Platform: Paperspace and Gradient ✅

I’m having a very similar problem now!!

Thanks Dillon. I re-tried and mine is working now.

@dillon This is not working…

Bad news: It will start one of my old machines - which doesn’t do me much good!! I need the recent one!

1 Like

… and now it’s taking longer to error out - 8 minutes

1 Like

What is the error you are seeing?

The one I showed earlier:
image

It can’t seem to find container nm7ij0zc

It was able to start the older one, nu0y82o1

This is still broken!

1 Like

Having same error.

I had to get Paperspace tech support to fix it for me

For Gradient I always need to start a new notebook each time from scratch because opening doesn’t work (as per other issues). Since the /storage is preserved this isn’t a big issue.

For lesson3-imdb I get an ascii decode error like this:

A workaround is to manually set the codec in the notebook:

import locale
locale.setlocale(locale.LC_ALL, 'C.UTF-8')
2 Likes

Before I saw your post I saw this:

and that worked for me.

It’s great to know that works in Gradient too. Where are you running this command? In a Jupyter terminal?

Yes, ran that in a terminal. But later when I returned to the IMDB nb I had a similar ascii codec error and used your method in the nb to fix it. Strange that it’s not a persistent setting… I do find it quicker to just run it in the nb.

Two things - I realized that one of my issues with fastai was using pytorch 0.4.1. The issue was that I HAD pytorch verion 1 installed (pytorch nightly) but did NOT uninstall pytorch 0.4.1. Simple but inintuitive to a beginner.

My second is the same issue as @ricknta was having where I cannot download the Amazon dataset - however when I click accept on the Kaggle page it simply does nothing! I’ve tried accepting on firefox and on chrome. Any ideas?

@agoldina : Did you upload the kaggle.json file into the ~/.kaggle directory?

Or maybe try recreating the API token and redo those steps.

Sasha, it’s been almost 2 weeks so I don’t recall exactly what happened, but I think I had just failed to find the Accept button way at the bottom of the Kaggle page. I think when I clicked it, nothing happened, but then when I ran the code in the planets nb it downloaded properly. Hope this helps.

I’ve clicked accept and the code still fails to run :confused:
I’ve downloaded data from kaggle successfully before too

Is anybody finding that the directory /notebooks/course-v3/nbs/dl1/ is empty? I have (or had) my own notebooks, in addition to those from the course here, but today I find this directory to be empty. If I do git pull, it is still empty. I found this issue a couple of weeks ago, but after some hours the ipynbs re-appeared. Thanks.

@dillon