLUDE(ExternalProject)
IF(NOT WITH_PYTHON)
return()
ENDIF()
INCLUDE(python_module)
FIND_PACKAGE(PythonInterp 2.7)
IF(WITH_PYTHON)
FIND_PACKAGE(PythonLibs 2.7)
# Fixme: Maybe find a static library. Get SHARED/STATIC by FIND_PACKAGE.
ADD_LIBRARY(python SHARED IMPORTED GLOBAL)
SET_PROPERTY(TARGET python PROPERTY IMPORTED_LOCATION ${PYTHON_LIBRARIES})
ENDIF(WITH_PYTHON)
FIND_PACKAGE(PythonLibs 2.7)
# Fixme: Maybe find a static library. Get SHARED/STATIC by FIND_PACKAGE.
ADD_LIBRARY(python SHARED IMPORTED GLOBAL)
SET_PROPERTY(TARGET python PROPERTY IMPORTED_LOCATION ${PYTHON_LIBRARIES})
SET(py_env "")
IF(PYTHONINTERP_FOUND)
......@@ -36,9 +37,5 @@ IF(PYTHONINTERP_FOUND)
ENDIF()
ENDIF(PYTHONINTERP_FOUND)
IF(WITH_PYTHON)
INCLUDE_DIRECTORIES(${PYTHON_INCLUDE_DIR})
INCLUDE_DIRECTORIES(${PYTHON_NUMPY_INCLUDE_DIR})
ELSE()
SET(PYTHON_LIBRARIES "")
ENDIF()
INCLUDE_DIRECTORIES(${PYTHON_INCLUDE_DIR})
INCLUDE_DIRECTORIES(${PYTHON_NUMPY_INCLUDE_DIR})
cmake/external/swig.cmake
浏览文件 @ 1ffdecf1
# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.
#
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#
# http://www.apache.org/licenses/LICENSE-2.0
#
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
IF(NOT WITH_SWIG_PY)
return()
ENDIF()
FIND_PACKAGE(SWIG)
IF(NOT SWIG_FOUND)
......
cmake/external/warpctc.cmake
浏览文件 @ 1ffdecf1
......@@ -16,25 +16,14 @@ INCLUDE(ExternalProject)
SET(WARPCTC_SOURCES_DIR ${THIRD_PARTY_PATH}/warpctc)
SET(WARPCTC_INSTALL_DIR ${THIRD_PARTY_PATH}/install/warpctc)
SET(WARPCTC_INCLUDE_DIR "${WARPCTC_INSTALL_DIR}/include" CACHE PATH "Warp-ctc Directory" FORCE)
INCLUDE_DIRECTORIES(${WARPCTC_INCLUDE_DIR})
SET(WARPCTC_LIB_DIR "${WARPCTC_INSTALL_DIR}/lib" CACHE PATH "Warp-ctc Library Directory" FORCE)
IF(WIN32)
SET(WARPCTC_LIBRARIES
"${WARPCTC_INSTALL_DIR}/lib/warpctc.dll" CACHE FILEPATH "Warp-ctc Library" FORCE)
ELSE(WIN32)
IF(APPLE)
SET(_warpctc_SHARED_SUFFIX dylib)
ELSE(APPLE)
SET(_warpctc_SHARED_SUFFIX so)
ENDIF(APPLE)
SET(WARPCTC_LIBRARIES
"${WARPCTC_INSTALL_DIR}/lib/libwarpctc.${_warpctc_SHARED_SUFFIX}" CACHE FILEPATH "Warp-ctc Library" FORCE)
ENDIF(WIN32)
SET(WARPCTC_INCLUDE_DIR "${WARPCTC_INSTALL_DIR}/include"
CACHE PATH "Warp-ctc Directory" FORCE)
# Used in unit test test_WarpCTCLayer
SET(WARPCTC_LIB_DIR "${WARPCTC_INSTALL_DIR}/lib"
CACHE PATH "Warp-ctc Library Directory" FORCE)
SET(WARPCTC_LIBRARIES "${WARPCTC_INSTALL_DIR}/lib/libwarpctc${CMAKE_SHARED_LIBRARY_SUFFIX}"
CACHE FILEPATH "Warp-ctc Library" FORCE)
IF(CMAKE_CXX_COMPILER_ID STREQUAL "Clang" OR CMAKE_CXX_COMPILER_ID STREQUAL "AppleClang" )
SET(USE_OMP OFF)
......@@ -46,25 +35,30 @@ ExternalProject_Add(
extern_warpctc
${EXTERNAL_PROJECT_LOG_ARGS}
GIT_REPOSITORY "https://github.com/gangliao/warp-ctc.git"
GIT_TAG b63a0644654a3e0ed624c85a1767bc8193aead09
PREFIX ${WARPCTC_SOURCES_DIR}
UPDATE_COMMAND ""
CMAKE_ARGS -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
CMAKE_ARGS -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
CMAKE_ARGS -DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
CMAKE_ARGS -DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${WARPCTC_INSTALL_DIR}
CMAKE_ARGS -DWITH_GPU=${WITH_GPU}
CMAKE_ARGS -DWITH_OMP=${USE_OMP}
CMAKE_ARGS -DWITH_TORCH=OFF
CMAKE_ARGS -DCMAKE_DISABLE_FIND_PACKAGE_Torch=ON
CMAKE_ARGS -DBUILD_SHARED=ON
CMAKE_ARGS -DCMAKE_POSITION_INDEPENDENT_CODE=ON
CMAKE_ARGS -DCMAKE_BUILD_TYPE=Release
CMAKE_CACHE_ARGS -DCMAKE_BUILD_TYPE:STRING=Release
-DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
-DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
-DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
-DCMAKE_INSTALL_PREFIX=${WARPCTC_INSTALL_DIR}
-DWITH_GPU=${WITH_GPU}
-DWITH_OMP=${USE_OMP}
-DWITH_TORCH=OFF
-DCMAKE_DISABLE_FIND_PACKAGE_Torch=ON
-DBUILD_SHARED=ON
-DCMAKE_POSITION_INDEPENDENT_CODE=ON
-DCMAKE_BUILD_TYPE=${THIRD_PARTY_BUILD_TYPE}
${EXTERNAL_OPTIONAL_ARGS}
CMAKE_CACHE_ARGS -DCMAKE_BUILD_TYPE:STRING=${THIRD_PARTY_BUILD_TYPE}
-DCMAKE_POSITION_INDEPENDENT_CODE:BOOL=ON
-DCMAKE_INSTALL_PREFIX:PATH=${WARPCTC_INSTALL_DIR}
)
MESSAGE(STATUS "warp-ctc library: ${WARPCTC_LIBRARIES}")
INCLUDE_DIRECTORIES(${WARPCTC_INCLUDE_DIR})
ADD_LIBRARY(warpctc STATIC IMPORTED GLOBAL)
SET_PROPERTY(TARGET warpctc PROPERTY IMPORTED_LOCATION ${WARPCTC_LIBRARIES})
ADD_DEPENDENCIES(warpctc extern_warpctc)
......
cmake/external/zlib.cmake
浏览文件 @ 1ffdecf1
# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.
#
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#
# http://www.apache.org/licenses/LICENSE-2.0
#
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
......@@ -34,18 +34,28 @@ ExternalProject_Add(
GIT_TAG "v1.2.8"
PREFIX ${ZLIB_SOURCES_DIR}
UPDATE_COMMAND ""
CMAKE_ARGS -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
CMAKE_ARGS -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
CMAKE_ARGS -DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
CMAKE_ARGS -DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${ZLIB_INSTALL_DIR}
CMAKE_ARGS -DBUILD_SHARED_LIBS=OFF
CMAKE_ARGS -DCMAKE_POSITION_INDEPENDENT_CODE=ON
CMAKE_ARGS -DCMAKE_MACOSX_RPATH=ON
CMAKE_ARGS -DCMAKE_BUILD_TYPE=Release
-DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
-DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}
-DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
-DCMAKE_INSTALL_PREFIX=${ZLIB_INSTALL_DIR}
-DBUILD_SHARED_LIBS=OFF
-DCMAKE_POSITION_INDEPENDENT_CODE=ON
-DCMAKE_MACOSX_RPATH=ON
-DCMAKE_BUILD_TYPE=${THIRD_PARTY_BUILD_TYPE}
${EXTERNAL_OPTIONAL_ARGS}
CMAKE_CACHE_ARGS -DCMAKE_INSTALL_PREFIX:PATH=${ZLIB_INSTALL_DIR}
-DCMAKE_POSITION_INDEPENDENT_CODE:BOOL=ON
-DCMAKE_BUILD_TYPE:STRING=Release
-DCMAKE_BUILD_TYPE:STRING=${THIRD_PARTY_BUILD_TYPE}
)
LIST(APPEND external_project_dependencies zlib)
IF(WITH_C_API)
INSTALL(DIRECTORY ${ZLIB_INCLUDE_DIR} DESTINATION third_party/zlib)
IF(ANDROID)
INSTALL(FILES ${ZLIB_LIBRARIES} DESTINATION third_party/zlib/lib/${ANDROID_ABI})
ELSE()
INSTALL(FILES ${ZLIB_LIBRARIES} DESTINATION third_party/zlib/lib)
ENDIF()
ENDIF()
cmake/flags.cmake
浏览文件 @ 1ffdecf1
......@@ -9,13 +9,6 @@ function(CheckCompilerCXX11Flag)
if(${CMAKE_CXX_COMPILER_VERSION} VERSION_LESS 4.8)
message(FATAL_ERROR "Unsupported GCC version. GCC >= 4.8 required.")
endif()
if(NOT ANDROID)
# TODO(qijun) gcc 4.9 or later versions raise SEGV due to the optimization problem.
# Use Debug mode instead for now.
if(CMAKE_CXX_COMPILER_VERSION VERSION_GREATER 4.9 OR CMAKE_CXX_COMPILER_VERSION VERSION_EQUAL 4.9)
set(CMAKE_BUILD_TYPE "Debug" CACHE STRING "" FORCE)
endif()
endif()
elseif(CMAKE_CXX_COMPILER_ID STREQUAL "AppleClang" OR CMAKE_CXX_COMPILER_ID STREQUAL "Clang")
# cmake >= 3.0 compiler id "AppleClang" on Mac OS X, otherwise "Clang"
# Apple Clang is a different compiler than upstream Clang which havs different version numbers.
......@@ -135,8 +128,10 @@ set(GPU_COMMON_FLAGS
)
if (APPLE)
# On Mac OS X build fat binaries with x86_64 architectures by default.
set (CMAKE_OSX_ARCHITECTURES "x86_64" CACHE STRING "Build architectures for OSX" FORCE)
if(NOT CMAKE_CROSSCOMPILING)
# On Mac OS X build fat binaries with x86_64 architectures by default.
set (CMAKE_OSX_ARCHITECTURES "x86_64" CACHE STRING "Build architectures for OSX" FORCE)
endif()
else()
set(GPU_COMMON_FLAGS
-Wall
......@@ -160,7 +155,7 @@ set(CUDA_PROPAGATE_HOST_FLAGS OFF)
# Release/Debug flags set by cmake. Such as -O3 -g -DNDEBUG etc.
# So, don't set these flags here.
LIST(APPEND CUDA_NVCC_FLAGS -std=c++11 --default-stream per-thread)
LIST(APPEND CUDA_NVCC_FLAGS -std=c++11)
LIST(APPEND CUDA_NVCC_FLAGS --use_fast_math)
if(CMAKE_BUILD_TYPE STREQUAL "Debug")
......
cmake/generic.cmake
浏览文件 @ 1ffdecf1
......@@ -106,22 +106,22 @@ function(merge_static_libs TARGET_NAME)
endforeach()
list(REMOVE_DUPLICATES libs_deps)
if(APPLE) # Use OSX's libtool to merge archives
# To produce a library we need at least one source file.
# It is created by add_custom_command below and will helps
# also help to track dependencies.
set(dummyfile ${CMAKE_CURRENT_BINARY_DIR}/${TARGET_NAME}_dummy.c)
# To produce a library we need at least one source file.
# It is created by add_custom_command below and will helps
# also help to track dependencies.
set(target_SRCS ${CMAKE_CURRENT_BINARY_DIR}/${TARGET_NAME}_dummy.c)
if(APPLE) # Use OSX's libtool to merge archives
# Make the generated dummy source file depended on all static input
# libs. If input lib changes,the source file is touched
# which causes the desired effect (relink).
add_custom_command(OUTPUT ${dummyfile}
COMMAND ${CMAKE_COMMAND} -E touch ${dummyfile}
add_custom_command(OUTPUT ${target_SRCS}
COMMAND ${CMAKE_COMMAND} -E touch ${target_SRCS}
DEPENDS ${libs})
# Generate dummy staic lib
file(WRITE ${dummyfile} "const char * dummy = \"${dummyfile}\";")
add_library(${TARGET_NAME} STATIC ${dummyfile})
file(WRITE ${target_SRCS} "const char *dummy = \"${target_SRCS}\";")
add_library(${TARGET_NAME} STATIC ${target_SRCS})
target_link_libraries(${TARGET_NAME} ${libs_deps})
foreach(lib ${libs})
......@@ -130,11 +130,14 @@ function(merge_static_libs TARGET_NAME)
endforeach()
add_custom_command(TARGET ${TARGET_NAME} POST_BUILD
COMMAND rm "${CMAKE_CURRENT_BINARY_DIR}/lib${TARGET_NAME}.a"
COMMAND /usr/bin/libtool -static -o "${CMAKE_CURRENT_BINARY_DIR}/lib${TARGET_NAME}.a" ${libfiles})
COMMAND /usr/bin/libtool -static -o "${CMAKE_CURRENT_BINARY_DIR}/lib${TARGET_NAME}.a" ${libfiles}
)
else() # general UNIX: use "ar" to extract objects and re-add to a common lib
set(target_DIR ${CMAKE_CURRENT_BINARY_DIR}/${TARGET_NAME}.dir)
foreach(lib ${libs})
set(objlistfile ${lib}.objlist) # list of objects in the input library
set(objdir ${lib}.objdir)
set(objlistfile ${target_DIR}/${lib}.objlist) # list of objects in the input library
set(objdir ${target_DIR}/${lib}.objdir)
add_custom_command(OUTPUT ${objdir}
COMMAND ${CMAKE_COMMAND} -E make_directory ${objdir}
......@@ -142,31 +145,32 @@ function(merge_static_libs TARGET_NAME)
add_custom_command(OUTPUT ${objlistfile}
COMMAND ${CMAKE_AR} -x "$<TARGET_FILE:${lib}>"
COMMAND ${CMAKE_AR} -t "$<TARGET_FILE:${lib}>" > ../${objlistfile}
COMMAND ${CMAKE_AR} -t "$<TARGET_FILE:${lib}>" > ${objlistfile}
DEPENDS ${lib} ${objdir}
WORKING_DIRECTORY ${objdir})
# Empty dummy source file that goes into merged library
set(mergebase ${lib}.mergebase.c)
add_custom_command(OUTPUT ${mergebase}
COMMAND ${CMAKE_COMMAND} -E touch ${mergebase}
DEPENDS ${objlistfile})
list(APPEND mergebases "${mergebase}")
list(APPEND target_OBJS "${objlistfile}")
endforeach()
add_library(${TARGET_NAME} STATIC ${mergebases})
# Make the generated dummy source file depended on all static input
# libs. If input lib changes,the source file is touched
# which causes the desired effect (relink).
add_custom_command(OUTPUT ${target_SRCS}
COMMAND ${CMAKE_COMMAND} -E touch ${target_SRCS}
DEPENDS ${libs} ${target_OBJS})
# Generate dummy staic lib
file(WRITE ${target_SRCS} "const char *dummy = \"${target_SRCS}\";")
add_library(${TARGET_NAME} STATIC ${target_SRCS})
target_link_libraries(${TARGET_NAME} ${libs_deps})
# Get the file name of the generated library
set(outlibfile "$<TARGET_FILE:${TARGET_NAME}>")
set(target_LIBNAME "$<TARGET_FILE:${TARGET_NAME}>")
foreach(lib ${libs})
add_custom_command(TARGET ${TARGET_NAME} POST_BUILD
COMMAND ${CMAKE_AR} cr ${outlibfile} *.o
COMMAND ${CMAKE_RANLIB} ${outlibfile}
WORKING_DIRECTORY ${lib}.objdir)
endforeach()
add_custom_command(TARGET ${TARGET_NAME} POST_BUILD
COMMAND ${CMAKE_AR} crs ${target_LIBNAME} `find ${target_DIR} -name '*.o'`
COMMAND ${CMAKE_RANLIB} ${target_LIBNAME}
WORKING_DIRECTORY ${target_DIR})
endif()
endfunction(merge_static_libs)
......@@ -196,7 +200,7 @@ function(cc_library TARGET_NAME)
add_style_check_target(${TARGET_NAME} ${cc_library_SRCS} ${cc_library_HEADERS})
else(cc_library_SRCS)
if (cc_library_DEPS)
if(cc_library_DEPS)
merge_static_libs(${TARGET_NAME} ${cc_library_DEPS})
else()
message(FATAL "Please specify source file or library in cc_library.")
