【ゼロから始める机器学习】(1)装环境,keras (win,cpu
你可能会问为啥不装gpu版本,你是不是傻?
喵的我显卡是AMD的,哈哈哈哈哈哈蛤蛤蛤。。。蛤蛤。蛤。。。。唉。。。。?????????辣鸡按摩店
安装过程中时不时会发出“喵喵喵喵喵????”的感慨,记录下遇到的问题
keras中文文档里的安装步骤:
http://keras-cn.readthedocs.io/en/latest/for_beginners/keras_windows/
1.装Anaconda,推荐装旧版本,Anaconda3 4.2.0,是python3.5的。这玩意集成了各种货,到时候就不用自己装了,爽不爽
https://repo.continuum.io/archive/index.html
然后你就去下载吧,能下得下来算我输,吼吼吼
所以我就滚去清华的镜像站了https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/
妈呀下得好快!
2.装TensorFlow
pip install --upgrade https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-1.0.1-cp35-cp35
装的时候你(是我)就发现,有个依赖项装不上的,忘了是哪个了
直接把那个单独弄一下
3.装keras
pip install keras -U --pre
然后,scipy并装不上,单独pip install也没有卵用,但是去下载下来whl装上就可以了
http://www.lfd.uci.edu/~gohlke/pythonlibs/#scipy
然后就假装自己装好了,然后去试一试呗
import numpy as np
from sklearn import datasets, linear_model
import matplotlib.pyplot as plt
然后报了这么个错
ImportError: cannot import name 'NUMPY_MKL'
问题是这里导致的from numpy._distributor_init import NUMPY_MKL # requires numpy+mkl
stackoverflow上一个大佬是这么说的
This means that you've installed the numpy
by pip
, but the scipy
was installed by precompiled archive, which expects numpy+mkl
This problem can be easy solved by installation for numpy+mkl
from whl file from here.
numpy是通过pip装的,scipy是下的whl装的,它要的是numpy+mkl,好有道理哦,刚才不是scipy装不上来着么。。。。
http://www.lfd.uci.edu/~gohlke/pythonlibs/#numpy
下载numpy+skl,装上就好(假装)了