I didn’t see it in the .jupyter directory. The jupyter-config.py is a file so the config file should be in .jupyter directory. Do I have that correct?
Thanks, I am now all set. Appreciate the very fast response time! 
Arturs
Yes absolutely correct…
In Windows it’s under you Username/.jupyter/jupyter-notebook-config.py
Not so sure about Linux but it should be something equivalent to that
If any body need promo code, here is mine: 2Q553C7
referral url
BTW, the FASTAI15 isn’t available now.
@leromerom I did not create a new machine but I uninstalled tornado and reinstall tornado 4.5.3. I am having a bit of progress. Thanks a lot!
However my kernel seems to be dieing still. So I probably will have to delete machine and create a new machine. But before doing that I wanted to check with you - were you charged $5 once again when you created the new machine? Sorry am a student so on a tight budget.
Hi @sshleifer. I’m having exactly the same issue. Did you find how to manage it without reinstalling cuda stuff?
No, gave up 
let me know if you do
The comments here are mildly helpful
https://stackoverflow.com/questions/48728426/tensorflow-device-lib-list-local-devices-doesnt-see-gpu
Suggest that may need tensorflow 1.6, but obviously I didn’t figure this out so maybe don’t listen to me!
The first time I created a machine I entered the code @reshama posted in her installation guide and got, I think like $15 credit,I created the second one and was charged again but I still have some credit left. I believe you only get billed proportionally to the time used for the month if you delete a machine.
Hi @dillon. I’m getting this “packet_write_wait: Connetion to (IP address) port 22: Broken pipe” output when trying to log-in paperspace machine (GPU+, fast.ai template, East Coast option).
I’m checking in the forum trying to figure out the solution. No success so far.
Thanks in advance for your support.
2 Likes
I’ve finally managed to make it work. It seems that I installed simple tensorflow while I had to install tensorflow-gpu. Now it works with tensorflow-gpu-1.5, keras-2.1.5 and cuda-9.0
+1 @dillon I’m also having the same issue waiting for account approval
Please do let me know if I need to provide any additional information.
Thanks
I have problems to start up Paperspace it takes too long I will try AWS, has it worked fine for you?
AWS worked great. I used these instructions: https://github.com/reshamas/fastai_deeplearn_part1/blob/master/tools/aws_ami_gpu_setup.md3
I did run into a bit of a problem launching jupyter notebook from my ec2 instance after setting up a new key pair on the instance in order to access an s3 bucket which had my downloaded images in it. I shut everything down and rebooted and that seemed to fix the problem. Hope it works for you!
I got my approval within 2 days. I guess you can wait or reapply again.
@keri how much is the cost for this instance, I say something about 7-9 dollars an hour is that right?
What type of machine did you chose?
Thank you for your help
So far, I have run a Linux p2.xlarge instance for 13 hours and have spent $11.89. I forgot to shut it down for a few hours, otherwise the use would be much less than that.
in AWS I looked for the “fastai-part1v2-p2 - ami-c6ac1cbc” instance since if I searched in the community AMI it would not be listed.
I installed the machine have access via ssh, however the files for the course were not copied.
Does anyone know what is the command to get all the files needed for the course copied to my machine that basically is blank?