安装

Unreadable Notebook: /Users/leon/study/AI/d2l-zh/tensorflow/index.ipynb NotJSONError(“Notebook does not appear to be JSON: ‘’…”)

按照安装步骤完成了但是出现两个问题。
OS:ubuntu-20.04.1
python3.8 | Miniconda3 Linux 64-bit

1.conda activate d2l后,python中
No module named ‘mxnet’.

2.运行jupyter notebook
jupyter command ‘jupyter-notebook’ not found.

请高手赐教。

以上问题是GPU安装时出现的。
又按照CPU的放发安装了一次,运行jupyter notebook成功。

问题如下:

python3.6 | Miniconda3-4.3.31-Linux-x86_64.sh

1.conda activate d2l后,python中import mxnet
lllegal instruction (core dumped)

请教高手,谢谢。

OS:ubuntu-20.04.1
python3.8 | Miniconda3 Linux 64-bit

第二个问题解决了,运行了
pip install -U d2l

第一个问题发生了变化
1.conda activate d2l后,python中import mxnet
lllegal instruction (core dumped)

请问一下,安装了mxnet过后,import的时候出现module ‘numpy’ has no attribute ‘histogram_bin_edges’,我把mxnet卸载了又安装,还是不行呢

在Win10上安装了基于CPU的环境,可以成功运行。大致步骤:

  1. 先安装miniconda -https://docs.conda.io/en/latest/miniconda.html , 我选的是python3.8
  2. 激活base环境 -cmdline窗口使用命令conda activate base
  3. 到pytorch主页(Start Locally | PyTorch ),获取安装pytorch的命令,我的大概是:conda install pytorch torchvision torchaudio cpuonly -c pytorch

    如果安装中遇到如下错误:
RemoveError: 'requests' is a dependency of conda and cannot be removed from
conda's operating environment.

请使用如下命令更新:conda update --force conda
然后再尝试安装pytorch
4. 然后安装jupyter 和d2l包,在base 环境下,输入pip install jupyter d2l


5. 下载教程的jupyter记事本,就在教材页面右上方工具栏中(《动手学深度学习》 — 动手学深度学习 2.0.0 documentation

找个合适的目录解压缩

  1. 在base环境下启动jupyter notebook
  2. 一切正常,浏览器自动弹出,即可打开各章节jupyter notebook进行学习和试验了。
    (注:由于是本地CPU版本,很多模型训练很慢,甚至是训练不出来)
1 Like

麻烦看看如下的问题,安装之后遇到RuntimeError,在浏览器出现了ImportError,如下图,求解答,谢谢。


(d2l) E:\AI\DeepLearning Tools\d2l-zh v2>jupyter notebook
[W 16:19:12.024 NotebookApp] Terminals not available (error was No module named ‘winpty.winpty’)
[I 16:19:12.026 NotebookApp] Serving notebooks from local directory: E:\AI\DeepLearning Tools\d2l-zh v2
[I 16:19:12.026 NotebookApp] Jupyter Notebook 6.4.0 is running at:
[I 16:19:12.026 NotebookApp] http://localhost:8888/?token=6147b72317382d92941f2d33458822da54cb725be075da4f
[I 16:19:12.026 NotebookApp] or http://127.0.0.1:8888/?token=6147b72317382d92941f2d33458822da54cb725be075da4f
[I 16:19:12.026 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
[C 16:19:12.078 NotebookApp]

To access the notebook, open this file in a browser:
    file:///C:/Users/GAME-1/AppData/Roaming/jupyter/runtime/nbserver-13528-open.html
Or copy and paste one of these URLs:
    http://localhost:8888/?token=6147b72317382d92941f2d33458822da54cb725be075da4f
 or http://127.0.0.1:8888/?token=6147b72317382d92941f2d33458822da54cb725be075da4f

ERROR:asyncio:Exception in callback <TaskWakeupMethWrapper object at 0x000001C799F35258>(<Future finis…bbf"\r\n\r\n’>)
handle: <Handle <TaskWakeupMethWrapper object at 0x000001C799F35258>(<Future finis…bbf"\r\n\r\n’>)>
Traceback (most recent call last):
File “c:\users\game-1.conda\envs\d2l\lib\asyncio\events.py”, line 88, in _run
self._context.run(self._callback, *self._args)
RuntimeError: Cannot enter into task <Task pending coro=<HTTP1ServerConnection._server_request_loop() running at c:\users\game-1.conda\envs\d2l\lib\site-packages\tornado\http1connection.py:823> wait_for=<Future finished result=b’GET /api/co…2bbf"\r\n\r\n’> cb=[IOLoop.add_future..() at c:\users\game-1.conda\envs\d2l\lib\site-packages\tornado\ioloop.py:688]> while another task <Task pending coro=<MappingKernelManager.start_kernel() running at c:\users\game-1.conda\envs\d2l\lib\site-packages\notebook\services\kernels\kernelmanager.py:176> cb=[IOLoop.add_future..() at c:\users\game-1.conda\envs\d2l\lib\site-packages\tornado\ioloop.py:688]> is being executed.
ERROR:asyncio:Exception in callback <TaskWakeupMethWrapper object at 0x000001C798AFD618>(<Future finis…bbf"\r\n\r\n’>)
handle: <Handle <TaskWakeupMethWrapper object at 0x000001C798AFD618>(<Future finis…bbf"\r\n\r\n’>)>
Traceback (most recent call last):
File “c:\users\game-1.conda\envs\d2l\lib\asyncio\events.py”, line 88, in _run
self._context.run(self._callback, *self._args)
RuntimeError: Cannot enter into task <Task pending coro=<HTTP1ServerConnection._server_request_loop() running at c:\users\game-1.conda\envs\d2l\lib\site-packages\tornado\http1connection.py:823> wait_for=<Future finished result=b’GET /kernel…2bbf"\r\n\r\n’> cb=[IOLoop.add_future..() at c:\users\game-1.conda\envs\d2l\lib\site-packages\tornado\ioloop.py:688]> while another task <Task pending coro=<MappingKernelManager.start_kernel() running at c:\users\game-1.conda\envs\d2l\lib\site-packages\notebook\services\kernels\kernelmanager.py:176> cb=[IOLoop.add_future..() at c:\users\game-1.conda\envs\d2l\lib\site-packages\tornado\ioloop.py:688]> is being executed.
[I 16:19:18.462 NotebookApp] Kernel started: 6f50232a-d0a4-4d0e-925d-396086f76a59, name: python3
[W 16:19:19.128 NotebookApp] 404 GET /static/components/MathJax/fonts/HTML-CSS/TeX/otf/MathJax_AMS-Regular.otf (::1) 5.980000ms referer=http://localhost:8888/notebooks/chapter_preliminaries/ndarray.ipynb

