投稿
  1. 极客之音首页
  2. 后端笔记

安装mingw+Theano+keras（多警告、错误解决办法）

seven_ • • 后端笔记 • 阅读 368

导读：本篇文章讲解 安装mingw+Theano+keras（多警告、错误解决办法），希望对大家有帮助，欢迎收藏，转发！站点地址：www.bmabk.com

安装mingw+Theano+keras

一、关于深度学习框架TensorFlow、Theano 和 Keras

https://www.cnblogs.com/shenxiaolin/p/11121152.html

Keras 可以基于两个Backend（也就是Keras 基于什么东西来做运算）：
一个是 Theano，一个是 Tensorflow（默认）。可以通过修改Backend将默认的Tensorflow改为Theano。
如果我们选择Theano作为Keras的Backend， 那么Keras就用 Theano 在底层搭建你需要的神经网络；
同样，如果选择 Tensorflow 的话呢，Keras 就使用 Tensorflow 在底层搭建神经网络。

二、可以安装theano、keras的python版本

theano、keras兼容的python版本：

 - theano -> python[version='>=2.7,<2.8.0a0|>=3.7,<3.8.0a0|>=3.8,<3.9.0a0|>=3.6,<3.7.0a0|>=3.5,<3.6.0a0']

 - keras -> python[version='>=3.5,<3.6.0a0|>=3.6,<3.7.0a0']

我们选择python3.5进行安装。

三、安装mingw+Theano+keras

1. 进入cmd，创建虚拟环境keras：

至于需要创建虚拟环境的原因，请参见文章：Anaconda、Jupyterlab、虚拟环境、Kernel

conda create --name keras python=3.5

2. 进入虚拟环境keras：

conda activate keras

3.安装 mingw 和 libpython ：

conda install mingw libpython

报错：

PackagesNotFoundError: The following packages are not available from current channels:

  - mingw

Current channels:

  - https://repo.anaconda.com/pkgs/msys2/win-64
  - https://repo.anaconda.com/pkgs/msys2/noarch
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/win-64
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/noarch
  - https://repo.anaconda.com/pkgs/main/win-64
  - https://repo.anaconda.com/pkgs/main/noarch
  - https://repo.anaconda.com/pkgs/r/win-64
  - https://repo.anaconda.com/pkgs/r/noarch

To search for alternate channels that may provide the conda package you are looking for, navigate to

    https://anaconda.org

and use the search bar at the top of the page.