......@@ -249,7 +253,7 @@ function(nv_library TARGET_NAME)
foreach(source_file ${nv_library_SRCS})
string(REGEX REPLACE "\\.[^.]*$" "" source ${source_file})
if(EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/${source}.h)
list(APPEND cc_library_HEADERS ${CMAKE_CURRENT_SOURCE_DIR}/${source}.h)
list(APPEND nv_library_HEADERS ${CMAKE_CURRENT_SOURCE_DIR}/${source}.h)
endif()
endforeach()
add_style_check_target(${TARGET_NAME} ${nv_library_SRCS} ${nv_library_HEADERS})
......@@ -385,13 +389,60 @@ function(go_test TARGET_NAME)
WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR})
endfunction(go_test)
# Modification of standard 'protobuf_generate_cpp()' with protobuf-lite support
# Usage:
# paddle_protobuf_generate_cpp(<proto_srcs> <proto_hdrs> <proto_files>)
function(paddle_protobuf_generate_cpp SRCS HDRS)
if(NOT ARGN)
message(SEND_ERROR "Error: paddle_protobuf_generate_cpp() called without any proto files")
return()
endif()
set(${SRCS})
set(${HDRS})
if (MOBILE_INFERENCE)
set(EXTRA_FLAG "lite:")
else()
set(EXTRA_FLAG "")
endif()
foreach(FIL ${ARGN})
get_filename_component(ABS_FIL ${FIL} ABSOLUTE)
get_filename_component(FIL_WE ${FIL} NAME_WE)
set(_protobuf_protoc_src "${CMAKE_CURRENT_BINARY_DIR}/${FIL_WE}.pb.cc")
set(_protobuf_protoc_hdr "${CMAKE_CURRENT_BINARY_DIR}/${FIL_WE}.pb.h")
list(APPEND ${SRCS} "${_protobuf_protoc_src}")
list(APPEND ${HDRS} "${_protobuf_protoc_hdr}")
add_custom_command(
OUTPUT "${_protobuf_protoc_src}"
"${_protobuf_protoc_hdr}"
COMMAND ${CMAKE_COMMAND} -E make_directory "${CMAKE_CURRENT_BINARY_DIR}"
COMMAND ${PROTOBUF_PROTOC_EXECUTABLE}
-I${CMAKE_CURRENT_SOURCE_DIR}
--cpp_out "${EXTRA_FLAG}${CMAKE_CURRENT_BINARY_DIR}" ${ABS_FIL}
DEPENDS ${ABS_FIL} protoc
COMMENT "Running C++ protocol buffer compiler on ${FIL}"
VERBATIM )
endforeach()
set_source_files_properties(${${SRCS}} ${${HDRS}} PROPERTIES GENERATED TRUE)
set(${SRCS} ${${SRCS}} PARENT_SCOPE)
set(${HDRS} ${${HDRS}} PARENT_SCOPE)
endfunction()
function(proto_library TARGET_NAME)
set(oneValueArgs "")
set(multiValueArgs SRCS DEPS)
cmake_parse_arguments(proto_library "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
set(proto_srcs)
set(proto_hdrs)
protobuf_generate_cpp(proto_srcs proto_hdrs ${proto_library_SRCS})
paddle_protobuf_generate_cpp(proto_srcs proto_hdrs ${proto_library_SRCS})
cc_library(${TARGET_NAME} SRCS ${proto_srcs} DEPS ${proto_library_DEPS} protobuf)
endfunction()
......
cmake/simd.cmake
浏览文件 @ 1ffdecf1
# This file is use to check all support level of AVX on your machine
# so that PaddlePaddle can unleash the vectorization power of muticore.
INCLUDE(CheckCXXSourceRuns)
INCLUDE(CheckCXXSourceCompiles)
include(CheckCXXSourceRuns)
include(CheckCXXSourceCompiles)
IF(CMAKE_COMPILER_IS_GNUCC OR CMAKE_COMPILER_IS_GNUCXX OR CMAKE_CXX_COMPILER_ID MATCHES "Clang")
if(CMAKE_COMPILER_IS_GNUCC OR CMAKE_COMPILER_IS_GNUCXX OR CMAKE_CXX_COMPILER_ID MATCHES "Clang")
set(MMX_FLAG "-mmmx")
set(SSE2_FLAG "-msse2")
set(SSE3_FLAG "-msse3")
SET(AVX_FLAG "-mavx")
SET(AVX2_FLAG "-mavx2")
ELSEIF(MSVC)
set(AVX_FLAG "-mavx")
set(AVX2_FLAG "-mavx2")
elseif(MSVC)
set(MMX_FLAG "/arch:MMX")
set(SSE2_FLAG "/arch:SSE2")
set(SSE3_FLAG "/arch:SSE3")
SET(AVX_FLAG "/arch:AVX")
SET(AVX2_FLAG "/arch:AVX2")
ENDIF()
endif()
set(CMAKE_REQUIRED_FLAGS_RETAINED ${CMAKE_REQUIRED_FLAGS})
# Check MMX
set(CMAKE_REQUIRED_FLAGS ${MMX_FLAG})
set(MMX_FOUND_EXITCODE 1 CACHE STRING "Result from TRY_RUN" FORCE)
CHECK_CXX_SOURCE_RUNS("
#include <mmintrin.h>
int main()
......@@ -32,6 +33,7 @@ int main()
# Check SSE2
set(CMAKE_REQUIRED_FLAGS ${SSE2_FLAG})
set(SSE2_FOUND_EXITCODE 1 CACHE STRING "Result from TRY_RUN" FORCE)
CHECK_CXX_SOURCE_RUNS("
#include <emmintrin.h>
int main()
......@@ -42,6 +44,7 @@ int main()
# Check SSE3
set(CMAKE_REQUIRED_FLAGS ${SSE3_FLAG})
set(SSE3_FOUND_EXITCODE 1 CACHE STRING "Result from TRY_RUN" FORCE)
CHECK_CXX_SOURCE_RUNS("
#include <pmmintrin.h>
int main()
......@@ -55,6 +58,7 @@ int main()
# Check AVX
set(CMAKE_REQUIRED_FLAGS ${AVX_FLAG})
set(AVX_FOUND_EXITCODE 1 CACHE STRING "Result from TRY_RUN" FORCE)
CHECK_CXX_SOURCE_RUNS("
#include <immintrin.h>
int main()
......@@ -67,6 +71,7 @@ int main()
# Check AVX 2
set(CMAKE_REQUIRED_FLAGS ${AVX2_FLAG})
set(AVX2_FOUND_EXITCODE 1 CACHE STRING "Result from TRY_RUN" FORCE)
CHECK_CXX_SOURCE_RUNS("
#include <immintrin.h>
int main()
......
cmake/system.cmake
浏览文件 @ 1ffdecf1
......@@ -24,11 +24,10 @@ IF(WIN32)
SET(HOST_SYSTEM "win32")
ELSE(WIN32)
IF(APPLE)
EXEC_PROGRAM (sw_vers ARGS -productVersion OUTPUT_VARIABLE MACOSX_VERSION)
STRING(REGEX MATCH "[0-9]+.[0-9]+" VERSION "${MACOSX_VERSION}")
SET(MACOS_VERSION ${VERSION})
SET(HOST_SYSTEM "macosx")
IF(NOT DEFINED ENV{MACOSX_DEPLOYMENT_TARGET})
EXEC_PROGRAM(sw_vers ARGS -productVersion OUTPUT_VARIABLE HOST_SYSTEM_VERSION)
STRING(REGEX MATCH "[0-9]+.[0-9]+" MACOS_VERSION "${HOST_SYSTEM_VERSION}")
IF(NOT DEFINED $ENV{MACOSX_DEPLOYMENT_TARGET})
# Set cache variable - end user may change this during ccmake or cmake-gui configure.
SET(CMAKE_OSX_DEPLOYMENT_TARGET ${MACOS_VERSION} CACHE STRING
"Minimum OS X version to target for deployment (at runtime); newer APIs weak linked. Set to empty string for default value.")
......@@ -49,6 +48,8 @@ ELSE(WIN32)
ELSEIF(LINUX_ISSUE MATCHES "Fedora")
SET(HOST_SYSTEM "fedora")
ENDIF()
STRING(REGEX MATCH "(([0-9]+)\\.)+([0-9]+)" HOST_SYSTEM_VERSION "${LINUX_ISSUE}")
ENDIF(EXISTS "/etc/issue")
IF(EXISTS "/etc/redhat-release")
......@@ -70,7 +71,7 @@ CMAKE_HOST_SYSTEM_INFORMATION(RESULT CPU_CORES QUERY NUMBER_OF_LOGICAL_CORES)
MARK_AS_ADVANCED(HOST_SYSTEM CPU_CORES)
MESSAGE(STATUS "Found Paddle host system: ${HOST_SYSTEM}")
MESSAGE(STATUS "Found Paddle host system: ${HOST_SYSTEM}, version: ${HOST_SYSTEM_VERSION}")
MESSAGE(STATUS "Found Paddle host system's CPU: ${CPU_CORES} cores")
# configuration for cross-compiling
......@@ -82,6 +83,9 @@ IF(DEFINED CMAKE_SYSTEM_NAME)
ELSEIF(${CMAKE_SYSTEM_NAME} STREQUAL "RPi")
SET(RPI TRUE)
INCLUDE(cross_compiling/raspberry_pi)
ELSEIF(${CMAKE_SYSTEM_NAME} STREQUAL "iOS")
SET(IOS TRUE)
INCLUDE(cross_compiling/ios)
ENDIF()
ENDIF()
......
cmake/util.cmake
浏览文件 @ 1ffdecf1
......@@ -25,7 +25,9 @@ function(target_circle_link_libraries TARGET_NAME)
endif()
endforeach()
if("${CMAKE_CXX_COMPILER_ID}" STREQUAL "Clang" OR "${CMAKE_CXX_COMPILER_ID}" STREQUAL "AppleClang")
list(APPEND LIBS "-undefined dynamic_lookup")
if(NOT IOS_ENABLE_BITCODE)
list(APPEND LIBS "-undefined dynamic_lookup")
endif()
endif()
list(REVERSE libsInArgn)
target_link_libraries(${TARGET_NAME}
......@@ -71,30 +73,52 @@ function(link_paddle_exe TARGET_NAME)
generate_rdma_links()
endif()
target_circle_link_libraries(${TARGET_NAME}
ARCHIVE_START
paddle_gserver
paddle_function
ARCHIVE_END
paddle_pserver
paddle_trainer_lib
paddle_network
paddle_math
paddle_utils
paddle_parameter
paddle_proto
paddle_cuda
paddle_optimizer
${EXTERNAL_LIBS}
${CMAKE_THREAD_LIBS_INIT}
${CMAKE_DL_LIBS}
${RDMA_LD_FLAGS}
${RDMA_LIBS})
if(MOBILE_INFERENCE)
target_circle_link_libraries(${TARGET_NAME}
ARCHIVE_START
paddle_gserver
paddle_function
ARCHIVE_END
paddle_math
paddle_utils
paddle_parameter
paddle_proto
paddle_cuda
${EXTERNAL_LIBS}
${CMAKE_THREAD_LIBS_INIT}
${CMAKE_DL_LIBS}
${RDMA_LD_FLAGS}
${RDMA_LIBS})
else()
target_circle_link_libraries(${TARGET_NAME}
ARCHIVE_START
paddle_gserver
paddle_function
ARCHIVE_END
paddle_pserver
paddle_trainer_lib
paddle_network
paddle_math
paddle_utils
paddle_parameter
paddle_proto
paddle_cuda
paddle_optimizer
${EXTERNAL_LIBS}
${CMAKE_THREAD_LIBS_INIT}
${CMAKE_DL_LIBS}
${RDMA_LD_FLAGS}
${RDMA_LIBS})
endif()
if(ANDROID)
target_link_libraries(${TARGET_NAME} log)
endif(ANDROID)
if(WITH_MKLDNN AND WITH_MKLML AND MKLDNN_IOMP_DIR)
target_link_libraries(${TARGET_NAME} "-L${MKLDNN_IOMP_DIR} -liomp5 -Wl,--as-needed")
endif()
add_dependencies(${TARGET_NAME} ${external_project_dependencies})
endfunction()
......
doc/about/index_cn.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
关于PaddlePaddle
================
PaddlePaddle是一个最早由百度科学家和工程师共同研发的并行分布式深度学习平台，兼备易用性、高效性、灵活性和可扩展性，目前已被百度内部多个产品线广泛使用。
PaddlePaddle目前已经开放源码, 但是远未完善，我们希望能在这个基础上不断的改进、扩展和延伸。
同时我们希望广大开发者积极提供反馈和贡献源代码，建立一个活跃的开源社区。
致谢
--------
在此，特别感谢PaddlePaddle的[所有贡献者](https://github.com/PaddlePaddle/Paddle/graphs/contributors)。
doc/about/index_en.rst 已删除 100644 → 0
浏览文件 @ 3cf01b5d
ABOUT
=======
PaddlPaddle is an easy-to-use, efficient, flexible and scalable deep learning platform,
which is originally developed by Baidu scientists and engineers for the purpose of applying deep learning to many products at Baidu.
PaddlePaddle is now open source but far from complete, which is intended to be built upon, improved, scaled, and extended.
We hope to build an active open source community both by providing feedback and by actively contributing to the source code.
Credits
--------
We owe many thanks to `all contributors and developers <https://github.com/PaddlePaddle/Paddle/graphs/contributors>`_ of PaddlePaddle!
doc/api/v1/index_cn.rst
浏览文件 @ 1ffdecf1
......@@ -21,7 +21,7 @@ Model Config API
trainer_config_helpers/optimizers.rst
trainer_config_helpers/data_sources.rst
trainer_config_helpers/layers.rst
trainer_config_helpers/activations.rst
trainer_config_helpers/activations.rst
trainer_config_helpers/poolings.rst
trainer_config_helpers/networks.rst
trainer_config_helpers/evaluators.rst
......
doc/api/v2/config/layer.rst
浏览文件 @ 1ffdecf1
......@@ -262,6 +262,11 @@ seq_concat
.. autoclass:: paddle.v2.layer.seq_concat
:noindex:
seq_slice
---------
.. autoclass:: paddle.v2.layer.seq_slice
:noindex:
kmax_sequence_score
-------------------
.. autoclass:: paddle.v2.layer.kmax_sequence_score
......@@ -345,6 +350,11 @@ clip
.. autoclass:: paddle.v2.layer.clip
:noindex:
resize
------
.. autoclass:: paddle.v2.layer.resize
:noindex:
slope_intercept
---------------
.. autoclass:: paddle.v2.layer.slope_intercept
......@@ -367,6 +377,11 @@ trans
.. autoclass:: paddle.v2.layer.trans
:noindex:
scale_shift
-----------
.. autoclass:: paddle.v2.layer.scale_shift
:noindex:
Sampling Layers
===============
......@@ -414,9 +429,14 @@ multi_binary_label_cross_entropy_cost
.. autoclass:: paddle.v2.layer.multi_binary_label_cross_entropy_cost
:noindex:
huber_cost
----------
.. autoclass:: paddle.v2.layer.huber_cost
huber_regression_cost
-------------------------
.. autoclass:: paddle.v2.layer.huber_regression_cost
:noindex:
huber_classification_cost
-------------------------
.. autoclass:: paddle.v2.layer.huber_classification_cost
:noindex:
lambda_cost
......@@ -424,9 +444,9 @@ lambda_cost
.. autoclass:: paddle.v2.layer.lambda_cost
:noindex:
mse_cost
square_error_cost
--------
.. autoclass:: paddle.v2.layer.mse_cost
.. autoclass:: paddle.v2.layer.square_error_cost
:noindex:
rank_cost
......
