일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | |||||
3 | 4 | 5 | 6 | 7 | 8 | 9 |
10 | 11 | 12 | 13 | 14 | 15 | 16 |
17 | 18 | 19 | 20 | 21 | 22 | 23 |
24 | 25 | 26 | 27 | 28 | 29 | 30 |
- ppc64le
- CUDA
- Windows 10
- Jupyter notebook
- gpu memory
- python
- dlib
- ubuntu
- object detection
- YouTube 8M
- TensorFlow
- urllib
- install
- keras
- linux
- windows
- python3
- pyTorch
- colaboratory
- colab
- Deep Learning
- 딥러닝
- face_recognition
- FIle
- error
- download
- shakeratos
- Anaconda
- dataset
- raspberry pi
- Today
- Total
Shakerato
Theano (using GPU) install in Windows 10 본문
1. Install Anaconda3, GPU driver, CUDA, cudnn
2. Run 'cmd' as administrator
3. conda create -n [myenv] python=3.6 anaconda
4. activate [myenv]
5. conda install theano pygpu
(if you got error, try this command: "conda install -c conda-forge pygpu")
6. conda install numpy scipy mkl-service libpython
7. Create a file "C:\Users\[Username]\.theanorc"
[global]
floatX = float32
device = cuda0
8. Test code on GPU (cuda0 = first GPU installed on your PC)
http://deeplearning.net/software/theano_versions/dev/tutorial/using_gpu.html#testing-theano-with-gpu
[when you got error like below, install old version theano as below]
theano.gradient.NullTypeGradError: tensor.grad encountered a NaN. This variable is Null because the grad method for input 0 (Reshape{2}.0) of the UndefinedGrad op is mathematically
[theano-gpu old version install method]
theano > 1.0.0 --> 0.9.0
pygpu 0.7.6 --> 0.6.9
----------
conda config --append channels conda-forge
conda install mingw
'Research' 카테고리의 다른 글
Video Action Classification using 3D ResNet (0) | 2019.02.18 |
---|---|
convert *.mat data file to json file (0) | 2019.02.14 |
딥러닝 객체인식 네트워크 retinanet 내 데이터셋으로 학습하는 방법 (1) | 2019.01.17 |
GPU devices checking tips for Deep Learning (Be careful!) (0) | 2019.01.16 |
Counting files with extension in linux (0) | 2019.01.13 |