解决：

anaconda search -t conda mingw

显示：

Using Anaconda API: https://api.anaconda.org
C:\ProgramData\Anaconda3\lib\site-packages\urllib3\connectionpool.py:857: InsecureRequestWarning: Unverified HTTPS request is being made. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.io/en/latest/advanced-usage.html#ssl-warnings
  InsecureRequestWarning)
Packages:
     Name                      |  Version | Package Types   | Platforms       | Builds
     ------------------------- |   ------ | --------------- | --------------- | ----------
     RMG/mingwpy               |    0.1.0 | conda           | win-32, win-64  | py27_3
                                          : GCC for Windows 64 & 32 bits
     bcilab/qt_mingw_32_release |   5.15.0 | conda           | win-32          | 0
     bcilab/qt_mingw_64_release |   5.15.0 | conda           | win-64          | 0
     bcilab/resonance_beta_mingw_32 |    3.3.1 | conda           | win-32          | 103
     bcilab/resonance_mingw_32 |    3.4.1 | conda           | win-32          | 11, 13, 138, 20, 0, 79, 18, 27, 23, 28, 132, 131, 89, 72, 134, 80, 78, 86, 87, 84, 85
     bcilab/resonance_mingw_64 |    3.4.1 | conda           | win-32, win-64  | 25, 27, 20, 18, 28, 0, 2, 4
     conda-forge/mingwpy       |    0.1.0 | conda           | win-32, win-64  | py27_2, py34_1, py34_0, py27_0, py27_1, py35_0, py34_2
                                          : GCC for Windows 64 & 32 bits
     free/mingw                |      4.7 | conda           | win-32, win-64  | 1, 0
     ijstokes/mingwpy          |   v0.1.0 | conda           | win-64          | 0.1.0
     lucaszw/mingw-runtime     |      4.7 | conda           | win-32          | 0
     mfix/libboost-mingw       |   1.67.0 | conda           | win-64          | 1
     mwm126/libboost-mingw     |   1.65.1 | conda           | win-64          | 1, 0
     mwojcikowski/mingwpy      |    0.1.0 | conda           | win-32, win-64  | py34_2, py34_1, py34_0, py27_0, py27_1, py35_0, py27_2
                                          : GCC for Windows 64 & 32 bits
     omnia/mingwpy             |    0.1.0 | conda           | win-32, win-64  | py27_3, py36_3, py34_3, py33_3
                                          : GCC for Windows 64 & 32 bits
     takluyver/mingw-w64       |  3.1.0.1 | conda           | linux-64        | 0
     tz-lom/qt_mingw_64_release |          | conda           | win-64          | 0
     tz-lom/resonance_mingw_32 |          | conda           | win-32          | 139
     tz-lom/resonance_mingw_64 |          | conda           | win-64          | 27
     ukoethe/mingw             |          | conda           | win-64          | 0, vc11
     ukoethe/mingw-for-msvc    |    4.7.2 | conda           | win-64          | 0
                                          : Configure MinGW as a subtool of Visual Studio (needed to compile packages that MSVC can't handle, e.g. Fortran sources)
     ukoethe/mingw-for-msvc-build |    4.7.2 | conda           | win-64          | 0
                                          : Configure MinGW as a subtool of Visual Studio (needed to compile packages that MSVC can't handle, e.g. Fortran sources)
     wheeler-microfluidics/mingw-runtime |      4.7 | conda           | win-32          | 0
     wheeler-microfluidics/tdm-mingw |  1.908.0 | conda           | win-32          | 1
     willyd/openblas-mingw     |   0.2.14 | conda           | win-64          | 0
Found 24 packages

Run 'anaconda show <USER/PACKAGE>' to get installation details

选择一个从上述路径中选一个可行的路径再次安装：例如选择 free/mingw | 4.7 | conda | win-32, win-64 | 1, 0

conda install -c https://conda.anaconda.org/free mingw

安装libpython：

conda install libpython

查看是否成功：

g++ --version

g++.exe (GCC) 4.7.0 20111220 (experimental)
Copyright (C) 2011 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

4. 安装theano：

conda install theano

有可能会出现连接失败的问题，重试即可。

5. 安装keras：

conda install keras

6. 检验是否安装成功：

1. 进入python，使用import theano检验theano是否安装成功：

(keras) C:\Users\huipu>python
Python 3.8.11 (default, Aug  6 2021, 09:57:55) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import theano
WARNING (theano.tensor.blas): Using NumPy C-API based implementation for BLAS functions.

出现了警告：Using NumPy C-API based implementation for BLAS functions.

警告解决方法：

a. 第一步：安装三个包

#在环境中安装这三个包
conda install mkl
conda install mkl-service
conda install blas

b. 第二步：在用户目录下放置一个文件
在桌面新建一个txt文档，输入如下内容：

#我建的是这个
[blas]
ldflags=-lmkl_rt

或者输入如下内容：

[blas]
ldflags=-lblas

将文档另存为文件名：.theanorc.txt ，文档类型为：所有文件（.） 。文件的存放路径为：C:\Users\你的电脑用户名 。

c. 验证，再次import theano就没有警告啦！

(keras) C:\Users\huipu>python
Python 3.8.11 (default, Aug  6 2021, 09:57:55) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import theano
>>> print (theano.config.blas.ldflags)
-lmkl_rt

2. theano.test()检验theano是否真的安装成功：

虽然现在import成功了。但是，大佬都用theano.test()来检验theano是否真的安装成功，即使出现了一堆错误，但是如果出现default=0 的话，就说明安装成功了~ ~