doc/api/v2/config/networks.rst
浏览文件 @ 1ffdecf1
......@@ -125,3 +125,8 @@ simple_attention
:members: simple_attention
:noindex:
dot_product_attention
---------------------
.. automodule:: paddle.v2.networks
:members: dot_product_attention
:noindex:
doc/api/v2/data.rst
浏览文件 @ 1ffdecf1
......@@ -2,112 +2,9 @@
Data Reader Interface and DataSets
==================================
.. toctree::
:maxdepth: 1
DataTypes
=========
.. automodule:: paddle.v2.data_type
:members:
:noindex:
DataFeeder
==========
.. automodule:: paddle.v2.data_feeder
:members:
:noindex:
Reader
======
.. automodule:: paddle.v2.reader
:members:
:noindex:
.. automodule:: paddle.v2.reader.creator
:members:
:noindex:
minibatch
=========
.. automodule:: paddle.v2.minibatch
:members:
:noindex:
Dataset
=======
.. automodule:: paddle.v2.dataset
:members:
:noindex:
mnist
+++++
.. automodule:: paddle.v2.dataset.mnist
:members:
:noindex:
cifar
+++++
.. automodule:: paddle.v2.dataset.cifar
:members:
:noindex:
conll05
+++++++
.. automodule:: paddle.v2.dataset.conll05
:members: get_dict,get_embedding,test
:noindex:
imdb
++++
.. automodule:: paddle.v2.dataset.imdb
:members:
:noindex:
imikolov
++++++++
.. automodule:: paddle.v2.dataset.imikolov
:members:
:noindex:
movielens
+++++++++
.. automodule:: paddle.v2.dataset.movielens
:members:
:noindex:
.. autoclass:: paddle.v2.dataset.movielens.MovieInfo
:noindex:
.. autoclass:: paddle.v2.dataset.movielens.UserInfo
:noindex:
sentiment
+++++++++
.. automodule:: paddle.v2.dataset.sentiment
:members:
:noindex:
uci_housing
+++++++++++
.. automodule:: paddle.v2.dataset.uci_housing
:members:
:noindex:
wmt14
+++++
.. automodule:: paddle.v2.dataset.wmt14
:members:
:noindex:
data/data_reader.rst
data/image.rst
data/dataset.rst
doc/api/v2/data/data_reader.rst 0 → 100644
浏览文件 @ 1ffdecf1
=====================
Data Reader Interface
=====================
DataTypes
=========
.. automodule:: paddle.v2.data_type
:members:
:noindex:
DataFeeder
==========
.. automodule:: paddle.v2.data_feeder
:members:
:noindex:
Reader
======
.. automodule:: paddle.v2.reader
:members:
:noindex:
.. automodule:: paddle.v2.reader.creator
:members:
:noindex:
minibatch
=========
.. automodule:: paddle.v2.minibatch
:members:
:noindex:
doc/api/v2/data/dataset.rst 0 → 100644
浏览文件 @ 1ffdecf1
Dataset
=======
.. automodule:: paddle.v2.dataset
:members:
:noindex:
mnist
+++++
.. automodule:: paddle.v2.dataset.mnist
:members:
:noindex:
cifar
+++++
.. automodule:: paddle.v2.dataset.cifar
:members:
:noindex:
conll05
+++++++
.. automodule:: paddle.v2.dataset.conll05
:members: get_dict,get_embedding,test
:noindex:
imdb
++++
.. automodule:: paddle.v2.dataset.imdb
:members:
:noindex:
imikolov
++++++++
.. automodule:: paddle.v2.dataset.imikolov
:members:
:noindex:
movielens
+++++++++
.. automodule:: paddle.v2.dataset.movielens
:members:
:noindex:
.. autoclass:: paddle.v2.dataset.movielens.MovieInfo
:noindex:
.. autoclass:: paddle.v2.dataset.movielens.UserInfo
:noindex:
sentiment
+++++++++
.. automodule:: paddle.v2.dataset.sentiment
:members:
:noindex:
uci_housing
+++++++++++
.. automodule:: paddle.v2.dataset.uci_housing
:members:
:noindex:
wmt14
+++++
.. automodule:: paddle.v2.dataset.wmt14
:members:
:noindex:
doc/api/v2/data/image.rst 0 → 100644
浏览文件 @ 1ffdecf1
Image Interface
===============
.. automodule:: paddle.v2.image
:members:
doc/design/api.md
浏览文件 @ 1ffdecf1
......@@ -3,7 +3,7 @@
## Ingredients
As our design principle is starting from the essence: how could we
allow users to express and solve their problems at neural networks.
allow users to express and solve their problems as neural networks.
Some essential concepts that our API have to provide include:
1. A *topology* is an expression of *layers*.
......@@ -233,7 +233,7 @@ paddle.dist_train(model,
num_parameter_servers=15)
```
The pseudo code if `paddle.dist_train` is as follows:
The pseudo code of `paddle.dist_train` is as follows:
```python
def dist_train(topology, parameters, trainer, reader, ...):
......
doc/design/auto_gradient_check.md
浏览文件 @ 1ffdecf1
## Auto Gradient Checker Design
## Backgraound：
- Operator forward computing is easy to check if the result is right because it has a clear definition. **But** backpropagation is a notoriously difficult algorithm to debug and get right:
- 1. you should get the right backpropagation formula according to the forward computation.
- 2. you should implement it right in CPP.
- 3. it's difficult to prepare test data.
- Generally, it is easy to check whether the forward computation of an Operator is correct or not. However, backpropagation is a notoriously difficult algorithm to debug and get right:
1. you should get the right backpropagation formula according to the forward computation.
2. you should implement it right in CPP.
3. it's difficult to prepare test data.
- Auto gradient check gets a numeric gradient by forward Operator and use it as a reference of the backward Operator's result. It has several advantages:
- 1. numeric gradient checker only need forward operator.
- 2. user only need to prepare the input data for forward Operator.
- Auto gradient checking gets a numerical gradient by forward Operator and use it as a reference of the backward Operator's result. It has several advantages:
1. numerical gradient checker only need forward operator.
2. user only need to prepare the input data for forward Operator.
## Mathematical Theory
The following two document from stanford has a detailed explanation of how to get numeric gradient and why it's useful.
The following two document from Stanford has a detailed explanation of how to get numerical gradient and why it's useful.
- [Gradient checking and advanced optimization(en)](http://deeplearning.stanford.edu/wiki/index.php/Gradient_checking_and_advanced_optimization)
- [Gradient checking and advanced optimization(cn)](http://ufldl.stanford.edu/wiki/index.php/%E6%A2%AF%E5%BA%A6%E6%A3%80%E9%AA%8C%E4%B8%8E%E9%AB%98%E7%BA%A7%E4%BC%98%E5%8C%96)
......@@ -20,7 +20,7 @@ The following two document from stanford has a detailed explanation of how to ge
## Numeric Gradient Implementation
### Python Interface
```python
def get_numeric_gradient(op,
def get_numerical_gradient(op,
input_values,
output_name,
input_to_check,
......@@ -30,13 +30,13 @@ def get_numeric_gradient(op,
Get Numeric Gradient for an operator's input.
:param op: C++ operator instance, could be an network
:param input_values: The input variables. Should be an dictionary, key is
variable name. Value is numpy array.
:param input_values: The input variables. Should be an dictionary, whose key is
variable name, and value is numpy array.
:param output_name: The final output variable name.
:param input_to_check: The input variable need to get gradient.
:param input_to_check: The input variable with respect to which to compute the gradient.
:param delta: The perturbation value for numeric gradient method. The
smaller delta is, the more accurate result will get. But if that delta is
too small, it could occur numerical stability problem.
too small, it will suffer from numerical stability problem.
:param local_scope: The local scope used for get_numeric_gradient.
:return: The gradient array in numpy format.
"""
......@@ -45,28 +45,28 @@ def get_numeric_gradient(op,
### Explaination:
- Why need `output_name`
- One Operator may have multiple Output, you can get independent gradient from each Output. So user should set one output to calculate.
- An Operator may have multiple Output, one can get independent gradient from each Output. So caller should specify the name of the output variable.
- Why need `input_to_check`
- One operator may have multiple inputs. Gradient Op can calculate the gradient of these Inputs at the same time. But Numeric Gradient needs to calculate them one by one. So `get_numeric_gradient` is designed to calculate the gradient for one input. If you need to compute multiple inputs, you can call `get_numeric_gradient` multiple times.
- One operator may have multiple inputs. Gradient Op can calculate the gradient of these inputs at the same time. But Numeric Gradient needs to calculate them one by one. So `get_numeric_gradient` is designed to calculate the gradient for one input. If you need to compute multiple inputs, you can call `get_numeric_gradient` multiple times.
### Core Algorithm Implementation
```python
# we only compute gradient of one element each time.
# we use a for loop to compute the gradient of every element.
# we only compute gradient of one element a time.
# we use a for loop to compute the gradient of each element.
for i in xrange(tensor_size):
# get one input element throw it's index i.
# get one input element by its index i.
origin = tensor_to_check.get_float_element(i)
# add delta to it, run op and then get the sum of the result tensor.
# add delta to it, run op and then get the new value of the result tensor.
x_pos = origin + delta
tensor_to_check.set_float_element(i, x_pos)
y_pos = get_output()
# plus delta to this element, run op and get the sum of the result tensor.
# plus delta to this element, run op and get the new value of the result tensor.
x_neg = origin - delta
tensor_to_check.set_float_element(i, x_neg)
y_neg = get_output()
......@@ -85,15 +85,15 @@ def get_numeric_gradient(op,
Each Operator Kernel has three kinds of Gradient:
- 1. Numeric Gradient
- 2. CPU Operator Gradient
- 3. GPU Operator Gradient(if supported)
1. Numerical gradient
2. CPU kernel gradient
3. GPU kernel gradient (if supported)
Numeric Gradient Only relies on forward Operator. So we use Numeric Gradient as the reference value.
The numerical gradient only relies on forward Operator. So we use the numerical gradient as the reference value. And the gradient checking is performed in the following three steps:
- 1. calculate the numeric gradient.
- 2. calculate CPU kernel Gradient with the backward Operator and compare it with the numeric gradient.
- 3. calculate GPU kernel Gradient with the backward Operator and compare it with the numeric gradient.(if support GPU)
1. calculate the numerical gradient
2. calculate CPU kernel gradient with the backward Operator and compare it with the numerical gradient
3. calculate GPU kernel gradient with the backward Operator and compare it with the numeric gradient (if supported)
#### Python Interface
......@@ -110,8 +110,8 @@ Numeric Gradient Only relies on forward Operator. So we use Numeric Gradient as
:param forward_op: used to create backward_op
:param input_vars: numpy value of input variable. The following
computation will use these variables.
:param inputs_to_check: inputs var names that should check gradient.
:param output_name: output name that used to
:param inputs_to_check: the input variable with respect to which to compute the gradient.
:param output_name: The final output variable name.
:param max_relative_error: The relative tolerance parameter.
:param no_grad_set: used when create backward ops
:param only_cpu: only compute and check gradient on cpu kernel.
......@@ -120,24 +120,24 @@ Numeric Gradient Only relies on forward Operator. So we use Numeric Gradient as
```
### How to check if two numpy array is close enough?
if `abs_numeric_grad` is nearly zero, then use abs error for numeric_grad, not relative
if `abs_numerical_grad` is nearly zero, then use abs error for numerical_grad
```python
numeric_grad = ...
numerical_grad = ...
operator_grad = numpy.array(scope.find_var(grad_var_name(name)).get_tensor())
abs_numeric_grad = numpy.abs(numeric_grad)
# if abs_numeric_grad is nearly zero, then use abs error for numeric_grad, not relative
abs_numerical_grad = numpy.abs(numerical_grad)
# if abs_numerical_grad is nearly zero, then use abs error for numeric_grad, not relative
# error.
abs_numeric_grad[abs_numeric_grad < 1e-3] = 1
abs_numerical_grad[abs_numerical_grad < 1e-3] = 1
diff_mat = numpy.abs(abs_numeric_grad - operator_grad) / abs_numeric_grad
diff_mat = numpy.abs(abs_numerical_grad - operator_grad) / abs_numerical_grad
max_diff = numpy.max(diff_mat)
```
#### Notes：
1，The Input data for auto gradient checker should be reasonable to avoid numeric problem.
The Input data for auto gradient checker should be reasonable to avoid numerical stability problem.
#### Refs:
......
doc/design/block.md 0 → 100644
浏览文件 @ 1ffdecf1
# Design Doc: Block and Scope
## The Representation of Computation
Both deep learning systems and programming languages help users describe computation procedures. These systems use various representations of computation:
- Caffe, Torch, and Paddle: sequences of layers.
- TensorFlow, Caffe2, Mxnet: graph of operators.
- PaddlePaddle: nested blocks, like C++ and Java programs.
## Block in Programming Languages and Deep Learning
In programming languages, a block is a pair of curly braces that includes local variables definitions and a sequence of instructions or operators.
Blocks work with control flow structures like `if`, `else`, and `for`, which have equivalents in deep learning:
| programming languages | PaddlePaddle |
|-----------------------|-----------------------|
| for, while loop | RNN, WhileOp |
| if, if-else, switch | IfElseOp, SwitchOp |
| sequential execution | a sequence of layers |
A key difference is that a C++ program describes a one pass computation, whereas a deep learning program describes both the forward and backward passes.
## Stack Frames and the Scope Hierarchy
The existence of the backward pass makes the execution of a block of PaddlePaddle different from traditional programs:
| programming languages | PaddlePaddle |
|-----------------------|---------------------------------|
| stack | scope hierarchy |
| stack frame | scope |
| push at entering block| push at entering block |
| pop at leaving block | destroy when minibatch completes|
1. In traditional programs:
- When the execution enters the left curly brace of a block, the runtime pushes a frame into the stack, where it realizes local variables.
- After the execution leaves the right curly brace, the runtime pops the frame.
- The maximum number of frames in the stack is the maximum depth of nested blocks.
1. In PaddlePaddle
- When the execution enters a block, PaddlePaddle adds a new scope, where it realizes variables.
- PaddlePaddle doesn't pop a scope after the execution of the block because variables therein are used by the backward pass. So it has a stack forest known as a *scope hierarchy*.
- The height of the highest tree is the maximum depth of nested blocks.
- After the processing of a minibatch, PaddlePaddle destroys the scope hierarchy.
## Use Blocks in C++ and PaddlePaddle Programs
Let us consolidate the discussion by presenting some examples.
### Blocks with `if-else` and `IfElseOp`
The following C++ programs shows how blocks are used with the `if-else` structure:
```c++
namespace pd = paddle;
int x = 10;
int y = 1;
int z = 10;
bool cond = false;
int o1, o2;
if (cond) {
int z = x + y;
o1 = z;
o2 = pd::layer::softmax(z);
} else {
int d = pd::layer::fc(z);
o1 = d;
o2 = d+1;
}
```
An equivalent PaddlePaddle program from the design doc of the [IfElseOp operator](./if_else_op.md) is as follows:
```python
import paddle as pd
x = minibatch([10, 20, 30]) # shape=[None, 1]
y = var(1) # shape=[1], value=1
z = minibatch([10, 20, 30]) # shape=[None, 1]
cond = larger_than(x, 15) # [false, true, true]
ie = pd.ifelse()
with ie.true_block():
d = pd.layer.add_scalar(x, y)
ie.output(d, pd.layer.softmax(d))
with ie.false_block():
d = pd.layer.fc(z)
ie.output(d, d+1)
o1, o2 = ie(cond)
```
In both examples, the left branch computes `x+y` and `softmax(x+y)`, the right branch computes `fc(x)` and `x+1` .
The difference is that variables in the C++ program contain scalar values, whereas those in the PaddlePaddle programs are mini-batches of instances.
### Blocks with `for` and `RNNOp`
The following RNN model in PaddlePaddle from the [RNN design doc](./rnn.md) :
```python
x = sequence([10, 20, 30]) # shape=[None, 1]
m = var(0) # shape=[1]
W = var(0.314, param=true) # shape=[1]
U = var(0.375, param=true) # shape=[1]
rnn = pd.rnn()
with rnn.step():
h = rnn.memory(init = m)
h_prev = rnn.previous_memory(h)
a = layer.fc(W, x)
b = layer.fc(U, h_prev)
s = pd.add(a, b)
act = pd.sigmoid(s)
rnn.update_memory(h, act)
rnn.output(a, b)
o1, o2 = rnn()
```
has its equivalent C++ program as follows
```c++
int* x = {10, 20, 30};
int* m = {0};
int* W = {0.314};
int* U = {0.375};
int mem[sizeof(x) / sizeof(x[0]) + 1];
int o1[sizeof(x) / sizeof(x[0]) + 1];
int o2[sizeof(x) / sizeof(x[0]) + 1];
for (int i = 1; i <= sizeof(x)/sizeof(x[0]); ++i) {
int x = x[i-1];
if (i == 1) mem[0] = m;
int a = W * x;
int b = Y * mem[i-1];
int s = fc_out + hidden_out;
int act = sigmoid(sum);
mem[i] = act;
o1[i] = act;
o2[i] = hidden_out;
}
```
## Compilation and Execution
Like TensorFlow, a PaddlePaddle program is written in Python. The first part describes a neural network as a protobuf message, and the rest executes the message for training or inference.
