Discriminative Embeddings of Latent Variable Models for Structured Data 代码实现

DiscriminativeEmbeddings of Latent Variable Models for Structured Data
代码实现

注意g++
要5.4版本

cd/usr/bin
sudo mvgcc gcc.bak  

sudo mvg++ g++.bak
sudo ln-s gcc-5 gcc  

sudo ln-s g++-5 g++
查看gcc版本指令

gcc--version

网址链接：https://github.com/Hanjun-Dai/pytorch_structure2vec

1        RDkit源码安装

sudo apt-get installbuild-essential python-numpy cmake python-dev sqlite3 libsqlite3-devlibboost-dev libboost-system-dev libboost-thread-dev libboost-serialization-devlibboost-python-dev libboost-regex-dev
 

配置系统环境变量

sudogedit ~/.bashrc

exportRDBASE=/home/gu/rdkit       

exportPYTHONPATH=$RDBASE   

exportLD_LIBRARY_PATH=$RDBASE/lib:/home/gu/rdkit/lib

source~/.bashrc

 

wgethttps://github.com/rdkit/rdkit/archive/Release_2017_09_1.tar.gz
tarzxvf Release_2017_09_1.tar.gz

mvrdkit-Release_2017_09_1 rdkit

cd/home/gu/rdkit

mkdirbuild

cdbuild

cmake..

cmake -DBOOST_ROOT=/home/gu/rdkit-DRDK_BUILD_PYTHON_WRAPPERS=OFF -DRDK_BUILD_INCHI_SUPPORT=ON-DRDK_BUILD_AVALON_SUPPORT=ON ..
make-j4

makeinstall

ctest

 

2        安装pytorch

按照官网链接选择对应版本安装http://pytorch.org/

pip install http://download.pytorch.org/whl/cu80/torch-0.3.0.post4-cp27-cp27mu-linux_x86_64.whl
pipinstall torchvision

3        Set UP

cds2v_lib

make-j4

 

exportRDBASE=/home/gu/rdkit

 

cdharvard_cep

make-j4

 

4        Prepare data

cdharvard_cep

pythonmol_lib.py

 

5        Model dump

执行主程序

pythonmain.py -gm mean_field -saved_model saved/mean_field.model -phase test -modecpu

 

6        可能遇到的问题

ELL解决：libstdc++.so.6: version`CXXABI_1.3.9' not found (required by bin/opencv_test_core)?

#删除原来的libstdc++.so.6

sudorm -rf anaconda2/lib/libstdc++.so.6

#拷贝新的动态库文件,注意自己的是哪个版本

sudocp /usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.21 /home/haotao/anaconda2/lib

建立软连接

cdanaconda2/lib/

sudochmod +r libstdc++.so.6.0.21

sudoln -sf libstdc++.so.6.0.21 libstdc++.so.6

sudoldconfig