去python里theano.test()：

(keras)C:\Users\huipu>python
Python 3.5.6 |Anaconda, Inc.| (default, Aug 26 2018, 16:05:27) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import theano
>>> theano.test()
Theano version 1.0.2
theano is installed in C:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\theano
NumPy version 1.14.2
NumPy relaxed strides checking option: True
NumPy is installed in C:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\numpy
Python version 3.5.6 |Anaconda, Inc.| (default, Aug 26 2018, 16:05:27) [MSC v.1900 64 bit (AMD64)]
nose version 1.3.7
EE...........E.EE.....EEEE....E..................................C:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\theano\gof\vm.py:887: UserWarning: CVM does not support memory profile, using Stack VM.
  'CVM does not support memory profile, using Stack VM.')
.E.........SS.............0.7415943278643314
0.7415943278643314
0.7415943278643314
0.7415943278643314
....E...E.....EEEEE..E......................................E...................................................E......................EC:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\theano\gof\vm.py:890: UserWarning: LoopGC does not support partial evaluation, using Stack VM.
  'LoopGC does not support partial evaluation, '
.C:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\theano\gof\vm.py:890: UserWarning: LoopGC does not support partial evaluation, using Stack VM.
  'LoopGC does not support partial evaluation, '
C:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\theano\gof\vm.py:890: UserWarning: LoopGC does not support partial evaluation, using Stack VM.
  'LoopGC does not support partial evaluation, '
.....EEEESEEEEEEEESEEEES.C:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\theano\gpuarray\type.py:895: UserWarning: config.experimental.unpickle_gpu_on_cpu is set to True. Unpickling GpuArray as numpy.ndarray
  "config.experimental.unpickle_gpu_on_cpu is set to True. "
.EEEEEES.....SEE...C:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\theano\sandbox\rng_mrg.py:1031: UserWarning: MRG_RandomStreams.multinomial_wo_replacement() is deprecated and will be removed in the next release of Theano. Please use MRG_RandomStreams.choice() instead.
  warnings.warn('MRG_RandomStreams.multinomial_wo_replacement() is '
.C:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\theano\sandbox\rng_mrg.py:1031: UserWarning: MRG_RandomStreams.multinomial_wo_replacement() is deprecated and will be removed in the next release of Theano. Please use MRG_RandomStreams.choice() instead.
  warnings.warn('MRG_RandomStreams.multinomial_wo_replacement() is '
...EES..E....E.......EEEE.EEEEEEEEEEEEEEEEEEEE.....EEEEE...E....E..E..S.....EEEEEEEEE......EEE......EEEE..........SEE..SSSS.......S.....E...
======================================================================
ERROR: Failure: ImportError (No module named 'parameterized')
======================================================================
ERROR: test_nnet (theano.compile.tests.test_misc.TestNnet)
======================================================================
ERROR: theano.compile.tests.test_mode.test_no_output_from_implace
======================================================================
ERROR: theano.compile.tests.test_monitormode.test_not_inplace
======================================================================
ERROR: theano.compile.tests.test_nanguardmode.test_NanGuardMode
======================================================================
ERROR: test_input_aliasing_affecting_inplace_operations (theano.compile.tests.test_pfunc.Test_aliasing_rules)
======================================================================
ERROR: test_partial_input_aliasing_affecting_inplace_operations (theano.compile.tests.test_pfunc.Test_aliasing_rules)
======================================================================
ERROR: test_profiling (theano.compile.tests.test_profiling.Test_profiling)
======================================================================
ERROR: test_constant (theano.gof.tests.test_compute_test_value.TestComputeTestValue)
======================================================================
ERROR: test_ndarray (theano.gof.tests.test_compute_test_value.TestComputeTestValue)
======================================================================
ERROR: test_scan_err2 (theano.gof.tests.test_compute_test_value.TestComputeTestValue)
======================================================================
ERROR: test_shared (theano.gof.tests.test_compute_test_value.TestComputeTestValue)
======================================================================
ERROR: test_string_var (theano.gof.tests.test_compute_test_value.TestComputeTestValue)
======================================================================
ERROR: test_variable_only (theano.gof.tests.test_compute_test_value.TestComputeTestValue)
======================================================================
ERROR: test_node_outputs_not_used (theano.gof.tests.test_fg.TFunctionGraph)
======================================================================
ERROR: theano.gof.tests.test_link.test_sort_schedule_fn
======================================================================
ERROR: theano.gof.tests.test_vm.test_partial_function
======================================================================
ERROR: Demonstrate stochastic gradient descent optimization for a multilayer
======================================================================
ERROR: theano.tensor.tests.test_misc.test_bug_2009_07_17_borrowed_output
----------------------------------------------------------------------
Ran 413 tests in 32.590s

