简体   繁体   English

如何在arm64,Ubuntu 16.04上安装Keras?

[英]How do I install Keras on arm64, Ubuntu 16.04?

I tried pip install --user keras==1.0.6 but it fails to build scipy. 我尝试了pip install --user keras==1.0.6但无法构建scipy。 It returns: 它返回:

numpy.distutils.system_info.NotFoundError: no lapack/blas resources found

and

failed with error code 1 in /tmp/pip-build-XjG1HN/scipy/ . failed with error code 1 in /tmp/pip-build-XjG1HN/scipy/

I also tried installing scipy alone, using these instructions : 我还尝试按照以下说明单独安装scipy:

export PATH="$PATH:/home/your_user/.local/bin"
pip install --user scipy

But received the same errors. 但是收到了同样的错误。

Note: While posting this I realized I didn't replace your_user with $USER . 注意:发布此内容时,我意识到我没有将$USER替换为your_user However, I found a different solution, see answer below. 但是,我找到了不同的解决方案,请参见下面的答案。

sudo apt-get install python-scipy

This successfully installed scipy on an arm64 Ubuntu 16.04 (on NVIDIA's Jetson TX1 with Jetpack 2.3). 这样可以在arm64 Ubuntu 16.04上成功安装scipy(在带有Jetpack 2.3的NVIDIA Jetson TX1上)。

The additional packages include lapack and blas packages, which the error alluded to. 其他软件包包括lapack和blas软件包,这些错误都暗示了该错误。

The following additional packages will be installed: lib blas -common lib blas 3 libgfortran3 lib lapack 3 python-decorator python-imaging python-numpy python-pil 将安装以下附加软件包:lib blas -common lib blas 3 libgfortran3 lib lapack 3 python-decorator python-imaging python-numpy python-pil

And then to install keras 1.0.6: 然后安装keras 1.0.6:

pip install --user keras==1.0.6


I found this answer by googling "install scipy "arm64"" and finding Xenial (16.04) : python-scipy package : Ubuntu - Launchpad . 我通过搜索“ install scipy“ arm64””并找到Xenial(16.04):python-scipy软件包:Ubuntu-Launchpad找到了这个答案。

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM