You have to make an hard link (mklink /d) that points from the notebooks directory to fastai library (or add fastai library directory to PYTHONPATH). But I advise to follow Jeremy’s new procedure from scratch. Look at the beginning of the thread, you’ll find a link!
i tried to create a link :
for lib, Library and libs to point to the environment fastai libraries but i still get the same error message.
Are these the right libraries?
Suppose you cloned your fastai repository in C:\users\your_account.
Git has now created a new directory: C:\users\your_account\fastai.
The hard link target has to be: C:\users\your_account\fastai\fastai
Let us know!
Newbie here. I cannot find the environment-win.yml. I see environment.yml, environment-cpu.yml, and environment-old.yml.
environment-win.yml has been replaced with environment.yml recently
I followed the directions to a T.
In the fastai folder I typed:
conda env update
and got: conda not found
Its been years since I’ve used Unix but is this a path problem? Is there some .xxxrc file I need to edit and add directories to a path?
When I typed: jupyter notebook
I got the same error about jupyter not found
By following the steps listed under Optional Steps I did get jupyter notebook to work.
I can try : conda env update
when I next ssh into paperspace
Thanks for this! “New Installation Instructions” also worked for me on MacOs for local cpu installation,
except environment-win.yml replaced by environment-cpu.yml
Today i have just managed to set up everything as per the instructions received on this link to set up the fast ai library and a python environment via conda.But i’m getting an issue wherein i have activated the new environment and from the kernel in the jupyter notebook,have selected the fast ai custom kernel.But i stll get an error saying that there is no module named fastAI.
Here are some pictures which can better describe the problem.
I can’t find an “environment-win.yml” file in the repo. I tried with the regular “environment.yml” file, but then jupyter notebook fails to import a bunch of libraries.
Anyone else have this problem? I’m running windows 10 and have an NVIDIA GeForce GTX 970 (although my assumption is still that it’s simply a problem with the environment setup).
Sorry if it’s an obvious mistake! I’m new to this!
Edit: I saw in the comments that “environment-win.yml” was replaced by “environment.yml”. So I guess my question is now: why are so many libraries missing when I try to import them in the jupyter notebook?
Check the step with creating symbolic link for fastai folder.
Looks like you need to start jupyter notebook from folder that contains symlink to fastai.
I had the same issue and I solved it by creating a symbolic link for the fastai folder like this:
mklink /J E:\fastai\courses\ml1 E:\fastai\fastai
Where the first link is where you want to have the link: E:\fastai\courses\ml
The second link is the source: E:\fastai\fastai\
Earlier I had a problem with the installation.
conda env create -f environment.yml
always ended with: Command “python setup.py egg_info” failed with error code 1
When I tried to run the bulldozer notebook some modules where not installed. I had to manually install bcolz, cv2, graphviz, sklearn_pandas, isoweek, pandas summary, torch, torchvision and torchtext to get it working.
Anyone having a fix for my problem?
Thanks for sharing this topic, I’m a beginner here. Hope this will work for me. Thank you. Go2top Power ISO Crack.
I didn’t try after that. I started using clouderizer for the course.
I’m trying it again now. Clouderizer is too buggy.
Looks like you don’t have Microsoft visual studio installed
Installed Microsoft visual studio still getting the same error.