The issue is when I evoke the ipynb from jupyter notebook (nbs then lessons1.ipynb) then error "Unreadable Notebook: /home/ubuntu/nbs/lesson1.ipynb NotJSONError(âNotebook does not appear to be JSON: :::â
I have checked the content of the file lesson1.ipynb, itâs been tranformed into HTML format file.
It seems that thereâs no tool to convert from HTML into ipynb (the other way round exists and works fine) ?
I know I can download the whole zip âcourses-master.zipâ onto my disk. Yes I have Anaconda installed on my laptop.
But I understand we are using the GPU serveur (P2) from AWS so I think I should download (wget) everthing onto my P2 machine (on the AWS) and run everything from there ?
Iâm having this same issue. Jade, were you able to resolve this?
I went to âhttp://files.fast.ai/files/nbsâ and saw the note âThis is not the place to download notebooks or spreadsheets from any more. Please use git, as mentioned in How To Get Started.â
In case anyone has this same problem, I figured it out. The actual ârawâ Python notebooks are stored in a different place on Github. You can access it by first clicking on the notebook (which will open it in your browser), then right-clicking âRawâ, and selecting Copy Link Address. Here it is for lesson 1:
For some reason I donât understand
==>This time, it lets me choose from my local disk (not on the AWS server) the ipynb files (notebook files),
As it 's from my PC, and the ipynb files were downloaded (a zip fie) from the Github (by clicking from the url shown on âPlease use git, as mentioned in How To Get Startedâ. It is working fine when I evoke it locally from my Jupyter Notebook (I had Anaconda installed as part of pre-requisite).
==>Therefore, since it lets me choose lesson1.ipynb from my local disk, itâs OK now.
The only thing I still donât understand is -
Before, when I had issue on opening the file after doing âwget and the url of github/lesson1.ipynbâ then got systematically the error "Unreadable Notebook: /home/ubuntu/nbs/lesson1.ipynb NotJSONError(âNotebook does not appear to be JSON: uâ\n\n\n\n\n\n\n<html lanâŚâ
It was because I chose the file from AWS serveur done by the wget. It seemed quite logic to me since I have launched jupyter notebook on AWSâs instance /home/ubuntu/nbs/lesson1.ipynb
Now, I do the same thing : i.e. I launch jupyter notebook on AWSâs instance -
e.g.
==> when I try to upload a notebook, it lets me choose from my local disk the ipynb files. No longer from ubuntu.
Do you have hints on why it happens like that ?
will there any impact on the following exercises if I choose ipynb files from local disk (and not from AWS?
Yes, Maxâs solution works :get lesson1 notebook from the below link to overcome issue with error "Unreadable Notebook: /home/ubuntu/nbs/lesson1.ipynb NotJSONError(âNotebook does not appear to be JSON: uâ\n\n\n\n\n\n\n<html lanâŚâ when evoke the lesson1.ipynb from Jupyter Notebook.
In order to recuperate the data, I actually installed Kaggle on the AWS server-instance:
Create profile at https://www.kaggle.com/
On server: pip install kaggle-cli
kg config -g -u [username] -p [password] -c dogs-vs-cats-redux-kernels-edition
mkdir ~/nbs/data/dogscats and cd to it
Go to Dogs vs. Cats Redux: Kernels Edition | Kaggle and accept competition requirements (by downloading the data)
On server: kg download
For data, you ll need to install unzip tool if it does not come along with the AMIS :
$ sudo apt-get install unzip
When the downloaded data are on the AWS server-instance, we can check the number of files:
ubuntu@ip-10-0-0-6:~/nbs/data/dogscats$ ls test | wc -l
12500
ubuntu@ip-10-0-0-6:~/nbs/data/dogscats$ ls train | wc -l
25000
Thank you so much for this Jade and Max. One question: when I click on lesson1.ipynb in my Jupyter notebook, I get the raw JSON version of the notebook, and not a nice looking Markdown file like the one depicted in the video. Does anybody know how to resolve this?
EDIT: Resolved the problem. Simply rename the notebook so it ends in the .ipynb extension. My notebook was actually a duplicate so it was named âlesson1.1.ipynbâ so it was being interpreted in raw JSON.
Hi everyone, having the same problem with the unreadable notebook. I tried the new file with the link Max posted and tried wget with that address and it worked.
My issue is that the file was working before and all of a sudden it stopped working. Anyone else had the same problem?