The generation of this protobuf message is similar to how a compiler generates a binary executable file. The execution of the message is similar to how the OS executes the binary file.
## The "Binary Executable File Format"
The definition of the protobuf message is as follows:
```protobuf
message BlockDesc {
repeated VarDesc vars = 1;
repeated OpDesc ops = 2;
}
```
The step net in above RNN example would look like
```
BlockDesc {
vars = {
VarDesc {...} // x
VarDesc {...} // h
VarDesc {...} // fc_out
VarDesc {...} // hidden_out
VarDesc {...} // sum
VarDesc {...} // act
}
ops = {
OpDesc {...} // matmul
OpDesc {...} // add_two
OpDesc {...} // sigmoid
}
};
```
Also, the RNN operator in above example is serialized into a protobuf message of type `OpDesc` and would look like:
```
OpDesc {
inputs = {0} // the index of x in vars of BlockDesc above
outputs = {5, 3} // indices of act and hidden_out in vars of BlockDesc above
attrs {
"states" : {1} // the index of h
"step_net" : <above step net>
}
};
```
This `OpDesc` value is in the `ops` field of the `BlockDesc` value representing the global block.
## The Compilation of Blocks
During the generation of the Protobuf message, the Block should store VarDesc (the Protobuf message which describes Variable) and OpDesc (the Protobuf message which describes Operator).
VarDesc in a block should have its name scope to avoid local variables affect parent block's name scope.
Child block's name scopes should inherit the parent's so that OpDesc in child block can reference a VarDesc that stored in parent block. For example:
```python
a = pd.Variable(shape=[20, 20])
b = pd.fc(a, params=["fc.w", "fc.b"])
rnn = pd.create_rnn()
with rnn.stepnet():
x = a.as_step_input()
# reuse fc's parameter
fc_without_b = pd.get_variable("fc.w")
rnn.output(fc_without_b)
out = rnn()
```
The method `pd.get_variable` can help retrieve a Variable by the name. The Variable may be stored in a parent block, but might be retrieved in a child block, so block should have a variable scope that supports inheritance.
In compiler design, the symbol table is a data structure created and maintained by compilers to store information about the occurrence of various entities such as variable names, function names, classes, etc.
To store the definition of variables and operators, we define a C++ class `SymbolTable`, like the one used in compilers.
`SymbolTable` can do the following:
- store the definitions (some names and attributes) of variables and operators,
- verify if a variable was declared,
- make it possible to implement type checking (offer Protobuf message pointers to `InferShape` handlers).
```c++
// Information in SymbolTable is enough to trace the dependency graph. So maybe
// the Eval() interface takes a SymbolTable is enough.
class SymbolTable {
public:
SymbolTable(SymbolTable* parent) : parent_(parent) {}
OpDesc* NewOp(const string& name="");
// TODO determine whether name is generated by python or C++.
// Currently assume that a unique name will be generated by C++ if the
// argument name is left default.
VarDesc* Var(const string& name="");
// find a VarDesc by name, if recursive is true, find parent's SymbolTable
// recursively.
// this interface is introduced to support InferShape, find protobuf messages
// of variables and operators, pass pointers into InferShape.
//
// NOTE maybe some C++ classes such as VarDescBuilder and OpDescBuilder should
// be proposed and embedded into pybind to enable python operation on C++ pointers.
VarDesc* FindVar(const string& name, bool recursive=true);
OpDesc* FindOp(const string& name);
BlockDesc Compile() const;
private:
SymbolTable* parent_;
map<string, OpDesc> ops_;
map<string, VarDesc> vars_;
};
```
After all the description of variables and operators is added into SymbolTable,
the block has enough information to run.
The `Block` class takes a `BlockDesc` as input, and provides `Run` and `InferShape` functions.
```c++
namespace {
class Block : OperatorBase {
public:
Block(const BlockDesc& desc) desc_(desc) {}
void InferShape(const framework::Scope& scope) const override {
if (!symbols_ready_) {
CreateVariables(scope);
CreateOperators();
}
// should run InferShape first.
for (auto& op : runtime_table_.ops()) {
op->InferShape(scope);
}
}
void Run(const framework::Scope& scope,
const platform::DeviceContext& dev_ctx) const override {
PADDLE_ENFORCE(symbols_ready_, "operators and variables should be created first.");
for (auto& op : runtime_table_.ops()) {
op->Run(scope, dev_ctx);
}
}
void CreateVariables(const framework::Scope& scope);
void CreateOperators();
// some other necessary interfaces of NetOp are listed below
// ...
private:
BlockDesc desc_;
bool symbols_ready_{false};
};
```
## The Execution of Blocks
Block inherits from OperatorBase, which has a Run method.
Block's Run method will run its operators sequentially.
There is another important interface called `Eval`, which takes some arguments called targets and generates a minimal graph which treats targets as the end points and creates a new Block. After `Run`, `Eval` will get the latest value and return the targets.
The definition of Eval is as follows:
```c++
// clean a block description by targets using the corresponding dependency graph.
// return a new BlockDesc with minimal number of operators.
// NOTE: The return type is not a Block but the block's description so that this can be distributed
// to a cluster.
BlockDesc Prune(const BlockDesc& desc, vector<string> targets);
void Block::Eval(const vector<string>& targets,
const framework::Scope& scope,
const platform::DeviceContext& dev_ctx) {
BlockDesc min_desc = Prune(desc_, targets);
Block min_block(min_desc);
min_block.Run(scope, dev_ctx);
}
```
doc/design/cluster_train/README.md
浏览文件 @ 1ffdecf1
......@@ -54,17 +54,18 @@ The life cycle of a single task is illustrated below:
<img src="src/paddle-task-states.png"/>
1. When a new pass of training starts, all tasks will be placed in the todo queue.
1. The master server will dispatch few tasks to each trainer at a time, puts them in the pending queue and waits for completion.
1. The trainer will work on its tasks and tell the master server once a task is completed. The master server will dispatch a new task to that trainer.
1. If a task timeout. the master server will move it back to the todo queue. The timeout count will increase by one. If the timeout count is above a threshold, the task is likely to cause a trainer to crash, so it will be discarded.
1. Upon trainer requests for new task, the master server will dispatch a task from todo queue to it, put the task in the pending queue and wait for completion.
1. The trainer will work on its task and tell the master server once the task is completed and ask for new task. The master server will dispatch a new task to that trainer.
1. If a task fails for any reason in trainer, or takes longer than a specific period of time, the master server will move the task back to the todo queue. The timeout count for that task will increase by one. If the timeout count is above a threshold, the task is likely to cause a trainer to crash, then it will be discarded.
1. The master server will move completed task to the done queue. When the todo queue is empty, the master server will start a new pass by moving all tasks in the done queue to todo queue and reset the timeout counter of all tasks to zero.
### Trainer Process
The trainer process will:
- Receive tasks from the master.
- Work on the tasks: calculate and upload gradient to parameter servers, and update local model by downloading new parameters from parameter servers.
- Request tasks from the master.
- Work on the tasks
- Upload gradient to parameter servers, and update local model by downloading new parameters from parameter servers.
### Parameter Server Process
......@@ -119,8 +120,8 @@ When the master is started by the Kubernetes, it executes the following steps at
1. Grabs a unique *master* lock in etcd, which prevents concurrent master instantiations.
1. Recovers the task queues from etcd if they already exist, otherwise, the master will create them.
1. Watches the trainer prefix keys `/trainer/` on etcd to find the live trainers.
1. Starts dispatching the tasks to the trainers, and updates task queue using an etcd transaction to ensure lock is held during the update.
1. Write its ip address to */master/addr* so that trainers can discover it.
1. Listens to trainers' request of task, dispatch one upon request, and updates task queue using an etcd transaction to ensure lock is held during the update.
When the master server process is dead for any reason, Kubernetes will restart it. It will be online again with all states recovered from etcd in few minutes.
......@@ -128,13 +129,11 @@ When the master server process is dead for any reason, Kubernetes will restart i
When the trainer is started by the Kubernetes, it executes the following steps at startup:
1. Watches the available parameter server prefix keys `/ps/` on etcd and waits until the count of parameter servers reaches the desired count.
1. Generates a unique ID, and sets key `/trainer/<unique ID>` with its contact address as value. The key will be deleted when the lease expires, so the master will be aware of the trainer being online and offline.
1. Waits for tasks from the master to start training.
1. Watches the available parameter server prefix keys `/ps/` on etcd and waits until the count of parameter servers reaches the desired count */ps_desired*.
1. Finds and watches */master/addr* to get master's address.
1. Requests for tasks from the master to start training.
If trainer's etcd lease expires, it will try set key `/trainer/<unique ID>` again so that the master server can discover the trainer again.
When a trainer fails, Kuberentes would try to restart it. The recovered trainer would fetch tasks from the TODO queue and go on training.
When a trainer fails, Kuberentes would try to restart it. The recovered trainer would fetch tasks from master and go on training.
### Parameter Server Process
......
doc/design/executor.md 0 → 100644
浏览文件 @ 1ffdecf1
# Executor Design Doc
## Motivation
We use executor to do the runtime evaluation of a `ProgramDesc`.
## Overview
An executor takes a `ProgramDesc`, a `block_id` and a `Scope`. The `ProgramDesc` is a list of blocks and each block contains the protobuf definition of all the parameters and operators. The `block_id` specifies the entrance block. And the `Scope` is the container of all the variable instance, which is persistent throughout different runs.
### What does executor do?
It evaluates all the operators in the `block_id`th block of a `ProgramDesc`.
### What does executor NOT do?
It does not do runtime optimization, meaning intelligently parse the dependency of each op a choose which one to be run and in which order they should be run.
It does not do graph partitioning, meaning dividing the `ProgramDesc` into several small pieces and executing them on different devices.
## Implementation
`Executor` evaluates a `ProgramDesc`. Essentially, it instantiates Variables and Operators, then run all the operators in sequence. [[code]](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/framework/executor.cc)
doc/design/float16.md 0 → 100644
浏览文件 @ 1ffdecf1
# Design Doc: float16
## Why float16
Half precision (float16) is a binary floating-point format that occupies 16 bits in memory. float16 is half the size of traditional 32-bit single precision format (float) and has lower precision and smaller range.
When high precision computation is not required, using float16 data type could potentially
- reduce storage space, memory bandwidth, and power usages;
- increase the chance of data fitting into a smaller cache of lower latency;
- provide arithmetic speed up if supported by hardware.
## Survey of current float16 support
A brief survey of float16 support on different compilers, hardwares, and libraries can be found below. Interested readers can refer to [link1](https://github.com/PaddlePaddle/Paddle/issues/4853) and [link2](https://github.com/Xreki/Xreki.github.io/blob/master/multi_data_types_in_dl_framework/ppt/float16_and_quantized_type.md) for more info.
The goal of float16 is to serve as a key for the executor to find and run the correct version of compute method specialized for float16 in operator kernel. It should be compatible with various natively supported float16 implementations including `__half` for cuda, `float16_t` for ARM, and `Eigen::half` for Eigen to make writing customized float16 kernels easier.
### Compiler
- nvcc supports `__half` data type after CUDA 7.5.
- `__fp16` or `float16_t` is supported as storage type for gcc >= 6.1 and clang >= 3.4.
- `__fp16` or `float16_t` is supported as arithmetic type for gcc >= 7.1 and clang >= 3.9.
### Hardware
- `__half` is supported on GPU with compute capability >= 5.3.
- `__fp16` is supported as storage type for ARMv7-A, ARMv8-A, and above.
- `__fp16` is supported as arithmetic type after ARMv8.2-A (currently, the only microarchitecture implementing ARMv8.2-A is ARM Cortex-A75, which is announced in May 2017. There seems to be no application processors currently available on market that adopts this architecture. It is reported that Qualcomm Snapdragon 845 uses Cortex-A75 design and will be available in mobile devices in early 2018).
### Libraries
- [Eigen](https://github.com/RLovelett/eigen) >= 3.3 supports float16 calculation on both GPU and CPU using the `Eigen::half` class. It is mostly useful for Nvidia GPUs because of the overloaded arithmetic operators using cuda intrinsics. It falls back to using software emulation on CPU for calculation and there is no special treatment to ARM processors.
- [ARM compute library](https://github.com/ARM-software/ComputeLibrary) >= 17.02.01 supports NEON FP16 kernels (requires ARMv8.2-A CPU).
## Implementation
The float16 class holds a 16-bit `uint16_t` data internally.
```
struct float16 {
uint16_t x;
};
```
float16 supports the following features:
- constructors / assignment operators that take input from primitive data types including bool, integers of various length, float, and double.
- constructors / assignment operators that take input from `__half` on cuda, `float16_t` on ARM, and `Eigen::half` on Eigen.
- conversion operators to primitive data types and half precision data types on cuda, ARM and Eigen.
- overloaded arithmetic operators for cuda, arm, and non-arm cpu, respectively. These operators will take advantage of the cuda and ARM intrinsics on the corresponding hardware.
To support the above features, two fundamental conversion functions are provided:
```
float16 float_to_half_rn(float f); // convert to half precision in round-to-nearest-even mode
float half_to_float(float16 h);
```
which provides one-to-one conversion between float32 and float16. These twos functions will do different conversion routines based on the current hardware. CUDA/ARM instrinsics will be used when the corresonding hardware is available. If the hardware or compiler level does not support float32 to float16 conversion, software emulation will be performed to do the conversion.
## To do
After float16 class is available, some of the future items are below:
- Update pybind/tensor_py.h to bind c++ float16 with numpy float16.
- Modify `GetKernelType()` method in `framework/operator.h` to make it compatible with float16.
- Create a type-casting operator that can convert the data type in tensor between float16 and other types.
doc/design/functions_operators_layers.md 0 → 100644
浏览文件 @ 1ffdecf1
# Design Doc: Functions, Operators, and Layers
In a DL system, we can compose one or more fine grained operators into a coarse grained one. For example, the FC layer can be composed of a multiplication operator and an add operator.
Historically, some fine grained operations are known as operators, and some coarse level ones are known as layers. But we need a well-defined separation.
In general, operators are those very fine grained operations, e.g., mul and add. In the implementation, we can write them as C++ functions:
```c++
template <typename T> T add(T x, T y) { return x + y; }
template <typename T> T mul(T x, T y) { return x * y; }
```
Then we can wrap them into operators which are C++ classes and can be created from Python bindings by name. A C macro can do this. For example, the following macro invocation
```c++
#define MAKE_FUNCTION_OPERATOR(mul);
```
generates
```c++
template <typename T> class mulOp : public OperatorBase {...};
REGISTER_OP(mulOp<float32>, "mul");
```
so that in Python we can create operator mul by:
```python
X1 = Var()
X2 = Var()
Y = Var()
paddle.cpp.create_operator("mul", input=[X1, X2], output=Y)
```
Also, at the same time, we can compose a coarse level C++ operator class by composing functions `mul` and `add`:
```c++
template <typename T>
class FCOp : public OperatorBase {
public:
void Run(...) {
add(mul(Input<T>("X"), Input<T>("W")), Input<T>("b");
}
};
REGISTER_OP(FCOp, "fc");
```
We need to support such composition in Python as well. To do so, we need a higher level Python wrapping of operator creation than `paddle.cpp.create_operator`. This higher level operator API should be compatible with the layer API.
Let's explain using an example. Suppose that we are going to compose the FC using mul and add in Python, we'd like to have Python functions `mul` and `add` defined in module `operator`:
```python
def operator.mul(X1, X2):
O = Var()
paddle.cpp.create_operator("mul", input={X1, Y1}, output=O)
return O
def operator.add(X1, X2):
O = Var()
paddle.cpp.create_operator("add", input={X1, X2}, output=O)
return O
```
Above code snippets are automatically generated. Given them, users can define
```python
def layer.fc(X):
W = Var()
b = Var()
return operator.add(operator.mul(X, W), b)
```
If we don't have `operator.mul` and `operator.add`, the definiton of `layer.fc` would be complicated:
```python
def layer.fc(X):
W = Var()
b = Var()
O1 = Var()
paddle.cpp.create_operator("mul", input=[X, W], output=O1)
O2 = Var()
paddle.cpp.create_operator("add", input=[O1, b], output=O2)
return O2
```
We'd like to have Python bindings to operators in package `paddle.operator`, and Python compositions of operators in package `paddle.layer`. So we have the following concepts in above illustrative example:
| C++ functions/functors | mul | add | | |
|------------------------|--------------|--------------|-------------|----------|
| C++ operator class | mulOp | addOp | FCOp | |
| Python binding | operator.mul | operator.add | operator.fc | |
| Python function | | | | layer.fc |
This is how we differentiate layer and operators in PaddlePaddle:
- those defined in C++ and have a lightweighted Python wrapper in module `operators` are operators; whereas
- those who don't have C++ implementations but a Python implementation that compose C++ operators are known as layers.
doc/design/gan_api.md 0 → 100644
浏览文件 @ 1ffdecf1
# Design for GAN
GAN (General Adversarial Net [https://arxiv.org/abs/1406.2661]) is an important model for unsupervised learning and widely used in many areas.
