Platform: Paperspace and Gradient ✅

@abhayvarma You can use the same machine type but not the same exact VM. Gradient runs inside a docker container, not a full VM. Containers remove a lot of the complexity around management which is why we’re moving in that direction. Here’s the P4000 machine types in Gradient.

@NandoBr Just submit a ticket here and we’ll take care of you right away:

https://support.paperspace.com/hc/en-us/requests/new

Be sure to include repro steps. Thanks!

@sahilk1610 The new course is based on a new framework (+ new NVIDIA drivers/CUDA,cuDNN etc.) so I would definitely recommend starting with a fresh Gradient Notebook. You can keep your old machine around though if you want to have it available for future reference. Hope that helps.

1 Like

@vijaysai You’re right, this is confusing. We thought it would be nice to be able to version Notebooks but it should be an option, not the default. I’ll create a ticket to make this change. FWIW, you are not charged anything for having these extra Notebook in your account.

1 Like

I had the promo code from last year, but I cannot see if it is still active nor the credit that is left. Where can I check?

@dkobran How do I access the data in storage? I was a bit freaked out when Gradient moved my data folder there but then I found the data - but don’t know how to set a path to it. The download_images nb uses Path like this:

path = Path(‘data/sayoc’)

which worked fine when I had my images in the data folder but if use ‘storage/sayoc’ it doesn’t find my images…

So I started the course once before (on a paperspace vm) and did the whole selenium, chromedriver, etc song and dance to be able to download my own image sets.

Cannot for the life of me get sudo, etc to work on a gradient notebook (neither via cells in a notebook, not via starting a terminal window.)

Any suggestions?

Thanks,

-L

Notebooks should be running as root so sudo is likely not necessary. What errors are you seeing?

The container is set to override the default fastai download location to the persistent /storage directory using the config.yml file. The fast.ai docs are here

and the config.yml file that we use in the Paperspace+Fast.ai Notebook can be found here: https://github.com/Paperspace/fastai-docker/blob/fastai/pytorch1.0/config.yml

Thanks but I have no idea what to do with that information…

How do I access data in the storage folder from a nb? Do I have to modify the config.yml file? If so, where is that located?

The /storage directory is a special folder that is mounted on every notebook. Anything that you put in that folder using and standard command will show up on every subsequent notebook.

Which notebook are you using? Let me see if I can reproduce your issue.

I’m running download_images RN. I just need to know how to reference the path to the data (storage) folder

Now I can’t get into my VM (“notebook”) at all - just spins…!!!

Is there a way to download all the files I have in a subfolder of the /storage folder? Meaning, download from my VM to my local PC.

OK I finally got it - in case anyone else has the same problem - you have to use /storage, not just storage…

OSError: cannot identify image file ‘/storage/oxford-iiit-pet/images/Siamese_47.jpg’

Hi I am getting above error while running fastai lesson1 in paperspace. Please help me out

1 Like

Hi Dillon – Situations like this:
Any suggestions? Thanks.

image

image

2 Likes

There’s not a way for me to swap between gpu and cpu for a core machine is there using Paperspace?

One of the reasons I would end up going to my own local machine is that coding on a gpu enabled seems like a waste of time. I just need the gpu when I’m running experiments.

Edit: Alright so I was a core user and Gradient seems like it does this very well. You launch your stored notebook with whatever resources you want to pay for. So I can code it up (i.e. make a bunch of mistakes) and then once everything is laid out correctly swap to a more powerful GPU and run it. Just deleted Core and swapped to Gradient. Nice!

Please advise how to update the course notebooks, I tried pip install update fastai but the notebooks still are only for lesson 1.

What is the correct github command and launch directory to update the course-v3/DL1/nbs