在 bash 中运行良好时在 Jupyter Notebook 中导入 TensorFlow 时出现问题答案

【问题标题】：Trouble with import TensorFlow in Jupyter Notebook while runs fine in bash在 bash 中运行良好时在 Jupyter Notebook 中导入 TensorFlow 时出现问题
【发布时间】：2017-04-30 13:16:43
【问题描述】：

我在没有安装 anaconda 的情况下使用“使用本机 pip 安装”安装了张量流，并且能够在从终端启动的 Python 控制台中导入 TensorFlow。 image

$ python
Python 2.7.13 (default, Apr  4 2017, 08:47:57)
[GCC 4.2.1 Compatible Apple LLVM 8.1.0 (clang-802.0.38)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
>>> hello = tf.constant('Hello, TensorFlow!')
>>> sess = tf.Session()
2017-04-30 20:54:36.278740: W     tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2017-04-30 20:54:36.278773: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
2017-04-30 20:54:36.278785: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.
2017-04-30 20:54:36.278794: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.
2017-04-30 20:54:36.423246: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:865] OS X does not support NUMA - returning NUMA node zero
2017-04-30 20:54:36.432456: I     tensorflow/core/common_runtime/gpu/gpu_device.cc:887] Found device 0 with     properties:
name: GeForce GTX 1080
major: 6 minor: 1 memoryClockRate (GHz) 1.86
pciBusID 0000:01:00.0
Total memory: 8.00GiB
Free memory: 6.12GiB
2017-04-30 20:54:36.432507: I tensorflow/core/common_runtime/gpu/gpu_device.cc:908] DMA: 0
2017-04-30 20:54:36.432511: I tensorflow/core/common_runtime/gpu/gpu_device.cc:918] 0:   Y
2017-04-30 20:54:36.432519: I tensorflow/core/common_runtime/gpu/gpu_device.cc:977] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1080, pci bus id: 0000:01:00.0)
>>> print(sess.run(hello))
Hello, TensorFlow!
>>>

但我在 Jupiter notebook 中运行了相同的代码，它显示了。 image

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-1-d7933b52e0de> in <module>()
----> 1 import tensorflow as tf
      2 hello = tf.constant('Hello, TensorFlow!')
      3 sess = tf.Session()
      4 print(sess.run(hello))

ImportError: No module named tensorflow

我该如何解决这个问题？

【问题讨论】：

你是如何运行 jupyter notebook 的？
只需在终端中运行“jupyter notebook”即可。
我认为你的 jupyter 是用 pip3(python3) 安装的，而 tensorflow 是用 pip(python2) 安装的
安装的jupyter版本是python2.7。这是图像。 i.stack.imgur.com/a9mzS.png
另外我建议不要太熟悉 jupyter 和 tensorflow。当您使用带有 tensorflow 的 GPU 时，TF 会贪婪地将所有内存分配给所有可见的 GPU（使用 CUDA_VISIBLE_DEVICES env 变量）。在 jupyter 中，GPU 分配将保持不变，直到您重新启动 Jupyter 内核。在你自己的机器上用一个 GPU 没什么大不了的。但是尝试在共享机器、多个 GPU 上运行您的代码……很容易让 GPU 完全分配/意外锁定。我建议这里的每个人都从命令行脚本而不是 jupyter 运行 TF 操作。

标签： python tensorflow jupyter

【解决方案1】：

Python、Tensorflow 和 Jupyter 都必须在保存版本 python2 / python3 上。以下是使用 python3 设置 jupyter 和 Tensorflow 的命令

sudo apt install python3-pip
pip3 install --upgrade pip   ##optional
pip3 install jupyter


##Install Tensorflow
pip3 install --upgrade tensorflow

Test
$ python3
Python 3.5.2 (default, Nov 17 2016, 17:05:23) 
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
>>>

##Launch jupyter
jupyter notebook   
(In Ubuntu you may need to launch ~/.local/bin/jupyter-notebook until you relaunch the bash i.e. PATH issue)
##Create new notepad and run
import tensorflow as tf

【讨论】：