Hi Jeremy I used the fast ai script to set up paperspace. But I am having issues in that script. It isn’t able to install Jupyter notebook. It shows the following error while installation
Traceback (most recent call last):
File “/home/paperspace/anaconda3/envs/fastai/bin/jupyter-notebook”, line 7, in
from notebook.notebookapp import main
File “/home/paperspace/anaconda3/envs/fastai/lib/python3.6/site-packages/notebook/notebookapp.py”, line 45, in
ioloop.install()
File “/home/paperspace/anaconda3/envs/fastai/lib/python3.6/site-packages/zmq/eventloop/ioloop.py”, line 210, in install
assert (not ioloop.IOLoop.initialized()) or
AttributeError: type object ‘IOLoop’ has no attribute ‘initialized’
I created a new machine in Paperspace with the template they have. I DID NOT “git pull” or “conda env update” and worked fine out of the box
When I ran the git pull and update in conda the environment did not work and was full of errors about paths, " "bcolz" package not found and dead kernels when running python3
@leromerom, @satish860 I am having a hard time connecting to jupyter notebook as well. I launched a fast.ai machine yesterday and have tried all of the suggestions listed above: ssh into the remote machine using both ssh options listed, pip uninstall tornado followed with pip install tornado==4.5.3. I even tried curl http://files.fast.ai/setup/paperspace | bash from the medium post. I am attempting to launch a new machine to start fresh and waiting for approval for that one. At which point, I will NOT do a git pull or a conda update as suggested in an earlier post. I’m sure there is a magic formula that works, but really appreciate any advice…
@satish860. Thanks for the advice. I finally got it working on an AWS instance following the reshamas instructions here: https://github.com/reshamas/fastai_deeplearn_part1/blob/master/tools/aws_ami_gpu_setup.md. The only thing I didn’t different from the instructions was in part III: jupyter notebook --generate-config command didn’t generate the config file so I added it with nano config,added the line c.NotebookApp.ip = '*' to the file, and saved in fastai directory. I then launched the jupyter notebook with jupyter notebook and pasted in the url with token (leaving the localhost:8888 as is) into a new window and it worked! I’m including all of these details in case it’s useful for someone else.
It did generate the config file but it isn’t in the current directory…
It’s probably hidden under the home directory (atleast in Windows) ../../.jupyter/jupyter-config.py
I am having errors importing libraries when running lesson1 script. Not sure if I need to change my paperspace set up. Does anyone have a solution?
I chose fast.ai template to install. Then I updated the environment and did git pull. Then I am trying to run the lesson 1 code. The kernel is dieing when importing the libraries. Then my subsequent code throws errors.
I’d like to rant about paperspace approval process, appologies if this is a wrong thread for this!
Been thinking whether I should keep quiet about this and stay patient. Decided to speak up.
Paperspace process sure sounds broken. I submitted a GPU request one week ago and still haven’t heard back from them. Do they want my money or not? I would recommend not using them in future courses unless they get their act together. I don’t care if they are cheaper or not, if I have to waste time complaining about them I’d rather pay more to someone with good service. And yes, my purpose is to do this course and run machine learning experiments for my schoolwork.
Hi Arturs, sorry for the less-than-ideal experience! We are getting through approvals as quickly as possible but have unfortunately been inundated by a lot of non Fast.ai users who are flooding the system making it harder for us to get you guys set up. Thanks for being patient and I assure you we are working around the clock to improve this experience!