問(wèn)題1
- The cuda backend is deprecated and will be removed in the next release (v0.10). Please switch to the gpuarray backend. You can get more information about how to switch at this URL:
https://github.com/Theano/Theano/wiki/Converting-to-the-new-gpu-back-end%28gpuarray%29 (還可能引起這個(gè)問(wèn)題:current context was not created by the StreamExecutor cuda_driver API: 0000028DC2D757B0; a CUDA runtime call was likely performed without using a StreamExecutor context） - 原因:
cuda后臺(tái)在下一個(gè)版本會(huì)被廢棄，需要改成gpuarray后臺(tái),異常說(shuō)的很清楚筷转。 - 解決方法:
直接查看鏈接姑原，照著鏈接上的改下。一般是這幾個(gè)步驟: - 安裝:
- 安裝libgpuarray
- 使用theano最新版本(0.9) Theano dev version
- 或者直接使用conda 命令安裝
conda install theano pygpu
- 把theano的配置文件
.theanorc.txt,把device=gpu改成device=cuda or cuda0, ...
問(wèn)題2
- The TensorFlow library wasn't compiled to use SSE instructions...
- 原因:tensorflow不是自己編譯的
- 解決方法:
自己編譯呜舒，或者買(mǎi)個(gè)顯卡锭汛，或者無(wú)視(如果有GPU可以忽略這個(gè)問(wèn)題)
問(wèn)題3
- Loaded runtime CuDNN library: 5005 (compatibility version 5000) but source was compiled with 5103 (compatibility version 5100). If using a binary install, upgrade your CuDNN library to match. If building from sources, make sure the library loaded at runtime matches a compatible version specified during compile configuration.
- 原因:cudnn版本和cuda不兼容
- 解決方法:
官網(wǎng)更新cudnn后,拷貝cuDNN庫(kù)文件到cuda目錄下
問(wèn)題4
- vs安裝完成后運(yùn)行deviceQuery，如果result=pass即為成功
問(wèn)題5(python3中出現(xiàn))
- 在import theano后出現(xiàn)Theano ImportError:cannot import name 'configparser'問(wèn)題
- 原因: 和python的configparser沖突
- 解決方法:
- 不要在環(huán)境變量添加PYTHONPATH變量
- 如果非要加環(huán)境變量，那就把theano下的configparser改名唤殴，如configparser_tf般婆，然后在運(yùn)行一段theano的代碼，依次改掉出錯(cuò)的地方
問(wèn)題6
- import時(shí)可能會(huì)出現(xiàn)WARNING (theano.configdefaults): g++ not detected ! Theano will be unable to execute optimized C-implementations (for both CPU and GPU) and will default to Python implementations. Performance will be severely degraded. To remove this warning, set Theano flags cxx to an empty string.
- 原因: 缺少mingw
- 解決方法:
- conda install mingw libpython
問(wèn)題7
- ERROR: Failure: ImportError (No module named nose_parameterized)
- 解決方法：
- pip install nose_parameterized
問(wèn)題8
- 如果遇到很奇怪的錯(cuò)朵逝，將 C:\Users\你的用戶(hù)名\AppData\local\Theano下的緩存清除
