Hello cheetha
were you able to solve the problem. I have the same problem here. Any help will be greatly appreciated
Hi,
To download D2L notebooks,
when i run the following command āunzip d2l-en.zip && rm d2l-en.zipā
it give the following error
āāunzipā is not recognized as an internal or external command, operable program or batch file.ā
when i try to install unzip using āpip install unzipā command, i get the following error again:
āunzipā is not recognized as an internal or external command,
operable program or batch file.
How to fix it please?
Hello. Whatās the mxnet distro for Cuda 11.4? I am on wsl2 and the latest cuda that I got officially installed is 11.4. But you seem to have mxnet cuda-adapted only until 11.2? Will it work also?
Hi,
Thanks for the book. I wonder if some guidance regarding the hardware requirements could make a difference. For example, running code from the Softmax-classification of chapter 3 in a AWS fully managed Jupyter notebook on a ml.t2.medium instance shows some limitations. Not surprisingly, upgrading to a ml.t2.large instance produces an improvement. This deserves further specification. What is the minimum hardware required to run all the examples in the book?
Thanks again,
Omar
Nice step by step tutorial! I wonder as a beginner, what is the recommendation of the framework? Do I need to install 3 of them or I can stick to just one of them?
Hello,
I wanted to install MXNet 2.0 for cuda version 11.4, but there were no instructions regarding how to install it(understandable since it just recently got in beta)
I thought of compiling it from source, however, I saw that the builds for centOS were failing for master(Iām using Fedora 34, so thatās what I concentrated on).
I fathom that the 2.0 release is not far away now since it has entered beta, so I donāt want to feel frustrated upon realizing that my effort to learn mxnet will be for nothing in a few months.
Does it matter that I stick to 1.8.0 for now? Does the book accommodate for the fact that MXNet is undergoing a major overhaul? Should I play it safe and just stick to Pytorch instead?
Thank you in advance for your help 
Regards,
Tejas Garhewal
i have a DELL G5 15 laptop with Geforce GTX 1060 GPU. i installed:
- cuda toolkit 10.1 update2 in windows 10
- cuda toolkit using conda in the active environment
- pip install mxnet-cu101==1.5.0 (as pip refuses to install any version higher than 1.5.0)
in python, i could import mxnet and access the gpu, but i cannot import np and npx.
any help, what is the appropriate installation and versions of cuda and mxnet are compatibel to my system configuration? or how may i import np and npx?
How to fix this error
OSError: libcudnn.so.8: cannot open shared object file: No such file or directory
I ran distributions.ipnyb
I installed mxnet 112
my gpu is nvidia gtx 1650
Reply for more info/
Hey Rachel,
I am in love with this book and would like to contribute towards the solution of exercise.If feasible could you please share the GitHub repository link with me for development for the same?
Thanks,
Al
Hello, Iāve successfully installed mxnet gpu, but I always get an error at the following code:
from d2l import mxnet as d2l
The error is:
ModuleNotFoundError: No module named ād2lā
The problem goes away if I write %cd .. though. Nevertheless, Iām interested in learning what the proper solution for this is.
Thanks in advance!
I can import modules in a Python REPL, but not in jupyter notebooks. To wit:
---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
<ipython-input-1-526f864be1f6> in <module>
1 get_ipython().run_line_magic('matplotlib', 'inline')
2 import random
----> 3 from mxnet import np, npx
4 from d2l import mxnet as d2l
5
ModuleNotFoundError: No module named 'mxnet'Helloļ¼I use āpip installā to install package, but the process crashed when pip start ādownloadingā.
Thank you!
Hello all,
I am running my setup on MacBook with appleās silicon. Itās GPU does not have CUDA, will I be okay with the the mxnet installation for CPU only?
