开发者社区> 问答> 正文

caffe有那些最佳实践


阿里云HPC服务器在交付时已经提供了一份编译好的Caffe源码(2016/5/8克隆于master分支),位于/disk1/deeplearning/caffe/,用户无需做任何额外工作即可直接运行。
(1) 命令行方式

  1. [backcolor=transparent]$ cd [backcolor=transparent]/[backcolor=transparent]disk1[backcolor=transparent]/[backcolor=transparent]deeplearning[backcolor=transparent]/[backcolor=transparent]caffe[backcolor=transparent]/
  2. [backcolor=transparent]$ [backcolor=transparent]./[backcolor=transparent]examples[backcolor=transparent]/[backcolor=transparent]mnist[backcolor=transparent]/[backcolor=transparent]train_lenet[backcolor=transparent].[backcolor=transparent]sh

(2)python方式
  1. [backcolor=transparent]$ cd [backcolor=transparent]/[backcolor=transparent]disk1[backcolor=transparent]/[backcolor=transparent]deeplearning[backcolor=transparent]/[backcolor=transparent]caffe[backcolor=transparent]/[backcolor=transparent]python
  2. [backcolor=transparent]$ [backcolor=transparent]/[backcolor=transparent]disk1[backcolor=transparent]/[backcolor=transparent]deeplearning[backcolor=transparent]/[backcolor=transparent]anaconda2[backcolor=transparent]/[backcolor=transparent]bin[backcolor=transparent]/[backcolor=transparent]python classify[backcolor=transparent].[backcolor=transparent]py
  3. [backcolor=transparent]$ [backcolor=transparent]/[backcolor=transparent]disk1[backcolor=transparent]/[backcolor=transparent]deeplearning[backcolor=transparent]/[backcolor=transparent]anaconda2[backcolor=transparent]/[backcolor=transparent]bin[backcolor=transparent]/[backcolor=transparent]python detect[backcolor=transparent].[backcolor=transparent]py

所有Caffe依赖包安装位置为/disk1/deeplearning/local_install ,其下README.txt描述了依赖包的详细信息:
  1. [backcolor=transparent]This[backcolor=transparent] is [backcolor=transparent]Caffe[backcolor=transparent] [backcolor=transparent]Dependency[backcolor=transparent].
  2. [backcolor=transparent]Include[backcolor=transparent] the binaries of [backcolor=transparent]:
  3. [backcolor=transparent]boost_1_56_0[backcolor=transparent].[backcolor=transparent]tar[backcolor=transparent].[backcolor=transparent]bz2
  4. [backcolor=transparent]cudnn[backcolor=transparent]-[backcolor=transparent]7.0[backcolor=transparent]-[backcolor=transparent]linux[backcolor=transparent]-[backcolor=transparent]x64[backcolor=transparent]-[backcolor=transparent]v3[backcolor=transparent].[backcolor=transparent]0[backcolor=transparent]-[backcolor=transparent]rc[backcolor=transparent].[backcolor=transparent]tgz
  5. [backcolor=transparent]gflags[backcolor=transparent]-[backcolor=transparent]2.1[backcolor=transparent].[backcolor=transparent]1.zip
  6. [backcolor=transparent]glog[backcolor=transparent]-[backcolor=transparent]0.3[backcolor=transparent].[backcolor=transparent]3.tar[backcolor=transparent].[backcolor=transparent]gz
  7. [backcolor=transparent]hdf5[backcolor=transparent]-[backcolor=transparent]1.8[backcolor=transparent].[backcolor=transparent]9.tar[backcolor=transparent].[backcolor=transparent]gz
  8. [backcolor=transparent]leveldb[backcolor=transparent]-[backcolor=transparent]1.7[backcolor=transparent].[backcolor=transparent]0.tar[backcolor=transparent].[backcolor=transparent]gz
  9. [backcolor=transparent]opencv[backcolor=transparent]-[backcolor=transparent]3.0[backcolor=transparent].[backcolor=transparent]0.zip[backcolor=transparent]    [backcolor=transparent].
  10. [backcolor=transparent]protobuf[backcolor=transparent]-[backcolor=transparent]2.5[backcolor=transparent].[backcolor=transparent]0.tar[backcolor=transparent].[backcolor=transparent]gz
  11. [backcolor=transparent]snappy[backcolor=transparent]-[backcolor=transparent]1.1[backcolor=transparent].[backcolor=transparent]1.tar[backcolor=transparent].[backcolor=transparent]gz
  12. [backcolor=transparent]lmdb[backcolor=transparent].[backcolor=transparent]zip
  13. [backcolor=transparent]GCC version[backcolor=transparent]:[backcolor=transparent] gcc version [backcolor=transparent]4.8[backcolor=transparent].[backcolor=transparent]3[backcolor=transparent] [backcolor=transparent]20140911[backcolor=transparent] [backcolor=transparent]([backcolor=transparent]Red[backcolor=transparent] [backcolor=transparent]Hat[backcolor=transparent] [backcolor=transparent]4.8[backcolor=transparent].[backcolor=transparent]3[backcolor=transparent]-[backcolor=transparent]9[backcolor=transparent])[backcolor=transparent] [backcolor=transparent]([backcolor=transparent]GCC[backcolor=transparent])
  14. [backcolor=transparent]OS [backcolor=transparent]:[backcolor=transparent] [backcolor=transparent]CentOS[backcolor=transparent] [backcolor=transparent]Linux[backcolor=transparent] release [backcolor=transparent]7.1[backcolor=transparent].[backcolor=transparent]1503[backcolor=transparent] [backcolor=transparent]([backcolor=transparent]Core[backcolor=transparent])

展开
收起
boxti 2017-10-20 10:55:52 16932 0
0 条回答
写回答
取消 提交回答
问答排行榜
最热
最新

相关电子书

更多
深度学习框架实战-Tensorflow 立即下载
低代码开发师(初级)实战教程 立即下载
阿里巴巴DevOps 最佳实践手册 立即下载