It applies several important concepts in machine learning system design, including building and running subgraphs, dependency tracing, different optimizers in one executor and so forth.
In our GAN design, we wrap it as a user-friendly easily customized python API to design different models. We take the conditional DC-GAN (Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks [https://arxiv.org/abs/1511.06434]) as an example due to its good performance on image generation.
<p align="center">
<img src="./test.dot.png" width = "35%" align="center"/><br/>
Figure 1. The overall running logic of GAN. The black solid arrows indicate the forward pass; the green dashed arrows indicate the backward pass of generator training; the red dashed arrows indicate the backward pass of the discriminator training. The BP pass of the green (red) arrow should only update the parameters in the green (red) boxes. The diamonds indicate the data providers. d\_loss and g\_loss marked in red and green are the two targets we would like to run.
</p>
The operators, layers and functions required/optional to build a GAN demo is summarized in https://github.com/PaddlePaddle/Paddle/issues/4563.
<p align="center">
<img src="./dcgan.png" width = "90%" align="center"/><br/>
Figure 2. Photo borrowed from the original DC-GAN paper.
</p>
## The Conditional-GAN might be a class.
This design we adopt the popular open source design in https://github.com/carpedm20/DCGAN-tensorflow and https://github.com/rajathkmp/DCGAN. It contains following data structure:
- DCGAN(object): which contains everything required to build a GAN model. It provides following member functions methods as API:
- __init__(...): Initialize hyper-parameters (like conv dimension and so forth), and declare model parameters of discriminator and generator as well.
- generator(z, y=None): Generate a fake image from input noise z. If the label y is provided, the conditional GAN model will be chosen.
Returns a generated image.
- discriminator(image):
Given an image, decide if it is from a real source or a fake one.
Returns a 0/1 binary label.
- build_model(self):
build the whole GAN model, define training loss for both generator and discrimator.
## Discussion on Engine Functions required to build GAN
- Trace the tensor and variable dependency in the engine executor. (Very critical, otherwise GAN can'be be trained correctly)
- Different optimizers responsible for optimizing different loss.
To be more detailed, we introduce our design of DCGAN as following:
### Class member Function: Initializer
- Set up hyper-parameters, including condtional dimension, noise dimension, batch size and so forth.
- Declare and define all the model variables. All the discriminator parameters are included in the list self.theta_D and all the generator parameters are included in the list self.theta_G.
```python
class DCGAN(object):
def __init__(self, y_dim=None):
# hyper parameters
self.y_dim = y_dim # conditional gan or not
self.batch_size = 100
self.z_dim = z_dim # input noise dimension
# define parameters of discriminators
self.D_W0 = pd.Variable(shape=[3,3, 1, 128], data=pd.gaussian_normal_randomizer())
self.D_b0 = pd.Variable(np.zeros(128)) # variable also support initialization using a numpy data
self.D_W1 = pd.Variable(shape=[784, 128], data=pd.gaussian_normal_randomizer())
self.D_b1 = pd.Variable(np.zeros(128)) # variable also support initialization using a numpy data
self.D_W2 = pd.Varialble(np.random.rand(128, 1))
self.D_b2 = pd.Variable(np.zeros(128))
self.theta_D = [self.D_W0, self.D_b0, self.D_W1, self.D_b1, self.D_W2, self.D_b2]
# define parameters of generators
self.G_W0 = pd.Variable(shape=[784, 128], data=pd.gaussian_normal_randomizer())
self.G_b0 = pd.Variable(np.zeros(128)) # variable also support initialization using a numpy data
self.G_W1 = pd.Variable(shape=[784, 128], data=pd.gaussian_normal_randomizer())
self.G_b1 = pd.Variable(np.zeros(128)) # variable also support initialization using a numpy data
self.G_W2 = pd.Varialble(np.random.rand(128, 1))
self.G_b2 = pd.Variable(np.zeros(128))
self.theta_G = [self.G_W0, self.G_b0, self.G_W1, self.G_b1, self.G_W2, self.G_b2]
```
### Class member Function: Generator
- Given a noisy input z, returns a fake image.
- Concatenation, batch-norm, FC operations required;
- Deconv layer required, which is missing now...
```python
class DCGAN(object):
def generator(self, z, y = None):
# input z: the random noise
# input y: input data label (optional)
# output G_im: generated fake images
if not self.y_dim:
z = pd.layer.concat(1, [z, y])
G_h0 = pd.layer.fc(z, self.G_w0, self.G_b0)
G_h0_bn = pd.layer.batch_norm(G_h0)
G_h0_relu = pd.layer.relu(G_h0_bn)
G_h1 = pd.layer.deconv(G_h0_relu, self.G_w1, self.G_b1)
G_h1_bn = pd.layer.batch_norm(G_h1)
G_h1_relu = pd.layer.relu(G_h1_bn)
G_h2 = pd.layer.deconv(G_h1_relu, self.G_W2, self.G_b2))
G_im = pd.layer.tanh(G_im)
return G_im
```
### Class member function: Discriminator
- Given a noisy input z, returns a fake image.
- Concatenation, Convolution, batch-norm, FC, Leaky-ReLU operations required;
```python
class DCGAN(object):
def discriminator(self, image):
# input image: either generated images or real ones
# output D_h2: binary logit of the label
D_h0 = pd.layer.conv2d(image, w=self.D_w0, b=self.D_b0)
D_h0_bn = pd.layer.batchnorm(h0)
D_h0_relu = pd.layer.lrelu(h0_bn)
D_h1 = pd.layer.conv2d(D_h0_relu, w=self.D_w1, b=self.D_b1)
D_h1_bn = pd.layer.batchnorm(D_h1)
D_h1_relu = pd.layer.lrelu(D_h1_bn)
D_h2 = pd.layer.fc(D_h1_relu, w=self.D_w2, b=self.D_b2)
return D_h2
```
### Class member function: Build the model
- Define data readers as placeholders to hold the data;
- Build generator and discriminators;
- Define two training losses for discriminator and generator, respectively.
If we have execution dependency engine to back-trace all tensors, the module building our GAN model will be like this:
```python
class DCGAN(object):
def build_model(self):
if self.y_dim:
self.y = pd.data(pd.float32, [self.batch_size, self.y_dim])
self.images = pd.data(pd.float32, [self.batch_size, self.im_size, self.im_size])
self.faked_images = pd.data(pd.float32, [self.batch_size, self.im_size, self.im_size])
self.z = pd.data(tf.float32, [None, self.z_size])
# step 1: generate images by generator, classify real/fake images with discriminator
if self.y_dim: # if conditional GAN, includes label
self.G = self.generator(self.z, self.y)
self.D_t = self.discriminator(self.images)
# generated fake images
self.sampled = self.sampler(self.z, self.y)
self.D_f = self.discriminator(self.G)
else: # original version of GAN
self.G = self.generator(self.z)
self.D_t = self.discriminator(self.images)
# generate fake images
self.sampled = self.sampler(self.z)
self.D_f = self.discriminator(self.images)
# step 2: define the two losses
self.d_loss_real = pd.reduce_mean(pd.cross_entropy(self.D_t, np.ones(self.batch_size))
self.d_loss_fake = pd.reduce_mean(pd.cross_entropy(self.D_f, np.zeros(self.batch_size))
self.d_loss = self.d_loss_real + self.d_loss_fake
self.g_loss = pd.reduce_mean(pd.cross_entropy(self.D_f, np.ones(self.batch_szie))
```
If we do not have dependency engine but blocks, the module building our GAN model will be like this:
```python
class DCGAN(object):
def build_model(self, default_block):
# input data in the default block
if self.y_dim:
self.y = pd.data(pd.float32, [self.batch_size, self.y_dim])
self.images = pd.data(pd.float32, [self.batch_size, self.im_size, self.im_size])
# self.faked_images = pd.data(pd.float32, [self.batch_size, self.im_size, self.im_size])
self.z = pd.data(tf.float32, [None, self.z_size])
# step 1: generate images by generator, classify real/fake images with discriminator
with pd.default_block().g_block():
if self.y_dim: # if conditional GAN, includes label
self.G = self.generator(self.z, self.y)
self.D_g = self.discriminator(self.G, self.y)
else: # original version of GAN
self.G = self.generator(self.z)
self.D_g = self.discriminator(self.G, self.y)
self.g_loss = pd.reduce_mean(pd.cross_entropy(self.D_g, np.ones(self.batch_szie))
with pd.default_block().d_block():
if self.y_dim: # if conditional GAN, includes label
self.D_t = self.discriminator(self.images, self.y)
self.D_f = self.discriminator(self.G, self.y)
else: # original version of GAN
self.D_t = self.discriminator(self.images)
self.D_f = self.discriminator(self.G)
# step 2: define the two losses
self.d_loss_real = pd.reduce_mean(pd.cross_entropy(self.D_t, np.ones(self.batch_size))
self.d_loss_fake = pd.reduce_mean(pd.cross_entropy(self.D_f, np.zeros(self.batch_size))
self.d_loss = self.d_loss_real + self.d_loss_fake
```
Some small confusion and problems with this design:
- D\_g and D\_f are actually the same thing, but has to be written twice; i.e., if we want to run two sub-graphs conceptually, the same codes have to be written twice if they are shared by the graph.
- Requires ability to create a block anytime, rather than in if-else or rnn only;
## Main function for the demo:
Generally, the user of GAN just need to the following things:
- Define an object as DCGAN class;
- Build the DCGAN model;
- Specify two optimizers for two different losses with respect to different parameters.
```python
# pd for short, should be more concise.
from paddle.v2 as pd
import numpy as np
import logging
if __name__ == "__main__":
# dcgan class in the default graph/block
# if we use dependency engine as tensorflow
# the codes, will be slightly different like:
# dcgan = DCGAN()
# dcgan.build_model()
with pd.block() as def_block:
dcgan = DCGAN()
dcgan.build_model(def_block)
# load mnist data
data_X, data_y = self.load_mnist()
# Two subgraphs required!!!
with pd.block().d_block():
d_optim = pd.train.Adam(lr = .001, beta= .1)
d_step = d_optim.minimize(dcgan.d_loss, dcgan.theta_D)
with pd.block.g_block():
g_optim = pd.train.Adam(lr = .001, beta= .1)
g_step = pd.minimize(dcgan.g_loss, dcgan.theta_G)
# executor
sess = pd.executor()
# training
for epoch in xrange(10000):
for batch_id in range(N / batch_size):
idx = ...
# sample a batch
batch_im, batch_label = data_X[idx:idx+batch_size], data_y[idx:idx+batch_size]
# sample z
batch_z = np.random.uniform(-1., 1., [batch_size, z_dim])
if batch_id % 2 == 0:
sess.run(d_step,
feed_dict = {dcgan.images: batch_im,
dcgan.y: batch_label,
dcgan.z: batch_z})
else:
sess.run(g_step,
feed_dict = {dcgan.z: batch_z})
```
# More thinking about dependency engine v.s. block design:
- What if we just want to run an intermediate result? Do we need to run the whole block/graph?
- Should we call eval() to get the fake images in the first stage? And then train the discriminator in the second stage?
doc/design/graph.md 0 → 100644
浏览文件 @ 1ffdecf1
# Design Doc: Computations as a Graph
A primary goal of the refactorization of PaddlePaddle is a more flexible representation of deep learning computation, in particular, a graph of operators and variables, instead of sequences of layers as before.
This document explains that the construction of a graph as three steps:
- construct the forward part
- construct the backward part
- construct the optimization part
## The Construction of a Graph
Let us take the problem of image classification as a simple example. The application program that trains the model looks like:
```python
x = layer.data("images")
l = layer.data("label")
y = layer.fc(x)
cost = layer.mse(y, l)
optimize(cost)
train(cost, reader=mnist.train())
```
### Forward Part
The first four lines of above program build the forward part of the graph.
![](images/graph_construction_example_forward_only.png)
In particular, the first line `x = layer.data("images")` creates variable x and a Feed operator that copies a column from the minibatch to x. `y = layer.fc(x)` creates not only the FC operator and output variable y, but also two parameters, W and b, and the initialization operators.
Initialization operators are kind of "run-once" operators -- the `Run` method increments a class data member counter so to run at most once. By doing so, a parameter wouldn't be initialized repeatedly, say, in every minibatch.
In this example, all operators are created as `OpDesc` protobuf messages, and all variables are `VarDesc`. These protobuf messages are saved in a `BlockDesc` protobuf message.
### Backward Part
The fifth line `optimize(cost)` calls two functions, `ConstructBackwardGraph` and `ConstructOptimizationGraph`.
`ConstructBackwardGraph` traverses the forward graph in the `BlockDesc` protobuf message and builds the backward part.
![](images/graph_construction_example_forward_backward.png)
According to the chain rule of gradient computation, `ConstructBackwardGraph` would
1. create a gradient operator G for each operator F,
1. make all inputs, outputs, and outputs' gradient of F as inputs of G,
1. create gradients for all inputs of F, except for those who don't have gradients, like x and l, and
1. make all these gradients as outputs of G.
### Optimization Part
For each parameter, like W and b created by `layer.fc`, marked as double circles in above graphs, `ConstructOptimizationGraph` creates an optimization operator to apply its gradient. Here results in the complete graph:
![](images/graph_construction_example_all.png)
## Block and Graph
The word block and graph are interchangable in the desgin of PaddlePaddle. A [Block](https://github.com/PaddlePaddle/Paddle/pull/3708) is a metaphore of the code and local variables in a pair of curly braces in programming languages, where operators are like statements or instructions. A graph of operators and variables is a representation of the block.
A Block keeps operators in an array `BlockDesc::ops`
```protobuf
message BlockDesc {
repeated OpDesc ops = 1;
repeated VarDesc vars = 2;
}
```
in the order that they appear in user programs, like the Python program at the beginning of this article. We can imagine that in `ops`, we have some forward operators, followed by some gradient operators, and then some optimization operators.
doc/design/graph_survey.md 0 → 100644
浏览文件 @ 1ffdecf1
## Survey on Graph
Neural network framework often provides symbolic API for users to write network topology conveniently. This doc manily focus on symbolic API in most popular neural network frameworks, and try to find out how to parse symbolic configuration to a portable file, such as protobuf or json.
### Mxnet
The core concept of symbolic API is `Symbol`. Mxnet implements `Symbol` class in C++, and export to Python using C-API. Please refer to the comments in Mxnet:
`Symbol` is help class used to represent the operator node in Graph.
`Symbol` acts as an interface for building graphs from different components like Variable, Functor and Group. `Symbol` is also exported to python front-end (while Graph is not) to enable quick test and deployment. Conceptually, symbol is the final operation of a graph and thus including all the information required (the graph) to evaluate its output value.
A simple network topology wrote by Symbol is as follows:
```python
def get_symbol(num_classes=10, **kwargs):
data = mx.symbol.Variable('data')
data = mx.symbol.Flatten(data=data)
fc1 = mx.symbol.FullyConnected(data = data, name='fc1', num_hidden=128)
act1 = mx.symbol.Activation(data = fc1, name='relu1', act_type="relu")
fc2 = mx.symbol.FullyConnected(data = act1, name = 'fc2', num_hidden = 64)
act2 = mx.symbol.Activation(data = fc2, name='relu2', act_type="relu")
fc3 = mx.symbol.FullyConnected(data = act2, name='fc3', num_hidden=num_classes)
mlp = mx.symbol.SoftmaxOutput(data = fc3, name = 'softmax')
return mlp
```
Varible here is actually a Symbol. Every basic Symbol will correspond to one Node, and every Node has its own NodeAttr. There is a op field in NodeAttr class, when a Symbol represents Variable(often input data), the op field is null.
Symbol contains a data member, std::vector<NodeEntry> outputs, and NodeEntry cantains a poniter to Node. We can follow the Node pointer to get all the Graph.
And Symbol can be saved to a Json file.
