使用每晚的TensorFlow构建进行Cloud ML Engine培训 [英] Using a nightly TensorFlow build for training with Cloud ML Engine
问题描述
如果我需要在Cloud ML Engine培训工作中使用每晚的TensorFlow构建,该怎么办?
If I need to use a nightly TensorFlow build in a Cloud ML Engine training job, how do I do it?
推荐答案
-
从 https://github.com/tensorflow/tensorflow#installation . 如何选择合适的版本:
Download a nightly build from https://github.com/tensorflow/tensorflow#installation. How to pick the right build:
- 根据是否需要使用GPU进行培训,使用仅Linux CPU"或"Linux GPU"
- 使用Python 2构建.
重命名.whl文件,例如
Rename the .whl file, for example
mv tensorflow-1.0.1-cp27-cp27mu-linux_x86_64.whl \ tensorflow-1.0.1-cp27-none-linux_x86_64.whl
(这里我们将
cpu27mu
重命名为none
.Pip解析该部分以检测.whl软件包是否适合平台,但是该特定名称在某些较早版本的pip上不起作用)(here we renamed the
cpu27mu
tonone
. Pip parses that part to detect whether a .whl package is suitable for a platform, but that particular name doesn't work on some older versions of pip)将.whl文件上传到GCS并将其指定为
package_uris
之一 提交Cloud ML Engine培训作业时.Upload the .whl file to GCS and specify it as one of the
package_uris
when submitting a Cloud ML Engine training job.请注意,除了使用夜间构建外,您还可以按照 https://中所述从源代码构建TensorFlow. www.tensorflow.org/install/install_sources .
Note that instead of using a nightly build you can also build TensorFlow from source as described in https://www.tensorflow.org/install/install_sources.
这篇关于使用每晚的TensorFlow构建进行Cloud ML Engine培训的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!