Paperspace setup help

(Mohammad Hossein Atlasi) #413

@dkobran @dillon

I’ve requested for a VM quite a few days ago but still haven’t received an approval.
I’m trying to connect to the European server since its closer to me.
I’m hoping for a P5000 with 250GB of storage on template, as indicated by the course guidelines.
My email that I used for signup is atlasi_hoseyn [at] yahoo .com


(Daniel) #414

@Atlasi You are all set though I would recommend also checking out Gradient :slight_smile:

(David) #415

hi @dkobran,

I’m finding that even for Gradient Notebooks, the types of machines available are limited. I can only pick C2 or C7.

The email I’ve signed up with starts with davidejimenezp.

If this could be changed, it would be much appreciated. Thank you.

(Mohammad Hossein Atlasi) #416

Hi @dkobran ,

Gradients also seem to be facing some limitations.
I tried to clone and continue the notebook for the course, and it told me that you have reached the capacity.

But the VM is working for me now, which is great!
Thanks for the support!


(Anouar) #417

As I read here, many people faced my same problem which is creating an account at paperspace and not being able to start right away. I am already waiting for three days for an approval to create a machine with GPU.
Is there another way or tool to speed things up? I really like to start immediately.

Thank you


Hi @dkobran @dillon, I’m also waiting for an account approval. My email starts with ‘emerac’.

Many thanks

(yifan zhang) #420

Hi @dillon and @dkobran , I’m encountering some payment issue right now.
I’ve tried all 4 of my cards, multiple times, none work. every time I clicks on ADD CARD, the page would show Error! Your card was declined.
I’ve also tried to my friend’s card, I’ve also tried to use my smartphone’s browser, still the same.
Additionally, every time after the page shows Declined, I can not edit my card info to try again because all the fields are freezed, so i’ll have to F5 and try again, think that’s probably a bug?
account is aeryen on gmail

(Tomasz Kaleczyc) #421

I’ve set up the Gradient notebook as advised. I started the course now but have run into two issues (possibly related) so far: it looks like NVidia CUDA is not set up for the notebook (or at least PyTorch does not seem to pick it up with torch.cuda.is_available()) and during the training of the first model, the kernel keeps dying before any training actually starts. Anyone else encountered this?


I have not had this problem. i’m using a P5000 with Gradient.

The one additional thing I did before running anything is run git pull.

(Robin Lin) #424

Greetings, @dkobran and @dillon could you please approve my request for a machine ? My email starts with robinrobin19980202@ Thanks a lot! Really looking forward to getting started!

(Gayathri Rengarajan) #425

@dkobran and @dillon Can you please approve my request for template machine. Email starts with gayathri.rengarajan .

Thank you!!


I’ve noticed that my kaggle dogbreed data as well as my installation of the kaggle API disappear every time I stop and clone my Gradient notebook. Also my installation of wget is gone.

Anyone else have that problem?

I was hoping to get a VM as well but it’s been a week and it looks like there are many of us waiting at this point.

(yifan zhang) #427

Just an update on this, I’ve since able to talk to Stripe Customer Support and they acknowledged it is a error on there side causing all the payment decline. But in theory it should be the business talk to them to work this out instead of end customer, because they don’t really have a support department interfacing with end users.

(abhash) #428

Hi @dkobran it has been around a week since I applied for a VM and am waiting for approval. I also tried gradient as suggested but I found it difficult to work with when downloading new data from the internet. Also with gradient, I could use only use the P5000 GPU a few times. Most of the time it was blacked out, perhaps due to a lot of demand. I was hoping for a P5000 gpu for my VM. My email is abhashjha[at]outlook[dot]com

Thank you for reading this

(Shaki) #429

I’m having issues setting up the paper space. I’m following the instructions from the video. The only machines I see are the CPU. There is no choice for GPU. Please advise


I signed up with paperspace and found the FASTAI15 coupon is expired. If the paperspace folks are here, is there any chance of a new one? It’s a nice gesture.

I managed to install pytorch-0.4 and all the conda stuff (cpu config) on a non-gpu machine. I might see if I can get the dog/cat model training to work on that, to see if the speed is tolerable. Otherwise paperspace looks nice.


This one still works


Thanks! And yes it works!

Added: Adding a gpu instance requires manual approval from them so I requested it and it’s still pending. They are out of the GPU+/P4000 instances so the lowest available is the P5000 which is 0.65/hour instead of 0.40. I guess that’s tolerable.

Another alternative might be OVH Public Cloud ( which starts about 0.57/hour for a GTX1070 instance. I went for paperspace as more affordable. It turns out to be almost the same but it does seem pretty nice. If I find I use it a lot, I might get a monthly gpu machine someplace instead of hourly.


Don’t wait for the approval (it takes more than a week), and get started with their Gradient notebook.


Thanks, I clicked Paperspace+Fast.AI gradient notebook and the only instances I could select were C2, C3, and K80 (from Google Cloud I think). I guess I can start with K80 if that’s what you’re suggesting.