Python Depth Learning Library Theano installation of vomiting blood experience __python

Theano is a Python library that can run fast numerical computations on a CPU or GPU. This is a key base in python depth learning, and you can use it directly to create a deep learning model or a wrapper library that simplifies the program greatly.

Pip Install Theano

It's an error. Theano relies on NumPy and scipy,numpy have been installed, now to install scipy.

SciPy is an open source Python algorithm library and math Toolkit. SCIPY contains modules such as optimization, linear algebra, integrals, interpolation, special functions, fast Fourier transforms, signal processing and image processing, ordinary differential equation solving, and other commonly used calculations in science and engineering. Software with similar functions is MATLAB, GNU Octave and Scilab.

https://www.lfd.uci.edu/~gohlke/pythonlibs/#scipy still download from here

Pip Install SCIPY-1.0.0-CP36-CP36M-WIN32.WHL

Run pip install again Theano

The installation is complete.

Pip Install nose

It is said that nose is a relatively cool unit test framework, the following figure shows the installation completed.

Thought it would be a quick wave. It's an error.

Let's install the g++. Https://sourceforge.net/projects/mingw-w64/files/latest/download

Run the code after the installation is complete:

Import Theano
theano.test ()

It's still an error. It seems that Libpython is not installed, but can not find the libpython3.6 version of.

or give up their own toss it. Direct use of Anaconda.

Anaconda is a free value-added open-source release of the Python language for large-scale data processing, predictive analysis, and scientific computing, and is dedicated to simplifying package management and deployment. Anaconda uses package management System Conda for package management.

After the installation is complete, most of the external libraries are basically included. Of course, there is no Theano, so still open the Anaconda command line.

All set up, test it. or an error.

WARNING (Theano.tensor.blas): Using NumPy C-api based implementation for Blas.
Theano version 1.0.1theano is installed in C:\Users\guanll\AppData\Local\Continuum\anaconda3\lib\site-packages\ Theanonumpy version 1.14.0NumPy relaxed strides checking option:true
numpy is installed in C:\Users\guanll\AppData\L Ocal\continuum\anaconda3\lib\site-packages\numpypython version 3.6.4 | Anaconda, inc.| (default, 16 2018, 10:22:32) [MSC v.1900-bit (AMD64)]
Nose version 1.3.7 ... Omit a long error.

Need to Blas, need Libpython, but no 3.6 version of Libpython. No solution yet.

Glanced at the Theano installation proposal. A mouthful of old blood sprayed the screen. Version 3.6 of Python is not supported. It is estimated that there is no 3.6 version of Libpython ...

Blas is an application interface standard that is used to standardize the publishing of numerical libraries for basic linear algebra operations. The assembly was originally published in 1979 and is used to create a larger numeric package. In the field of High-performance computing, Blas is widely used. For example, the performance of Linpack is largely determined by the Blas neutron program DGEMM.

It is said that the third party NumPy library, if not automatically installed through the PIP, but to download their own numpy words will bring blas,atlas and so on. Go and download it.

Well, the download completes the installation successfully. But the error of Theano.test () is just not the same.

Since you do not support the 3.6 version, it will not toss it. Http://www.aibbt.com/a/22258.html