FAILED (SKIP=15, errors=101)
<nose.result.TextTestResult run=413 errors=101 failures=0>

我的确出现了这么一大堆错误，但是我也没看到default=0字样。
未完待续。。。

3.如果用高于3.5.X的python环境进行安装， theano.test()时会出现以下错误，并附带解决办法：

这是之前在Python 3.8环境中进行安装时发生的错误：

(theano) C:\Users\huipu>python
Python 3.8.11 (default, Aug  6 2021, 09:57:55) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import theano
>>> theano.test()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\ProgramData\Anaconda3\envs\theano\lib\site-packages\theano\__init__.py", line 148, in test
    raise ImportError("The nose module is not installed."
ImportError: The nose module is not installed. It is needed for Theano tests.

test（）出错了：提示nose模块没有安装。

错误解决方法：

不慌，首先我们来看看我们安装nose了没：

用pip list和conda list查看了一下，列表里都显示已安装nose 1.3.7。而且在python里单独import nose没出错：

(theano) C:\Users\huipu>python
Python 3.8.11 (default, Aug  6 2021, 09:57:55) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import nose
>>> nose
<module 'nose' from 'C:\\ProgramData\\Anaconda3\\envs\\theano\\lib\\site-packages\\nose\\__init__.py'>
>>> result=nose.run()

----------------------------------------------------------------------
Ran 0 tests in 0.069s

OK

所以我不由得怀疑：可能nose 1.3.7和当前的python环境3.8.11不兼容？？？

  1. 所以我先尝试升级nose包：conda update nose，升级完之后仍然是1.3.7版本。升级nose行不通。
  2. 既然nose升不了，那就降级python吧：Currently, Theano Release 1.0 only supports >=3.4 and < 3.6. If your Python version is higher than 3.5.X, try downgrading and then re-run the Theano test. Do let everyone know if that works.
    重新安装一个3.5.X版本的python，就是重复之前的步骤就好。
  3. 再去test()，发现错误：没有nose模块。解决：安装nose模块。
(keras) C:\Users\huipu>python
Python 3.5.6 |Anaconda, Inc.| (default, Aug 26 2018, 16:05:27) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import theano
>>> theano.test()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\ProgramData\Anaconda3\envs\theano_2\lib\site-packages\theano\tests\main.py", line 107, in test
    from nose.config import Config
ImportError: No module named 'nose'

#安装nose模块
conda install nose

装好之后，再去test（），会发现之前的ImportError: The nose module is not installed. It is needed for Theano tests.这个错误消失了！！！
说明降级python是有效的。不能用太高版本的python安装theano。

参考链接

WARNING (theano.tensor.blas): Using NumPy C-API based implementation for BLAS functions.

版权声明：本文内容由互联网用户自发贡献，该文观点仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容， 请发送邮件至 举报，一经查实，本站将立刻删除。

文章由极客之音整理，本文链接：https://www.bmabk.com/index.php/post/119069.html

赞 (0)
seven_的头像seven_bm
0 0

相关推荐

发表回复

登录后才能评论
极客之音——专业性很强的中文编程技术网站，欢迎收藏到浏览器，订阅我们！