英特尔的Caffe优化是否支持这些网络？

你好，Sindhu。没有担心。
对于英特尔Caffe，您选择的网络无关紧要。
英特尔Caffe本身就是针对英特尔硬件的优化，因此它支持AlexNet和其他网络。我对您尝试的内容有几个疑问： 
你是如何定义网络的？
你在哪里寻找状态输出？
你有没有另一个版本的Caffe独立于DevCloud工作？问候，凯文（英特尔）



以下为原文

Hello, Sindhu.

No worries. For Intel Caffe, the network you choose shouldn't matter. Intel Caffe itself is optimization for Intel hardware, and so it supports AlexNet and other networks.

I have a couple of questions about what you are attempting:

• How are you defining your network?
  
• Where are you looking for the status output?
  
• Have you gotten another version of Caffe working independently of the DevCloud?
  

Regards,
Kevin (Intel)

请看答案
您是如何定义网络的？我们在原型文件中定义我们的网络（通常命名为train_val.prototxt）。
解算器参数在同一目录中的单独solver.prototxt中定义。
你在哪里寻找状态输出？在命令行本身。
但没有地位。
你有没有另一个版本的Caffe独立于DevCloud工作？
是。



以下为原文

Please  see the answers
 

• How are you defining your network?
  

We are defining our network in a prototxt file (normally named as train_val.prototxt). The solver parameters are defined in a separate solver.prototxt in the same directory.

• Where are you looking for the status output?
  

In the commandline itself. But there is no status.

• Have you gotten another version of Caffe working independently of the DevCloud?
  

       Yes.

1）请创建一个虚拟环境.2）按如下方式安装intel caffe：conda install -c intel caffe尝试运行代码，如果仍然遇到问题，请分享您正在遵循的步骤。



以下为原文

1) Please create a virtual environment.
2) Install intel caffe as follows:
   conda install -c intel caffe

Try running your code and if you still find an issue, please share the steps which you are following.

请看结果....
创建了一个虚拟环境并安装了所述软件包。
[uXXXXX @ c002~] $ source activate test
（测试）[uXXXXX @ c002项目] $
（测试）[uXXXXX @ c002 project] $ caffe~ / .conda / envs / test / bin / caffe
（测试）[uXXXXX @ c002项目] $ l***vlc_alexnet.caffemodel deploy.prototxt info.json input_lmdb labels.txt mean.binaryproto solver.prototxt train_val.prototxt
（测试）[uXXXXX @ c002项目] $ caffe train --solver = / home / uXXXXX / project / solver.prototxt |
qsub5196.c002
（测试）[uXXXXX @ c002项目] $ qstat
（测试）[uXXXXX @ c002项目] $ ls
bvlc_alexnet.caffemodel deploy.prototxt info.json input_lmdb labels.txt mean.binaryproto solver.prototxt STDIN.e5196 STDIN.o5196 train_val.prototxt
（测试）[uXXXXX @ c002项目] $ caffe train --solver = / home / uXXXXX / project / solver.prototxt浮点异常
（测试）[uXXXXX @ c002项目] $ pythonPython 3.6.3 |英特尔公司|
（默认情况下，2018年5月4日，04：22：28）[GCC 4.8.2 20140120（红帽4.8.2-15）]关于linuxType“帮助”，“版权”，“信用”或“许可证”的更多信息。
Intel（R）Python发行版由英特尔公司提供给您。请查看：https：//software.intel.com/en-us/python-distribution >>>导入caffeFloating点异常



以下为原文

Please see the results....
 
[size=13.3333px]Created a virtual environment and installed the said packages.
 
