Jupyter：ImportError：libcusolver.so.8.0：无法打开共享对象文件：没有这样的文件或目录

While I run the following code in Jupyter notebook:

我在Jupyter笔记本中运行以下代码:

import tensorflow as tf

a = tf.constant("hello world!")

sess = tf.Session()
print(sess.run(a))

I got the following error messages:

我收到以下错误消息:

ImportError: Traceback (most recent call last):
  File "/home/ac/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow.py", line 41, in <module>
    from tensorflow.python.pywrap_tensorflow_internal import *
  File "/home/ac/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in <module>
    _pywrap_tensorflow_internal = swig_import_helper()
  File "/home/ac/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper
    _mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
  File "/home/ac/anaconda3/lib/python3.6/imp.py", line 242, in load_module
    return load_dynamic(name, filename, file)
  File "/home/ac/anaconda3/lib/python3.6/imp.py", line 342, in load_dynamic
    return _load(spec)
ImportError: libcusolver.so.8.0: cannot open shared object file: No such file or directory

However, the same code if I run directly through python, it works as expected.

但是,如果我直接通过python运行相同的代码,它按预期工作。

Here is my environment:

这是我的环境:

Ubuntu 16.04
Tensorflow 1.5
Cuda 9.0
Cudnn 7
Nvidia Driver 390.12

Nvidia Driver 390.12

Python 3.6
Anaconda 3

The issue happens after I recently upgrade the Tensorflow. Before I was using Tensorflow 1.4 with Cuda 8 and Cudnn 6. However, I feel this is more like Jupyter notebook issue than Tensorflow installation.

我最近升级Tensorflow后问题就出现了。在我使用Tensorflow 1.4与Cuda 8和Cudnn 6之前。但是,我觉得这比Tensorflow安装更像是Jupyter笔记本问题。

I also tried other solution like export LD_LIBRARY_PATH, no luck. I wonder if the Jupyter using the different library than the avocado environment? Or this issue is caused by some failure installation?

我也试过像导出LD_LIBRARY_PATH这样的其他解决方案,没有运气。我想知道Jupyter是否使用了不同于鳄梨环境的库?或者这个问题是由安装失败引起的?

1 个解决方案

#1

The solution I found is following:

我找到的解决方案如下:

First, I installed the Jupyter extension nb_conda via conda install nb_conda, which will add the ability to view the current kernel environment in Jupyter. Then I realize the Jupyter doesn't use the correct environment I expected.

首先,我通过conda install nb_conda安装了Jupyter扩展nb_conda,这将添加查看Jupyter中当前内核环境的功能。然后我意识到Jupyter没有使用我期望的正确环境。

Second, install Jupyter under the wanted environment conda install Jupyter.

其次,在想要的环境下安装Jupyter conda安装Jupyter。

Finally, the Jupyter is running under the environment where calls jupyter notebook.

最后,Jupyter在调用jupyter notebook的环境下运行。

Hope this will help anyone encountered the same issue as I did.

希望这会帮助任何人遇到与我相同的问题。

#1