你好,我按照你的步骤进行了安装,安装的时候一切正常。但是到了激活环境的时候,就不行了
报错如下
输入:conda activate d2l
报错:Could not find conda environment: d2l
You can list all discoverable environments with conda info --envs.
请问您有遇到这样的问题吗?

d2l是咱们教程用的Python 包,而不是conda虚拟环境。你需要激活的是base虚拟环境,命令是:conda activate base

非常感谢,现在activate base和deactivate都正常了。
但是我在打开jupyter notebook里的项目例子时,
报错:
[W 13:43:39.079 NotebookApp] 404 GET /api/contents/d2l-zh/pytorch/chapter_linear-networks?type=directory&=1628573992115 (::1): No such file or directory: d2l-zh/pytorch/chapter_linear-networks
[W 13:43:39.087 NotebookApp] No such file or directory: d2l-zh/pytorch/chapter_linear-networks
[W 13:43:39.091 NotebookApp] 404 GET /api/contents/d2l-zh/pytorch/chapter_linear-networks?type=directory&
=1628573992115 (::1) 16.120000ms referer=http://localhost:8888/tree/d2l-zh/pytorch/chapter_linear-networks
[W 13:43:58.612 NotebookApp] Notebook pytorch/chapter_linear-networks/linear-regression-concise.ipynb is not trusted
Bad file descriptor (C:\projects\libzmq\src\epoll.cpp:100)
请指点,谢谢!

错误码404一般表示服务端资源不存在。看样子你是不是没有把下载的jupyter notebooks解压?
我的是解压zip到我的windows用户目录,如图

我也是 请问你后来解决了嘛?我是win10系统 不知道你是什么系统……

解决从Ubuntu虚拟机上的jupyter映射到window的方法https://www.youtube.com/watch?v=qeJUsahqzw8&t=315s

1 Like

按照视频在ubuntu下安装完成了,jupyter notebook在本机也可以打开,可是在局域网内其他机子访问 ip地址:8888 提示拒绝连接

请问如果不是Nvidia显卡就不能下载GPU嘛?我是AMD显卡怎么办,没有GPU有什么影响吗

正式版,支持CUDA的,好像是只到1.7.0,更高版本只有CPU版本。
而且D2L这个项目,是基于1.7.0 的。

希望安装带有 GPU 支持的 MXNET,可以参考这篇文章:

MXNET不同版本对 CUDA 和 MKL 支持的情况,可以看 MXNET 网站:

我的完成步骤,根据知乎的文章和本文的教程:
0. 安装 Anaconda 3.0

  1. 安装 VisualStudio 2017

  2. 安装 NVidia CUDA 10.1

  3. 安装 NVidia CUDNN 8.05 For CUDA 10.1

  4. 安照本文的教程安装 MXNET 1.7.0

  5. 在验证 MXNET GPU 的时候( 参见 MXNET网站 Validate Your MXNet Installation 的 Python 部分)
    import mxnet as mx
    总是报错, libmxnet.dll 无法找到,其实是缺少了 CUDA最新的DLL库文件 CUDART64_101.dll
    从 C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin 把这个文件复制到 Anaconda 创建的 d2l 环境中(libmxnet.dll所在的目录)就OK,我的目录是
    E:\anaconda3\envs\d2l\Lib\site-packages\mxnet

    安装成功后,启动 jupyter notebook
    创建一个新的 Notebook文件,输入以下内容:

from mxnet import nd, npx
npx.set_np()
import mxnet as mx
a = nd.ones((2, 3), mx.gpu())
a

输出结果应该为:
[[1. 1. 1.] [1. 1. 1.]] <NDArray 2x3 @gpu(0)>

说明已经是使用 GPU 上的数组,安装成功!


安装d2l出现这个怎么解决

安装d2l时出现这个问题,怎么解决呀

下载安装d2l安装包出错,numpy和pip都是最新,wheel也提前安装了
怎么解决?