[size=13.3333px][uXXXXX@c002 ~]$ source activate test
[size=13.3333px] 
[size=13.3333px](test) [uXXXXX@c002 project]$
[size=13.3333px] 
[size=13.3333px](test) [uXXXXX@c002 project]$ which caffe
~/.conda/envs/test/bin/caffe
[size=13.3333px] 
[size=13.3333px] 
[size=13.3333px](test) [uXXXXX@c002 project]$ ls
bvlc_alexnet.caffemodel  deploy.prototxt  info.json  input_lmdb  labels.txt  mean.binaryproto  solver.prototxt  train_val.prototxt
[size=13.3333px] 
[size=13.3333px] 
[size=13.3333px](test) [uXXXXX@c002 project]$ caffe train --solver=/home/uXXXXX/project/solver.prototxt | qsub
5196.c002
[size=13.3333px] 
[size=13.3333px] 
[size=13.3333px](test) [uXXXXX@c002 project]$ qstat
[size=13.3333px] 
[size=13.3333px] 
[size=13.3333px](test) [uXXXXX@c002 project]$ ls
[size=13.3333px]bvlc_alexnet.caffemodel  deploy.prototxt  info.json  input_lmdb  labels.txt  mean.binaryproto  solver.prototxt  STDIN.e5196  STDIN.o5196  train_val.prototxt
[size=13.3333px] 
[size=13.3333px] 
[size=13.3333px](test) [uXXXXX@c002 project]$ caffe train --solver=/home/uXXXXX/project/solver.prototxt
Floating point exception
[size=13.3333px] 
[size=13.3333px] 
[size=13.3333px](test) [uXXXXX@c002 project]$ python
Python 3.6.3 |Intel Corporation| (default, May  4 2018, 04:22:28)
[GCC 4.8.2 20140120 (Red Hat 4.8.2-15)] on linux
Type "help", "copyright", "credits" or "license" for more information.
Intel(R) Distribution for Python is brought to you by Intel Corporation.
Please check out: https://software.intel.com/en-us/python-distribution
>>> import caffe
Floating point exception
 

嗨Sindhu，英特尔caffe可以在devcloud中使用。您可以使用命令'which caffe'检查caffe路径。
路径为'/glob/intel-python/python2/bin/caffe'.Kindly看到附带的截图。谢谢Aswathy     
caffe.PNG 
16.1 K.



以下为原文

Hi Sindhu,

Intel caffe is available in devcloud.
You can check the caffe path using the command 'which caffe'. The path will be '/glob/intel-python/python2/bin/caffe'.
Kindly see the screenshot attached.

Thanks
Aswathy

 

•   caffe.PNG 16.1 K
  
  

嗨Sindhu，
正如我们没有收到你的消息，假设问题已经解决。如果你想工作onintel optimize tensorflow，你可以按照以下步骤安装它：1。创建一个虚拟环境conda create -n -c intel python
= 3.6 pip numpy2.Activate环境源激活3.安装Intel优化的tensorflow（for tensorflow 1.6）：pip installhttps：//anaconda.org/intel/tensorflow/1.6.0/download/tensorflow-1.6.0-cp36-cp36m
-linux_x86_64.whl4.在terminal5.import tensorflow中键入'python'有关详细信息，请参阅：https：//software.intel.com/en-us/articles/intel-optimization-for-tensorflow-installation-guide?page=
1请确认这是否有帮助。
谢谢，Aswathy



以下为原文

Hi Sindhu,
As we haven't heard from you, assume that the issue is resolved.

If you would like to work on intel optimized tensorflow, you can follow the below steps to install it:
1.Create a virtual environment
   conda create -n -c intel python=3.6 pip numpy
2.Activate the environment
   source activate
3.Install Intel optimized tensorflow(for tensorflow 1.6):
   pip install https://anaconda.org/intel/tensorflow/1.6.0/download/tensorflow-1.6.0-cp36-cp36m-linux_x86_64.whl
4.Type 'python' in terminal
5.import tensorflow

For further details, kindly refer: https://software.intel.com/en-us/articles/intel-optimization-for-tensorflow-installation-guide?page=1

Please confirm if this helps.
Thanks,
Aswathy

嗨Sindhu，我们希望所提供的解决方案对您有所帮助。
请确认。谢谢。



以下为原文

Hi Sindhu,

We hope that the solution provided was helpful to you. Kindly confirm.

Thanks.