如何在Windows 10上为Python 3.5安装Theano [英] How to install Theano for Python 3.5 on Windows 10

问题描述

我找不到应该如何仅为Python 2.7安装Theano for Python 3.5(Windows 10).

I can't find how one is supposed to install Theano for Python 3.5 (Windows 10) only for Python 2.7.

我不想降级到2.7，那么有人知道如何为Python 3.5安装Theano(在Windows 10上)吗?

I don't want to have to downgrade to 2.7 so does anyone know how to install Theano for Python 3.5 (on Windows 10)?

我们将非常感谢您提供清晰的分步说明集，以及我需要的其他内容以及如何安装它们.

A clear set of step-by-step instructions would be greatly appreciated, as well as other things I need and how to install them.

推荐答案

要使这个问题更加独立，我觉得值得添加以下答案:

To make this question more self contained I find it worth to add this answer:

Winpython 最好！，实际上是在Windows上推荐的安装，在此处阅读.它随python 3.5一起加载，其中包含了所有依赖项，包括numpy，scipy，scikit，theano，keras，ipython，jupyter等，实际上，您不需要执行任何操作或更改任何配置，只需要做(您要启用GPU的对象)是下载 CUDA (当然，您应该使用Visual C ++)最后将此变量添加到您的环境变量中:

Winpython is the best!!, actually it is the recommended installation over windows, read here. It comes with python 3.5 fully loaded with all the dependencies including numpy, scipy, scikit, theano, keras, ipython, jupyter and more, actually you don't need to do anything or to change any configurations, all you need to do (if you want to enable GPU) is to download CUDA (and of course you should have visual C++) and finally add this variable to your environment variables:

变量名称:THEANO_FLAGS

Variable name: THEANO_FLAGS

变量值:floatX = float32，device = gpu，nvcc.fastmath = True，lib.cnmem = 0.8

Variable value: floatX=float32,device=gpu,nvcc.fastmath=True,lib.cnmem=0.8

(lib.cnmem = 0.8)可以等于任何值，这意味着cuda将使用您80％的GPU内存

(lib.cnmem=0.8) can be equal to any value, this means that cuda will use 80% of your GPU memory

如果您需要其他说明，请在评论中告诉我

If you need any other clarification let me know in the comments

这篇关于如何在Windows 10上为Python 3.5安装Theano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！