Tensorflow Syntaxnet模型安装测试方法

#install dev env
apt-get update
apt-get install build-essential vim -y


#install jdk8
apt-get install software-properties-common -y
#need to confirm
add-apt-repository ppa:webupd8team/java
apt-get update
#need to confirm
apt-get install oracle-java8-installer -y


#install bazel
echo "deb http://storage.googleapis.com/bazel-apt stable jdk1.8" | tee /etc/apt/sources.list.d/bazel.list
apt-get install curl
curl https://storage.googleapis.com/bazel-apt/doc/apt-key.pub.gpg | apt-key add -
apt-get update && apt-get install bazel -y
apt-get upgrade bazel -y


#install swig
apt-get install swig -y


#install pip
apt-get install python-pip python-devel python-numpy -y


#install protobuf, asciitree, numpy
pip freeze | grep protobuf
pip install -U protobuf==3.0.0b2
pip install asciitree


#install SyntaxNet
apt-get install git -y
git clone --recursive https://github.com/tensorflow/models.git
cd models/syntaxnet/tensorflow


apt-get install libcurl3-dev -y
# we should check whether python path is right
./configure
cd ..
#need big memory(>=16G)
bazel test syntaxnet/... util/utf8/...
#test parse tree

echo 'Bob brought the pizza to Alice.' | syntaxnet/demo.sh

 

参考1和2介绍了在syntaxnet中用定制universal dependencies英文语料重新训练POS-tagger和parser的方法，有兴趣的继续下载定制语料进一步深入测试。

 

参考：

1. https://github.com/tensorflow/models/tree/master/syntaxnet

2. https://github.com/dsindex/syntaxnet  定制韩语语料的测试，方法可借鉴。