Here is a detailed example:
```
>>> import mxnet as mx
>>> data = mx.symbol.Variable('data')
>>> print data.debug_str()
Variable:data
>>> data = mx.symbol.Flatten(data=data)
>>> print data.debug_str()
Symbol Outputs:
output[0]=flatten0(0)
Variable:data
--------------------
Op:Flatten, Name=flatten0
Inputs:
arg[0]=data(0) version=0
>>> fc1 = mx.symbol.FullyConnected(data = data, name='fc1', num_hidden=128)
>>> print fc1.debug_str()
Symbol Outputs:
output[0]=fc1(0)
Variable:data
--------------------
Op:Flatten, Name=flatten0
Inputs:
arg[0]=data(0) version=0
Variable:fc1_weight
Variable:fc1_bias
--------------------
Op:FullyConnected, Name=fc1
Inputs:
arg[0]=flatten0(0)
arg[1]=fc1_weight(0) version=0
arg[2]=fc1_bias(0) version=0
Attrs:
num_hidden=128
```
### TensorFlow
The core concept of symbolic API is `Tensor`. Tensorflow defines `Tensor` in Python. Please refer to the comments in TensorFlow:
A `Tensor` is a symbolic handle to one of the outputs of an `Operation`. It does not hold the values of that operation's output, but instead provides a means of computing those values in a TensorFlow [Session](https://www.tensorflow.org/api_docs/python/tf/Session).
A simple example is as follows:
```python
# Build a dataflow graph.
c = tf.constant([[1.0, 2.0], [3.0, 4.0]])
d = tf.constant([[1.0, 1.0], [0.0, 1.0]])
e = tf.matmul(c, d)
# Construct a `Session` to execute the graph.
sess = tf.Session()
# Execute the graph and store the value that `e` represents in `result`.
result = sess.run(e)
```
The main method of `Tensor` is as follows:
```python
@property
def op(self):
"""The `Operation` that produces this tensor as an output."""
return self._op
@property
def dtype(self):
"""The `DType` of elements in this tensor."""
return self._dtype
@property
def graph(self):
"""The `Graph` that contains this tensor."""
return self._op.graph
@property
def name(self):
"""The string name of this tensor."""
if not self._op.name:
raise ValueError("Operation was not named: %s" % self._op)
return "%s:%d" % (self._op.name, self._value_index)
@property
def device(self):
"""The name of the device on which this tensor will be produced, or None."""
return self._op.device
```
Tensor can be taken as target to run by session. Tensor contains all the information of Graph, and tracks data dependency.
Here is a detailed example:
```
>>> import tensorflow as tf
>>> c = tf.constant([[1.0, 2.0], [3.0, 4.0]])
>>> print c.graph
<tensorflow.python.framework.ops.Graph object at 0x10f256d50>
>>> d = tf.constant([[1.0, 1.0], [0.0, 1.0]])
>>> print d.graph
<tensorflow.python.framework.ops.Graph object at 0x10f256d50>
>>> e = tf.matmul(c, d)
>>> print e.graph
<tensorflow.python.framework.ops.Graph object at 0x10f256d50>
```
### Dynet
The core concept of symbolic API is `Expression`, and Dynet defines `Expression` class in C++.
A simple example is as follows:
```cpp
ComputationGraph cg;
Expression W = parameter(cg, pW);
Expression in = input(cg, xs[i]);
Expression label = input(cg, ys[i]);
Expression pred = W * in;
Expression loss = square(pred - label);
```
The input data and parameter are also represented by Expression. Every basci Expression corresponds to a Node. And input data is also a Node.
Expression has a data member ComputationGraph, and ComputationGraph will be modified in users' configuring process. Expression can be a running target, beacuse Expression contains all dependency.
Here is a detailed example:
write topology in C++
```
ComputationGraph cg;
Expression W = parameter(cg, pW);
cg.print_graphviz();
Expression pred = W * xs[i];
cg.print_graphviz();
Expression loss = square(pred - ys[i]);
cg.print_graphviz();
```
compile and print
```
# first print
digraph G {
rankdir=LR;
nodesep=.05;
N0 [label="v0 = parameters({1}) @ 0x7ffe4de00110"];
}
# second print
digraph G {
rankdir=LR;
nodesep=.05;
N0 [label="v0 = parameters({1}) @ 0x7ffe4de00110"];
N1 [label="v1 = v0 * -0.98"];
N0 -> N1;
}
# third print
digraph G {
rankdir=LR;
nodesep=.05;
N0 [label="v0 = parameters({1}) @ 0x7ffe4de00110"];
N1 [label="v1 = v0 * -0.98"];
N0 -> N1;
N2 [label="v2 = -1.88387 - v1"];
N1 -> N2;
N3 [label="v3 = -v2"];
N2 -> N3;
N4 [label="v4 = square(v3)"];
N3 -> N4;
}
```
### Conclusion
Actually, Symbol/Tensor/Expression in Mxnet/TensorFlow/Dynet are the same level concepts. We use a unified name Expression here, this level concept has following features:
- Users wirte topoloy with symbolic API, and all return value is Expression, including input data and parameter.
- Expression corresponds with a global Graph, and Expression can also be composed.
- Expression tracks all dependency and can be taken as a run target
doc/design/if_else_op.md 0 → 100644
浏览文件 @ 1ffdecf1
# The `IfElse` Operator
PaddlePaddle's `IfElse` operator differs from TensorFlow's:
- the TensorFlow version takes a scalar boolean value as the condition so that the whole mini-batch goes to either the true or the false branch, whereas
- the PaddlePaddle version takes a vector of boolean value as the condition, and instances corresponding to true values go to the true branch, those corresponding to false values go to the false branch.
## Example
The following PaddlePaddle program shows the usage of the IfElse operator:
```python
import paddle as pd
x = minibatch([10, 20, 30]) # shape=[None, 1]
y = var(1) # shape=[1], value=1
z = minibatch([10, 20, 30]) # shape=[None, 1]
cond = larger_than(x, 15) # [false, true, true]
ie = pd.ifelse()
with ie.true_block():
d = pd.layer.add(x, y)
ie.output(d, pd.layer.softmax(d))
with ie.false_block():
d = pd.layer.fc(z)
ie.output(d, d+1)
o1, o2 = ie(cond)
```
A challenge to implement the `IfElse` operator is to infer those variables to be split, or, say, to identify the variable of the mini-batch or those derived from the mini-batch.
An equivalent C++ program is as follows:
```c++
namespace pd = paddle;
int x = 10;
int y = 1;
int z = 10;
bool cond = false;
int o1, o2;
if (cond) {
int d = x + y;
o1 = z;
o2 = pd::layer::softmax(z);
} else {
int d = pd::layer::fc(z);
o1 = d;
o2 = d+1;
}
```
doc/design/images/graph_construction_example.bash 0 → 100755
浏览文件 @ 1ffdecf1
cat ./graph_construction_example.dot | \
sed 's/color=red/color=red, style=invis/g' | \
sed 's/color=green/color=green, style=invis/g' | \
dot -Tpng > graph_construction_example_forward_only.png
cat ./graph_construction_example.dot | \
sed 's/color=green/color=green, style=invis/g' | \
dot -Tpng > graph_construction_example_forward_backward.png
cat ./graph_construction_example.dot | \
dot -Tpng > graph_construction_example_all.png
doc/design/images/graph_construction_example.dot 0 → 100644
浏览文件 @ 1ffdecf1
digraph ImageClassificationGraph {
///////// The forward part /////////
FeedX [label="Feed", color=blue, shape=box];
FeedY [label="Feed", color=blue, shape=box];
InitW [label="Init", color=blue, shape=diamond];
Initb [label="Init", color=blue, shape=diamond];
FC [label="FC", color=blue, shape=box];
MSE [label="MSE", color=blue, shape=box];
x [label="x", color=blue, shape=oval];
l [label="l", color=blue, shape=oval];
y [label="y", color=blue, shape=oval];
W [label="W", color=blue, shape=doublecircle];
b [label="b", color=blue, shape=doublecircle];
cost [label="cost", color=blue, shape=oval];
FeedX -> x -> FC -> y -> MSE -> cost [color=blue];
FeedY -> l [color=blue];
InitW -> W [color=blue];
Initb -> b [color=blue];
W -> FC [color=blue];
b -> FC [color=blue];
l -> MSE [color=blue];
////////// The backward part /////////
MSE_Grad [label="MSE_grad", color=red, shape=box];
FC_Grad [label="FC_grad", color=red, shape=box];
d_cost [label="d cost", color=red, shape=oval];
d_y [label="d y", color=red, shape=oval];
d_b [label="d b", color=red, shape=oval];
d_W [label="d W", color=red, shape=oval];
cost -> MSE_Grad [color=red];
d_cost -> MSE_Grad [color=red];
l -> MSE_Grad [color=red];
y -> MSE_Grad -> d_y [color=red];
x -> FC_Grad [color=red];
y -> FC_Grad [color=red];
d_y -> FC_Grad [color=red];
W -> FC_Grad -> d_W [color=red];
b -> FC_Grad -> d_b [color=red];
////////// The optimizaiton part //////////
OPT_W [label="SGD", color=green, shape=box];
OPT_b [label="SGD", color=green, shape=box];
W -> OPT_W [color=green];
b -> OPT_b [color=green];
d_W -> OPT_W -> W [color=green];
d_b -> OPT_b -> b [color=green];
////////// Groupings //////////
subgraph clusterMSE {
style=invis;
MSE;
MSE_Grad;
}
subgraph clusterFC {
style=invis;
FC;
FC_Grad;
}
}
doc/design/infer_var_type.md 0 → 100644
浏览文件 @ 1ffdecf1
# Design Doc: InferVarType
## The Problem Posed
The variable in our design can hold variant types. Such as `LoDTensor` and `SelectedRows`. An operator should be able to inference the variable types of its output.
For example, a `lookup table` operator takes two `LoDTensor`; one is a float tensor as the embedding table, the other is an int tensor as word ID. The gradient operator of `lookup table` will generate a `SelectedRows` as its output. A `sum` operator can take both `LoDTensor` and `SelectedRows` as its inputs and will generate a `LoDTensor` if any of its inputs is `LoDTensor`, otherwise, the `sum` operator will generate `SelectedRows` as its output.
The variable type will be constant at runtime. Every variable's type can either be set by the user (input data and parameter) or be inferred by the operator in compile time.
## Proposed Solution
The `InferVarType` is a compile-time function which is registered to each operator. The inferface of that function is:
```c++
using InferVarTypeFN = std::function<
void (const OpDescBind& /*op_desc*/, BlockDescBind* /*block*/)>;
```
It takes an operator description as its input and will write the output variable type and store them in block description.
The `InferVarTypeFN` will be registered in `OpInfo`, to replace `infer_var_type_` field. The `OpInfo` should be
```cpp
struct OpInfo {
InferVarTypeFN infer_var_type_;
...
};
```
The default `InferVarType` will set output type as `LoDTensor`. It can be done by `GetInferVarType()`.
```cpp
void DefaultInferVarType(const OpDescBind& op_desc, BlockDescBind* block) {
// set the output type of variable as `LoDTensor`.
// ...
}
struct OpInfo {
InferVarTypeFN infer_var_type_;
InferVarTypeFN GetInferVarType() const {
if (infer_var_type_) {
return infer_var_type_;
} else {
return DefaultInferVarType;
}
}
};
```
## Register InferVarType
We provide a thin base class for registering an `InferVarTypeFN`. To use a base class will ease the implementation of registry since we can detect the registry entry is an `InferVarTypeFN` or not.
```cpp
class VarTypeInferer {
public:
virtual void operator()(const OpDescBind& op_desc, BlockDescBind* block) const = 0;
}
```
Operator developers can write the specialize `VarTypeInferer` as follow.
```cpp
class SpecialVarTypeInferer : public VarTypeInferer {
public:
virtual void operator()(const OpDescBind& op_desc, BlockDescBind* block) const {
// .. own logic
}
}
```
Then user can register the `InferVarType` just like `GradOpDescMaker` and `OpInfoMaker`.
```
REGISTER_OPERATOR(some_op, OpType, SpecialVarTypeInferer, ...);
```
doc/design/mkldnn/README.MD
浏览文件 @ 1ffdecf1
......@@ -101,6 +101,7 @@ if use_mkldnn
5. 在**Argument**里添加两个`MkldnnMatrixPtr`，取名为`mkldnnValue`和`mkldnnGrad`，用于存放`MkldnnLayer`会用到的memory buffer。 并且添加函数cvt(会修改为一个更加合适的函数名)，用于处理"CPU device"和"MKL-DNN device"之间memory的相互转化。
6. 在父类`Layer`中的`getOutput`函数中添加一段逻辑，用于判断`deviceId`，并针对device在MKL-DNN和CPU之间不统一的情况，做一个前期转换。 也就是调用`Argument`的cvt函数把output统一到需要的device上。
7. 在原来的`FLAGS`中添加一个`use_mkldnn`的flag，用于选择是否使用MKL-DNN的相关功能。
8. 关于MKLDNN参数的保存。由于MKLDNN参数的格式与PaddlePaddle原有的格式存在不一样的情况，所以需要在保存参数时同时保存该格式信息。目前准备扩展[Header](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/parameter/Parameter.h#L247)里面的`int32_t version`。这个值不管是在v1还是在v2里面，一直保存的是0，所以可以充分利用这个信息，定义一个枚举处理所有MKLDNN的参数格式，从而`MKLDNNLayer`就可以从输入的参数中获取需要的格式信息。
## References
......
doc/design/model_format.md 0 → 100644
浏览文件 @ 1ffdecf1
# Design Doc: Model Format
## Motivation
A model is an output of the training process. One complete model consists of two parts, the **topology** and the **parameters**. In order to support industrial deployment, the model format must be self-complete and must not expose any training source code.
As a result, In PaddlePaddle, the **topology** is represented as a [ProgramDesc](https://github.com/PaddlePaddle/Paddle/blob/1c0a4c901c9fc881d120249c703b15d1c50dae7d/doc/design/program.md), which describes the model structure. The **parameters** contain all the trainable weights in the model. We must support large size parameters and efficient serialization/deserialization of parameters.
## Implementation
The topology is saved as a plain text in a detailed self-contain protobuf file.
The parameters are saved as a binary file. As we all know, the protobuf message has a limit of [64M size](https://developers.google.com/protocol-buffers/docs/reference/cpp/google.protobuf.io.coded_stream#CodedInputStream.SetTotalBytesLimit.details). We have done a [benchmark experiment](https://github.com/PaddlePaddle/Paddle/pull/4610), which shows that protobuf is not fit for the task.
As a result, we design a particular format for tensor serialization. By default, an arbitrary tensor in Paddle is a [LoDTensor](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/framework/lod_tensor.md), and has a description information proto of [LoDTensorDesc](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/framework/framework.proto#L99). We save the DescProto as the byte string header. It contains all the necessary information, such as the `dims`, and the `LoD` information in [LoDTensor](https://github.com/PaddlePaddle/Paddle/blob/1c0a4c901c9fc881d120249c703b15d1c50dae7d/paddle/framework/lod_tensor.md). A tensor stores values in a continuous memory buffer. For speed we dump the raw memory to disk and save it as the byte string content. So, the binary format of one tensor is,
The table below shows a tensor's byte view in detail. Note that all the signed values are written in the little-endian format.
|field name | type | description |
| --- | --- | --- |
| version | uint32_t | Version of saved file. Always 0 now. |
| tensor desc length | uint32_t | TensorDesc(Protobuf message) length in bytes. |
| tensor desc | void* | TensorDesc protobuf binary message |
| tensor data | void* | Tensor's data in binary format. The length of `tensor_data` is decided by `TensorDesc.dims()` and `TensorDesc.data_type()` |
| lod_level | uint64_t | Level of LoD |
| length of lod[0] | uint64_t | [Optional] length of lod[0] in bytes. |
| data of lod[0] | uint64_t* | [Optional] lod[0].data() |
| ... | ... | ... |
## Summary
- We introduce a model format.
- The model represented by its forward-pass computation procedure is saved in a **ProgramDesc** protobuf message.
