MacOS+miniconda+theano安装（tensorflow）

1. brew update --debug --verbose

2. brew install wget

3. xcode-select --install

4. 以上都不行

   Error: No such file or directory @ dir_chdir - /usr/local/Cellar

    

   卸载重装homwbrew
   ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/uninstall)"
   /usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)" 
   这样，在最新的HomeBrew中，安装的时候就将需要的目录都已经创建完成并且分配了所需的权限，那么在用brew去安装其他软件的时候就可以不需要sudo权限直接进行了

5. 替换核心软件仓库

   cd "$(brew --repo)/Library/Taps/homebrew/homebrew-core"
   git remote set-url origin https://mirrors.ustc.edu.cn/homebrew-core.git
   替换 cask 软件仓库（提供 macOS 应用和大型二进制文件）

   cd "$(brew --repo)"/Library/Taps/caskroom/homebrew-cask
   git remote set-url origin https://mirrors.ustc.edu.cn/homebrew-cask.git
   替换 Bottles 源（Homebrew 预编译二进制软件包）

   bash（默认 shell）用户：

   echo 'export HOMEBREW_BOTTLE_DOMAIN=https://mirrors.ustc.edu.cn/homebrew-bottles' >> ~/.bash_profile
   source ~/.bash_profile

6. brew install wget
   
   
7. https://docs.conda.io/en/latest/miniconda.html找到合适版本; cd ~

8. wget https://repo.continuum.io/miniconda/Miniconda3-latest-MacOSX-x86_64.sh

9. bash Miniconda3-latest-MacOSX-x86_64.sh
   
   Miniconda3 will now be installed into this location:

10. /Users/jingyuli/miniconda3.

11. 手动加载环境变量来使得conda生效.  source ~/.bash_profile
12. conda -h. 检验安装成功

 

 

接下来theano

• conda create -n theano python=3.5
• conda-env list
• source activate theano

Install requirements and optional packages

• conda install numpy scipy mkl nose sphinx
• pip install parameterized

Install and configure the GPU drivers (recommended) CUDA安装

macbookpro没有英伟达显卡 不支持gpu

• conda install theano pygpu (CondaHTTPError)

• git clone https://github.com/Theano/libgpuarray.git
  cd libgpuarray
  git checkout tags/v0.6.5 -b v0.6.9

• conda install pygpu

• brew install cmake

• mkdir Build
  cd Build
  # you can pass -DCMAKE_INSTALL_PREFIX=/path/to/somewhere to install to an alternate location
  cmake .. -DCMAKE_BUILD_TYPE=Release # or Debug if you are investigating a crash
  make
  make install
  cd ..

• pip install Theano
• conda install mkl-service
• test:  

  ERROR: Failure: ImportError (No module named 'parameterized')

• pip install nose-parameterized

 

 

 

 tensorboard的使用

import tensorflow as tf
from tensorboard import summary
with tf.name_scope('input1'):
    input1=tf.constant([1.0,2.0,3.0],name="input1")
with tf.name_scope('input2'):
    input2=tf.Variable(tf.random_uniform([3]),name="input2")
output=tf.add_n([input1,input2],name="add")
writer=tf.summary.FileWriter(r"./summary",tf.get_default_graph())
writer.close()

 

source activate tf

python /Users/jingyuli/miniconda3/envs/tf/bin/tensorboard  --logdir /Users/jingyuli/PycharmProjects/tf-session/summary

报错

ModuleNotFoundError: No module named 'tensorboard'

 

是因为没有source activate tf
http://localhost:6006/

 

安装tf2

• conda create -n tf2 python=3.7
• conda-env list
• source activate tf2

 

 

 

 

安装pytorch

• conda create -n pytorch python=3.6.5
• conda-env list
• source activate pytorch

conda install pytorch torchvision -c pytorch