========================================
-
Install visual studio 2010
-
Download and install CUDA 7.0
-
Install TDM GCC
Theano C code compiler currently requires a GCC installation. We have used the build TDM GCC which is provided for both 32- and 64-bit platforms. A few caveats to watch for during installation:
- Install to a directory without spaces (we have placed it in C:\SciSoft\TDM-GCC-64)
- If you don’t want to clutter your system PATH un-check add to path option.
- Enable OpenMP support by checking the option openmp support option.
-
create virtual env for deep learn python, so it will not mess up your current python open Anaconda Command Prompt conda create --name dl_python python=2
-
Install Anaconda with Anaconda Accelerate, which is Fast Python for GPUs and multi-core with NumbaPro and MKL Optimizations)
activate dl_python
conda install mkl
conda install accelerate
conda install mingw libpython
conda install pydot-ng
conda install scikit-learn- Create .theanorc.txt under %USERPROFILE% folder. Copy following into file.
[global]
floatX = float32
device = gpu
[nvcc]
flags = --use-local-env --cl-version=2010- Create a batch file setenv.bat to set the following environment
REM configuration of paths
set VSFORPYTHON="C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC"
set SCISOFT=%~dp0
REM add tdm gcc stuff
set PATH=c:\TDM-GCC-64\bin;c:\TDM-GCC-64\x86_64-w64-mingw32\bin;%PATH%
REM add anaconda env
CALL C:\Users\simuser\Anaconda\Scripts\anaconda.bat
REM configure path for msvc compilers
REM for a 32 bit installation change this line to
REM CALL %VSFORPYTHON%\vcvarsall.bat
CALL %VSFORPYTHON%\vcvarsall.bat amd64
CALL activate dl_python
REM return a shell
cmd.exe /k
- Run the setenv.bat and install Theano
conda install pydot-ng
pip install git+https://github.com/Theano/Theano.git- In setenv.bat: Run python check1.py and you should see that you python script are using gpu.
python check1.py
Using gpu device 0: Quadro 4000
[GpuElemwise{exp,no_inplace}(<CudaNdarrayType(float32, vector)>), HostFromGpu(GpuElemwise{exp,no_inplace}.0)]
Looping 1000 times took 0.910000085831 seconds
Result is [ 1.23178029 1.61879349 1.52278066 ..., 2.20771813 2.29967761
1.62323296]
Used the gpu- Run the setenv.bat and install Lasagne
pip install https://github.com/Lasagne/Lasagne/archive/master.zip
conda install matplotlib
python mnist.py- Run python mnist.py to test installation of lasagne, you should see output information after several nvcc compile information
python mnist.py
Starting training...
Epoch 1 of 50 took 2.630s
training loss: 1.230027
validation loss: 0.410091
validation accuracy: 88.64 %
- Run the setenv.bat and install pandas, scikit-learn, nolearn
conda install pandas
pip install git+https://github.com/dnouri/nolearn.git@master#egg=nolearn==0.7.git- Run python mnist_nolearn.py to test installation of nolearn
python mnist_nolearn.py
Using gpu device 0: Quadro 4000 (CNMeM is disabled)
Loading data...
DEBUG: nvcc STDOUT mod.cu
Creating library C:/Users/simuser/AppData/Local/Theano/compiledir_Windows-7-6.1.7601-SP1-Intel64_Family_6_Model_44_St
epping_2_GenuineIntel-2.7.10-64/tmpozxlxp/88b4f8c09b81504243de03738d23d308.lib and object C:/Users/simuser/AppData/Local
/Theano/compiledir_Windows-7-6.1.7601-SP1-Intel64_Family_6_Model_44_Stepping_2_GenuineIntel-2.7.10-64/tmpozxlxp/88b4f8c0
9b81504243de03738d23d308.exp
# Neural Network with 1276810 learnable parameters
## Layer information
# name size
--- ----------- -------
0 input 1x28x28
1 inputDrop 1x28x28
2 hidden1 800
3 hidden1Drop 800
4 hidden2 800
5 hidden2Drop 800
6 output 10
epoch train loss valid loss train/val valid acc dur
------- ------------ ------------ ----------- ----------- -----
1 ←[36m0.69431←[0m ←[32m0.24281←[0m 2.85946 0.92890 4.79s
2 ←[36m0.35514←[0m ←[32m0.17866←[0m 1.98783 0.94858 4.86s
3 ←[36m0.28206←[0m ←[32m0.14359←[0m 1.96431 0.96064 4.79s
4 ←[36m0.23673←[0m ←[32m0.12444←[0m 1.90238 0.96598 4.88s
5 ←[36m0.21256←[0m ←[32m0.11248←[0m 1.88966 0.96954 4.91s
6 ←[36m0.19146←[0m ←[32m0.10138←[0m 1.88847 0.97093 4.83s
7 ←[36m0.17600←[0m ←[32m0.09579←[0m 1.83726 0.97231 4.91s
8 ←[36m0.16370←[0m ←[32m0.08897←[0m 1.83999 0.97439 4.77s
9 ←[36m0.15022←[0m ←[32m0.08424←[0m 1.78316 0.97488 4.78s
10 ←[36m0.14097←[0m ←[32m0.08132←[0m 1.73363 0.97528 4.93s
11 ←[36m0.13568←[0m ←[32m0.07851←[0m 1.72806 0.97696 4.86s
12 ←[36m0.13062←[0m ←[32m0.07406←[0m 1.76359 0.97785 4.89s- Run the setenv.bat and install keras
pip install keras-
Configure pycharm to run keras
- Run the setenv.bat
- echo %PATH%
- echo %lib%
- echo %include%
- copy these three variable values into pycharm run env vaiables