- A bunch of specified format binary tensors describe the **parameters**.
doc/design/ops/images/2_level_rnn.dot 0 → 100644
浏览文件 @ 1ffdecf1
digraph G {
rnn [label="1-th level RNN" shape=box]
subgraph cluster0 {
label = "time step 0"
sent0 [label="sentence"]
sent1 [label="sentence"]
rnn1 [label="2-th level RNN" shape=box]
sent0 -> rnn1
sent1 -> rnn1
}
subgraph cluster1 {
label = "time step 1"
sent2 [label="sentence"]
sent3 [label="sentence"]
rnn2 [label="2-th level RNN" shape=box]
sent2 -> rnn2
sent3 -> rnn2
}
subgraph cluster2 {
label = "time step 2"
sent4 [label="sentence"]
sent5 [label="sentence"]
rnn3 [label="2-th level RNN" shape=box]
sent4 -> rnn3
sent5 -> rnn3
}
para0 [label="paragraph info 0"]
para1 [label="paragraph info 1"]
para2 [label="paragraph info 2"]
rnn1 -> para0
rnn2 -> para1
rnn3 -> para2
para0 -> rnn
para1 -> rnn
para2 -> rnn
chapter [label="chapter info"]
rnn -> chapter
}
doc/design/ops/images/rnn.dot 0 → 100644
浏览文件 @ 1ffdecf1
digraph G {
label = "simple RNN implementation"
ranksep=2;
//graph [nodesep=1, ranksep=1];
node[nodesep=1]
subgraph cluster0 {
label = "global scope"
rankdir = TB
W
boot_memory
input
output
}
subgraph cluster1 {
label = "step-scope 0"
rankdir = TB
memory0[label="memory"]
prememory0[label="pre-memory"]
step_input0[label="step input"]
step_output0[label="step output"]
}
subgraph cluster2 {
label = "step-scope 1"
rankdir = TB
memory1[label="memory"]
prememory1[label="pre-memory"]
step_input1[label="step input"]
step_output1[label="step output"]
}
subgraph cluster3 {
label = "step-scope 2"
rankdir = TB
memory2[label="memory"]
prememory2[label="pre-memory"]
step_input2[label="step input"]
step_output2[label="step output"]
}
stepnet [shape=box]
stepnet0 [shape=box, style=dashed]
stepnet1 [shape=box, style=dashed]
stepnet2 [shape=box, style=dashed]
edge[color=blue]
boot_memory -> prememory0 [label="init" color="blue"]
memory0 -> prememory1 [label="copy/reference" color="blue"]
memory1 -> prememory2 [label="copy/reference" color="blue"]
edge[color=black]
W -> stepnet0[constraint=false, style=dashed]
W -> stepnet1[constraint=false, style=dashed]
W -> stepnet2[constraint=false, style=dashed]
memory0 -> stepnet0[style=dashed]
prememory0 -> stepnet0 -> step_output0[style=dashed]
memory1 -> stepnet1[style=dashed]
prememory1 -> stepnet1 -> step_output1[style=dashed]
memory2 -> stepnet2[style=dashed]
prememory2 -> stepnet2 -> step_output2[style=dashed]
input -> step_input0
input -> step_input1
input -> step_input2
step_input0 -> stepnet0 [style=dashed]
step_input1 -> stepnet1[style=dashed]
step_input2 -> stepnet2[style=dashed]
step_output0 -> output
step_output1 -> output
step_output2 -> output
stepnet0 -> stepnet[style=dashed]
stepnet1 -> stepnet[style=dashed]
stepnet2 -> stepnet[style=dashed]
}
doc/design/ops/images/rnn_2level_data.dot 0 → 100644
浏览文件 @ 1ffdecf1
digraph G {
chapter [label="chapter"]
subgraph cluster0 {
label = "paragraph 0"
top_rnn0[label="top rnn step 0" shape=box]
p0 [label="paragraph 0"]
p1 [label="paragraph 1"]
}
subgraph cluster1{
label = "paragraph 1"
top_rnn1[label="top rnn step 1" shape=box]
p2 [label="paragraph 0"]
p3 [label="paragraph 1"]
}
subgraph cluster_p0 {
label = "sentence 0"
low_rnn0 [label="low rnn step 0" shape=box]
s00 [label="sentence 0"]
s01 [label="sentence 1"]
low_rnn0 -> s00
low_rnn0 -> s01
}
subgraph cluster_p1 {
label = "sentence 1"
low_rnn1 [label="low rnn step 1" shape=box]
s10 [label="sentence 0"]
s11 [label="sentence 1"]
low_rnn1 -> s10
low_rnn1 -> s11
}
subgraph cluster_p2 {
label = "sentence 1"
low_rnn2 [label="low rnn step 0" shape=box]
s20 [label="sentence 0"]
s21 [label="sentence 1"]
low_rnn2 -> s20
low_rnn2 -> s21
}
subgraph cluster_p3 {
label = "sentence 1"
low_rnn3 [label="low rnn step 1" shape=box]
s30 [label="sentence 0"]
s31 [label="sentence 1"]
low_rnn3 -> s30
low_rnn3 -> s31
}
chapter -> top_rnn0
chapter -> top_rnn1
top_rnn0 -> p0
top_rnn0 -> p1
top_rnn1 -> p2
top_rnn1 -> p3
p0 -> low_rnn0
p1 -> low_rnn1
p2 -> low_rnn2
p3 -> low_rnn3
}
doc/design/ops/rnn.md 0 → 100644
浏览文件 @ 1ffdecf1
# RNNOp design
This document is about an RNN operator which requires that instances in a mini-batch have the same length. We will have a more flexible RNN operator.
## RNN Algorithm Implementation
<p aligh="center">
<img src="./images/rnn.jpg"/>
</p>
The above diagram shows an RNN unrolled into a full network.
There are several important concepts:
- *step-net*: the sub-graph to run at each step,
- *memory*, $h_t$, the state of the current step,
- *ex-memory*, $h_{t-1}$, the state of the previous step,
- *initial memory value*, the ex-memory of the first step.
### Step-scope
There could be local variables defined in step-nets. PaddlePaddle runtime realizes these variables in *step-scopes* -- scopes created for each step.
<p aligh="center">
<img src="./images/rnn.png"/><br/>
Figure 2 the RNN's data flow
</p>
Please be aware that all steps run the same step-net. Each step
1. creates the step-scope,
2. realizes local variables, including step-outputs, in the step-scope, and
3. runs the step-net, which could use these variables.
The RNN operator will compose its output from step outputs in step scopes.
### Memory and Ex-memory
Let's give more details about memory and ex-memory via a simply example:
$$
h_t = U h_{t-1} + W x_t
$$,
where $h_t$ and $h_{t-1}$ are the memory and ex-memory of step $t$'s respectively.
In the implementation, we can make an ex-memory variable either "refers to" the memory variable of the previous step,
or copy the value of the previous memory value to the current ex-memory variable.
### Usage in Python
For more information on Block, please refer to the [design doc](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/block.md).
We can define an RNN's step-net using Block:
```python
import paddle as pd
X = some_op() # x is some operator's output, and is a LoDTensor
a = some_op()
# declare parameters
W = pd.Variable(shape=[20, 30])
U = pd.Variable(shape=[20, 30])
rnn = pd.create_rnn_op(output_num=1)
with rnn.stepnet():
x = rnn.add_input(X)
# declare a memory (rnn's step)
h = rnn.add_memory(init=a)
# h.pre_state() means previous memory of rnn
new_state = pd.add_two( pd.matmul(W, x) + pd.matmul(U, h.pre_state()))
# update current memory
h.update(new_state)
# indicate that h variables in all step scopes should be merged
rnn.add_outputs(h)
out = rnn()
```
Python API functions in above example:
- `rnn.add_input` indicates the parameter is a variable that will be segmented into step-inputs.
- `rnn.add_memory` creates a variable used as the memory.
- `rnn.add_outputs` mark the variables that will be concatenated across steps into the RNN output.
### Nested RNN and LoDTensor
An RNN whose step-net includes other RNN operators is known as an *nested RNN*.
For example, we could have a 2-level RNN, where the top level corresponds to paragraphs, and the lower level corresponds to sentences.
The following figure illustrates the feeding of text into the lower level, one sentence each step, and the feeding of step outputs to the top level. The final top level output is about the whole text.
<p aligh="center">
<img src="./images/2_level_rnn.png"/>
</p>
```python
import paddle as pd
W = pd.Variable(shape=[20, 30])
U = pd.Variable(shape=[20, 30])
W0 = pd.Variable(shape=[20, 30])
U0 = pd.Variable(shape=[20, 30])
# a is output of some op
a = some_op()
# chapter_data is a set of 128-dim word vectors
# the first level of LoD is sentence
# the second level of LoD is chapter
chapter_data = pd.Variable(shape=[None, 128], type=pd.lod_tensor, level=2)
def lower_level_rnn(paragraph):
'''
x: the input
'''
rnn = pd.create_rnn_op(output_num=1)
with rnn.stepnet():
sentence = rnn.add_input(paragraph, level=0)
h = rnn.add_memory(shape=[20, 30])
h.update(
pd.matmul(W, sentence) + pd.matmul(U, h.pre_state()))
# get the last state as sentence's info
rnn.add_outputs(h)
return rnn
top_level_rnn = pd.create_rnn_op(output_num=1)
with top_level_rnn.stepnet():
paragraph_data = rnn.add_input(chapter_data, level=1)
low_rnn = lower_level_rnn(paragraph_data)
paragraph_out = low_rnn()
h = rnn.add_memory(init=a)
h.update(
pd.matmul(W0, paragraph_data) + pd.matmul(U0, h.pre_state()))
top_level_rnn.add_outputs(h)
# just output the last step
chapter_out = top_level_rnn(output_all_steps=False)
```
in above example, the construction of the `top_level_rnn` calls `lower_level_rnn`. The input is a LoD Tensor. The top level RNN segments input text data into paragraphs, and the lower level RNN segments each paragraph into sentences.
By default, the `RNNOp` will concatenate the outputs from all the time steps,
if the `output_all_steps` set to False, it will only output the final time step.
<p align="center">
<img src="images/rnn_2level_data.png"/>
</p>
doc/design/ops/sequence_decoder.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/optimizer.md 0 → 100644
浏览文件 @ 1ffdecf1
## Optimizer Design
### The Problem
A PaddlePaddle program, or a block, is a sequence of operators operating variables. A training program needs to do three kinds of works:
1. the forward pass, which computes intermediate results and the cost(s),
1. the backward pass, which derives gradients from intermediate results and costs, and
1. the optimization pass, which update model parameters to optimize the cost(s).
These works rely on three kinds of operators:
1. forward operators,
1. gradient operators, and
1. optimization operators.
It's true that users should be able to create all these operators manually by calling some low-level API, but it would be much more convenient if they could only describe the forward pass and let PaddlePaddle create the backward and optimization operators automatically.
In this design, we propose a high-level API that automatically derives the optimisation pass and operators from the forward pass.
### High-level Python API to describe the training process
1. User write code to describe the network:
```python
images = layer.data("images")
labels = layer.data("labels")
w1 = pd.var("w1")
b1 = pd.var("b1")
hidden = layer.fc(images, w=w1, b=b1)
cost = layer.mse(hidden, labels)
```
The above code snippet will create forward operators in [Block](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/block.md).
2. Users create a certain kind of Optimizer with some argument.
```python
optimizer = AdagradOptimizer(learing_rate=0.001)
```
3. Users use the optimizer to `minimize` a certain `cost` through updating parameters in parameter_list.
```python
opt_op_list = optimizer.minimize(cost, parameter_list=[w1, b1])
```
The above code snippet will create gradient and optimization operators in Block. The return value of `minimize()` is list of optimization operators that will be run by session.
4. Users use Session/Executor to run this opt_op_list as target to do training.
```python
sess.run(target= opt_op_list, ...)
```
#### Optimizer Python interface:
```python
class Optimizer(object):
"""Optimizer Base class.
"""
def __init__(self):
pass
def create_optimization_pass(self, parameters_and_grads):
"""Add optimization operators to update gradients to variables.
Args:
parameters_and_grads: a list of (variable, gradient) pair to update.
Returns:
optmization_op_list: a list of optimization operator that will update parameter using gradient.
"""
return None
def minimize(self, loss, parameter_list):
"""Add operations to minimize `loss` by updating `parameter_list`.
This method combines interface `append_backward_ops()` and
`create_optimization_pass()` into one.
"""
params_grads = self.create_backward_pass(loss, parameter_list)
update_ops = self.create_optimization_pass(params_grads)
return update_ops
```
Users can inherit the Optimizer above to create their own Optimizer with some special logic, such as AdagradOptimizer.
doc/design/parameter_average.md 0 → 100644
浏览文件 @ 1ffdecf1
# Averaging Parameter in PaddlePaddle
## Why Averaging
In a large scale machine learning setup where the size of the training data is huge, it could take us a large number of iterations over the training data before we can achieve the optimal values of parameters of our model. Looking at the problem setup, it is desirable if we can obtain the optimal values of parameters by going through the data in as few passes as we can.
Polyak and Juditsky (1992) showed that the test performance of simple average of parameters obtained by Stochastic Gradient Descent (SGD) is as good as that of parameter values that are obtained by training the model over and over again, over the training dataset.
Hence, to accelerate the speed of Stochastic Gradient Descent, Averaged Stochastic Gradient Descent (ASGD) was proposed in Polyak and Juditsky (1992). For ASGD, the running average of parameters obtained by SGD, is used as the estimator for <img src="./images/theta_star.gif"/><br/> . The averaging is done as follows:
<img src="./images/asgd.gif" align="center"/><br/>
We propose averaging for any optimizer similar to how ASGD performs it, as mentioned above.
### How to perform Parameter Averaging in PaddlePaddle
Parameter Averaging in PaddlePaddle works in the following way during training :
1. It will take in an instance of a normal optimizer as an input, e.g. RMSPropOptimizer
2. The optimizer itself is responsible for updating the parameters.
3. The ParameterAverageOptimizer maintains a separate copy of the parameters for itself:
1. In concept, the values of this copy are the average of the values of the parameters in the most recent N batches.
2. However, saving all the N instances of the parameters in memory is not feasible.
3. Therefore, an approximation algorithm is used.
Hence, overall we have have two copies of the parameters: one for the optimizer itself, and one for the ParameterAverageOptimizer. The former should be used in back propagation, while the latter should be used during testing and should be saved.
During the testing/ saving the model phase, we perform the following steps:
1. Perform the delayed operations.
2. Save current values of the parameters to a temporary variable.
3. Replace the values of the parameters with the averaged values.
4. Perform testing and/or save the parameters.
5. Restore the values of the parameters once done.
### How to implement Averaging of Parameter in PaddlePaddle
We can add the ParameterAverageOptimizer op to the graph through Python API. Using this approach, we manually add this op to the graph and direct the output of the optimizer op to this op during training.
**Advantages**:
- Allows for greater flexibility to the users of PaddlePaddle. Using this approach, the users can plug different optimizers into ParameterAverageOptimizer by passing in the optimizer to the op.
- Makes it easy for the users to customize and extend the framework.
**Disadvantages**:
- Implementation requires re-writing the averaging methodology in Python.
### Low-Level implementation
In the new design, we propose to create a new operation for averaging parameter updates (ParameterAverageOptimizer). For now, we can add an op that takes in the following as input:
- the optimizer
- the window_size to keep the updates
The ParameterAverageOptimizer op can be like any other operator with its own CPU/GPU implementation either using Eigen or separate CPU and GPU kernels. As the initial implementation, we can implement the kernel using Eigen following the abstraction pattern implemented for [Operators](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/rmsprop_op.h). We also want to support the case when the Trainer/Optimizer runs on the GPU while ParameterAverageOptimizer runs on a CPU.
The idea of building an op for averaging is in sync with the refactored PaddlePaddle philosophy of using operators to represent any computation unit. The way the op will be added to the computation graph will be decided by the [layer functions](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/python_api.md#layer-function) in Python API.
### Python API implementation for ParameterAverageOptimizer
Based on Polyak and Juditsky (1992), we can generalize the averaging of updates to any optimizer. The input to the op would be the following:
- Any optimizer (RMSProp , AdaGrad etc.)
- A window size. The op keeps accumulating updated parameter values over a window of N batches and takes an average. Move the averaged value to a buffer when window is full to avoid loss of precision.
Using the ParameterAverageOptimizer op, any user can add the operation to their computation graphs. However, this will require a lot of lines of code and we should design Python APIs that support averaging. As per the PaddlePaddle [Python API design](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/python_api.md), the layer functions are responsible for creating operators, operator parameters and variables. Since ParameterAverageOptimizer will be an operator, it makes sense to create it in the layer functions.
We will have a wrapper written in Python that will support the functionality and implement the actual core computation in C++ core as we have done for other [Optimizers](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/rmsprop_op.cc)
#### Creation of the ParameterAverageOptimizer operator
There are two ways for creating the ParameterAverageOptimizer op:
1. We create the op immediately while building the computation graph.
2. We add the op in a lazy manner, just before the backward pass, similar to the way the optimization ops are added.
The proposal is to add the op immediately while building the computation graph.
#### High-level API
In PaddlePaddle Python API, users will primarily rely on [layer functions](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/python_api.md#layer-function) to create neural network layers. Hence, we also need to provide parameter average functionality in layer functions.
doc/design/parameters_in_cpp.md
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/program.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/prune.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/python_api.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/reader/README.md
浏览文件 @ 1ffdecf1
......@@ -52,7 +52,7 @@ Here are valid outputs:
# a mini batch of three data items, each data item is a list (single column).
[([1,1,1],),
([2,2,2],),
([3,3,3],),
([3,3,3],)]
```
Please note that each item inside the list must be a tuple, below is an invalid output:
......
doc/design/refactor/distributed_architecture.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/parameter_server.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/session.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/dist-graph.graffle 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/dist-graph.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/distributed_architecture.graffle 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/distributed_architecture.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/local-graph.graffle 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/local-graph.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/local_architecture.graffle 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/local_architecture.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/paddle-compile.graffle 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactor/src/paddle-compile.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/refactorization.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/register_grad_op.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/regularization.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/releasing_process.md
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/scope.md
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/selected_rows.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/simple_op_design.md
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/tensor_array.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/test.dot 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/test.dot.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/design/var_desc.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/faq/build_and_install/index_cn.rst 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/faq/cluster/index_cn.rst 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/faq/index_cn.rst
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/faq/local/index_cn.rst 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/faq/model/index_cn.rst 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/faq/parameter/index_cn.rst 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/dev/build_cn.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/dev/build_en.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/dev/contribute_to_paddle_en.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/howto/dev/contribute_to_paddle_en.md 0 → 120000
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/dev/new_op_cn.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/dev/new_op_en.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/dev/use_eigen_cn.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/dev/use_eigen_en.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/dev/write_docs_cn.rst
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/index_cn.rst
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/index_en.rst
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/usage/cluster/src/trainer.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/usage/cluster/src/trainer_cn.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/usage/cluster/src/word2vec/api_train_v2.py 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/howto/usage/cluster/src/word2vec/prepare.py 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/index_cn.rst
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/index_en.rst
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/mobile/cross_compiling_for_android_en.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/mobile/cross_compiling_for_ios_cn.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/mobile/cross_compiling_for_raspberry_en.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/mobile/index_cn.rst 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/mobile/index_en.rst 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/survey/cluster_bootstrapping_tools.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
doc/tutorials/image_classification/cifar.png 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/image_classification/lenet.png 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/image_classification/plot.png 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/image_classification/src/plot.png 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/index_cn.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/index_en.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/rec/ml_dataset_cn.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/rec/ml_dataset_en.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/rec/ml_regression_cn.rst 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/rec/ml_regression_en.rst 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/rec/rec_regression_network.png 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/sentiment_analysis/index_cn.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/sentiment_analysis/index_en.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/sentiment_analysis/lstm.png 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/sentiment_analysis/src/lstm.png 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/text_generation/index_cn.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/tutorials/text_generation/index_en.md 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
doc/v1_api_tutorials/README.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
go/.gitignore
浏览文件 @ 1ffdecf1
此差异已折叠。
go/cmd/master/master.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/cmd/pserver/pserver.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/glide.lock
浏览文件 @ 1ffdecf1
此差异已折叠。
go/glide.yaml
浏览文件 @ 1ffdecf1
此差异已折叠。
go/master/c/client.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/master/client.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/master/client_internal_test.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/master/client_test.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/master/etcd_client.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/master/service.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/proto/.gitignore 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/client/c/cclient.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/client/client.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/client/client_test.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/client/etcd_client.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/etcd_client.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/optimizer.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/optimizer_test.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/service.go
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/service_internal_test.go 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
go/pserver/service_test.go
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/api/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/api/Util.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/capi/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/capi/Matrix.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/capi/gradient_machine.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/capi/gradient_machine.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/capi/matrix.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/capi/paddle_capi.map 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/capi/tests/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/capi/tests/test_Matrix.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/cuda/include/hl_cnn.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/cuda/include/hl_cpu_gru.cuh
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/cuda/include/hl_cuda_cudnn.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/cuda/include/hl_matrix.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/cuda/include/hl_tensor_ops.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/cuda/src/hl_cuda_cnn.cu
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/cuda/src/hl_cuda_cudnn.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/cuda/src/hl_cuda_matrix.cu
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/attribute.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/attribute.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/backward.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/backward.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/backward.md
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/backward_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/block_desc.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/block_desc.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/data_type.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/ddim.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/ddim.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/ddim_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/details/op_registry.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/dim.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/dim_test.cu
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/eigen.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/eigen_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/executor.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/executor.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/feed_fetch_method.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/feed_fetch_type.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/framework.proto
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/grad_op_builder.cc 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
paddle/framework/grad_op_builder_test.cc 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
paddle/framework/grad_op_desc_maker.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/lod_rank_table.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/lod_rank_table.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/lod_tensor.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/lod_tensor.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/lod_tensor.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/lod_tensor_array.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/lod_tensor_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/lod_tensor_test.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/op_desc.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/op_desc.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/op_info.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/op_info.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/op_proto_maker.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/op_proto_maker.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/op_proto_maker_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/op_registry.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/op_registry.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/operator.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/operator.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/operator_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/program_desc.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/program_desc.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/program_desc_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/proto_desc.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/prune.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/prune_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/pybind.cc 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
paddle/framework/scope.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/scope.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/scope_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/selected_rows.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/selected_rows.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/selected_rows_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/shape_inference.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/shape_inference.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/tensor.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/tensor_array.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/tensor_array.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/tensor_array_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/tensor_impl.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/tensor_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/type_defs.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/var_desc.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/var_desc.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/var_type.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/var_type_inference.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/var_type_inference_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/variable.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/framework/variable.md
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/BlockExpandOp.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/CosSimOp.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/CropOp.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/DepthwiseConvOp.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/EigenGemm.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/FunctionTest.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/GemmConvOp.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/GemmConvOpTest.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/GemmFunctor.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/GemmFunctor.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/GruFunctor.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/Im2Col.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/Im2ColTest.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/MulOp.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/PadOp.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/RowConvOp.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/ScaleSubRegionOp.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/ScaleSubRegionOp.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/ScaleSubRegionOpGpu.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/ScaleSubRegionOpTest.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/SwitchOp.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/SwitchOp.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/SwitchOpGpu.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/SwitchOpTest.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/neon/NeonDepthwiseConv.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/neon/NeonDepthwiseConv.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/neon/NeonDepthwiseConvTranspose.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/function/neon/neon_util.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/activations/MKLDNNActivation.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/activations/MKLDNNActivation.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/CRFLayer.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/Conv3DLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/Conv3DLayer.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/CostLayer.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/CostLayer.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/CrossEntropyOverBeam.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/CrossEntropyOverBeam.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/DeConv3DLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/DeConv3DLayer.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/ExpandConvBaseLayer.cpp 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
paddle/gserver/layers/Layer.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/Layer.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNAddtoLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNAddtoLayer.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNBase.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNBatchNormLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNBatchNormLayer.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNConvLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNConvLayer.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNLayer.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNPoolLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/MKLDNNPoolLayer.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/Pool3DLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/PoolLayer.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/ScaleShiftLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/ScaleSubRegionLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/ScaleSubRegionLayer.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/SequenceSliceLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/SwitchOrderLayer.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/layers/SwitchOrderLayer.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/tests/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/tests/MKLDNNTester.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/tests/mkldnn_branch_net.conf 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/tests/mkldnn_simple_net.conf 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/tests/test_CrossEntropyOverBeamGrad.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/tests/test_Expand.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/gserver/tests/test_SeqSliceLayerGrad.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/Allocator.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/BaseMatrix.cu
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/MKLDNNMatrix.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/MKLDNNMatrix.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/MathFunctions.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/MathFunctions.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/Matrix.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/Matrix.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/NEONFunctions.cpp 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/NEONFunctions.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/RowBuffer.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/SparseMatrix.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/Vector.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/Vector.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/tests/TensorCheck.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/tests/test_Matrix.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/math/tests/test_Tensor.cu
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/memory/.clang-format 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
paddle/memory/.clang-format 0 → 120000
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/memory/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/memory/detail/meta_cache.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/memory/memcpy.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/memory/memcpy.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/memory/memory.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/memory/memory.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/memory/memory_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/.clang-format 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
paddle/operators/.clang-format 0 → 120000
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/accuracy_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/accuracy_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/accuracy_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/activation_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/activation_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/activation_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adadelta_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adadelta_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adagrad_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adagrad_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adam_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adam_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adam_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adamax_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adamax_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/adamax_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/array_operator.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/array_to_lod_tensor_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/auc_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/auc_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/batch_norm_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/batch_norm_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/batch_norm_op.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/cast_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/cast_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/cast_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/chunk_eval_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/chunk_eval_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/clip_by_norm_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/clip_by_norm_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/clip_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/clip_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/clip_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/compare_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/compare_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/compare_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/concat_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/concat_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/concat_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/cond_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/cond_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv2d_transpose_cudnn_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv2d_transpose_cudnn_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_cudnn_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_shift_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_shift_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_shift_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_transpose_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_transpose_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/conv_transpose_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/cos_sim_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/cos_sim_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/crf_decoding_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/crf_decoding_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/crop_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/crop_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/crop_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/cross_entropy_op.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/decayed_adagrad_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/decayed_adagrad_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/decayed_adagrad_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/detail/strided_memcpy.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/dropout_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/dropout_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/dropout_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/dynamic_recurrent_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/dynamic_recurrent_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/dynamic_recurrent_op_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_add_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_add_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_add_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_div_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_div_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_div_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_mul_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_mul_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_mul_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_op_function.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_sub_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_sub_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/elementwise_sub_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/feed_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/fetch_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/fill_constant_batch_size_like_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/fill_constant_batch_size_like_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/fill_constant_batch_size_like_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/fill_constant_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gather.cu.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gather.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gather_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gather_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gather_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gru_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gru_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gru_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gru_unit_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gru_unit_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/gru_unit_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/huber_loss_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/huber_loss_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/huber_loss_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/images/batch_norm_fork.dot 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/images/batch_norm_fork.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/images/batch_norm_op_kernel.png 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/increment_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/l1_norm_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/l1_norm_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/linear_chain_crf_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/linear_chain_crf_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/linear_chain_crf_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/load_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lod_array_length_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lod_rank_table_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lod_tensor_to_array_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lookup_table_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lookup_table_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lookup_table_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lrn_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lrn_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lrn_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lstm_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lstm_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lstm_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lstm_unit_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lstm_unit_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/lstm_unit_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/margin_rank_loss_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/margin_rank_loss_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/margin_rank_loss_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/context_project.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/context_project.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/context_project.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/cross_entropy.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/cross_entropy.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/cross_entropy.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/detail/CMakeLists.txt 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/detail/activation_functions.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/detail/avx_functions.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/detail/gru_cpu_kernel.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/detail/gru_gpu_kernel.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/detail/gru_kernel.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/detail/lstm_cpu_kernel.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/detail/lstm_gpu_kernel.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/detail/lstm_kernel.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/gru_compute.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/gru_compute.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/gru_compute.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/im2col.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/im2col.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/im2col.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/im2col_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/lstm_compute.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/lstm_compute.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/lstm_compute.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/math_function_test.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/matmul.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/pooling.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/pooling.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/pooling.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/selected_rows_functor.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/selected_rows_functor.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/selected_rows_functor.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/selected_rows_functor_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/selected_rows_functor_test.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/sequence2batch.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/sequence2batch.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/sequence2batch.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/sequence_pooling.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/sequence_pooling.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/sequence_pooling.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/softmax.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/softmax.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/softmax.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/vol2col.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/vol2col.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/vol2col.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/math/vol2col_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/matmul_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/matmul_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/matmul_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/mean_op.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/mean_op.cu
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/mean_op.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/minus_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/minus_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/minus_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/modified_huber_loss_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/modified_huber_loss_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/modified_huber_loss_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/momentum_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/momentum_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/momentum_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/mul_op.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/mul_op.cu
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/mul_op.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/multiplex_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/multiplex_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/multiplex_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/name_convention.md 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/nccl/CMakeLists.txt 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/nccl/nccl_gpu_common.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/nccl/nccl_gpu_common.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/nccl_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/nccl_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/nccl_op_test.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/net_op.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/net_op.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/net_op_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pad_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pad_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pad_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pool_cudnn_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pool_cudnn_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pool_cudnn_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pool_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pool_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pool_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pool_with_index_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pool_with_index_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/pool_with_index_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/positive_negative_pair_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/positive_negative_pair_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/precision_recall_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/precision_recall_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/prelu_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/prelu_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/prelu_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/proximal_adagrad_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/proximal_adagrad_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/proximal_adagrad_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/proximal_gd_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/proximal_gd_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/proximal_gd_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/rank_loss_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/rank_loss_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/rank_loss_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/recurrent_op.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/recurrent_op.h 已删除 100644 → 0
浏览文件 @ 3cf01b5d
此差异已折叠。
paddle/operators/reduce_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/reduce_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/reduce_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/reshape_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/reshape_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/reshape_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/rmsprop_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/rmsprop_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/rmsprop_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/rnn_memory_helper_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/save_load_op_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/save_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/scale_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/scale_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/scale_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/scatter.cu.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/scatter.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/scatter_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/scatter_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/scatter_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/scatter_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/seq_expand_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/seq_expand_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/seq_expand_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_concat_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_concat_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_concat_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_conv_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_conv_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_conv_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_pool_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_pool_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_pool_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_softmax_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_softmax_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sequence_softmax_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sgd_op.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sgd_op.cu
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sgd_op.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/shrink_rnn_memory_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sigmoid_cross_entropy_with_logits_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sign_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sign_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/smooth_l1_loss_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/smooth_l1_loss_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/smooth_l1_loss_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/softmax_op.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/softmax_op.cu
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/softmax_op.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/softmax_with_cross_entropy_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/softmax_with_cross_entropy_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/softmax_with_cross_entropy_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/split_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/split_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/split_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/squared_l2_distance_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/squared_l2_distance_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/squared_l2_distance_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/squared_l2_norm_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/squared_l2_norm_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/squared_l2_norm_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/strided_memcpy.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/strided_memcpy_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sum_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sum_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/sum_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/tensor_array_read_write_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/top_k_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/top_k_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/top_k_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/transpose_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/transpose_op.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/transpose_op.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/operators/while_op.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/optimizer/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/optimizer/adam_optimizer.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/optimizer/adam_optimizer.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/optimizer/lr_policy.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/optimizer/optimizer.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/optimizer/optimizer.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/optimizer/sgd_optimizer.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/optimizer/sgd_optimizer.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/optimizer/tensor.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/parameter/Argument.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/parameter/Argument.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/parameter/Parameter.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/parameter/Parameter.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/cuda_helper.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/cudnn_helper.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/cudnn_helper_test.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/details/device_ptr_cast.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/device_context.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/device_context.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/dynload/cublas.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/dynload/cudnn.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/dynload/nccl.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/dynload/nccl.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/enforce.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/enforce_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/gpu_info.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/gpu_info.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/hostdevice.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/macros.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/nccl_test.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/place.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/place.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/transform.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/transform_test.cu 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/platform/variant.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pserver/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pserver/LightNetwork.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pserver/ParameterClient2.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pserver/ParameterClient2.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pserver/ParameterServer2.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pserver/test/SocketTest.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pybind/.clang-format 0 → 120000
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pybind/CMakeLists.txt 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pybind/exception.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pybind/exception.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pybind/print_operators_doc.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pybind/protobuf.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pybind/protobuf.h 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/pybind/pybind.cc 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/scripts/cluster_train_v2/fabric/conf.py 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/scripts/cluster_train_v2/fabric/run.sh 0 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/scripts/docker/build.sh
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/scripts/submit_local.sh.in
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/scripts/travis/build_doc.sh
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/scripts/travis/build_ios.sh 0 → 100755
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/string/.clang-format 0 → 120000
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/string/piece.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/string/printf_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/string/to_string_test.cc
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/trainer/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/trainer/MergeModel.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/utils/Flags.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/utils/Flags.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/utils/Util.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/utils/Util.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/utils/Version.h
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/utils/arch/linux/Locks.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
paddle/utils/arch/osx/Excepts.cpp
浏览文件 @ 1ffdecf1
此差异已折叠。
proto/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
proto/ModelConfig.proto
浏览文件 @ 1ffdecf1
此差异已折叠。
proto/TrainerConfig.proto
浏览文件 @ 1ffdecf1
此差异已折叠。
python/CMakeLists.txt
浏览文件 @ 1ffdecf1
此差异已折叠。
python/paddle/trainer/recurrent_units.py 100755 → 100644
浏览文件 @ 1ffdecf1
文件模式从 100755 更改为 100644
python/paddle/trainer_config_helpers/layers.py 100755 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
python/paddle/trainer_config_helpers/networks.py 100755 → 100644
浏览文件 @ 1ffdecf1
此差异已折叠。
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或