Platform: Kaggle Kernels

init_27 · January 31, 2019, 6:34am

Added a note in the top post

miwojc · January 31, 2019, 8:12am

Seems time limit was increased to 9 hours as per kaggle docs, didn’t test it yet…

Simmo · January 31, 2019, 9:13am

I think this is a result of insufficient shared memory, you can work around it my setting num_workers=0

https://www.kaggle.com/product-feedback/72606#latest-429470

basitis3 · February 2, 2019, 8:50pm

Hello,
i want to try out my own data set:
so i uploaded it as a .zip, and it shows in …input directory
my question now is, how do i specify it as a path and if i need to do any unzipping for the images?

Edit:
i tried to:

path = Path(’…/input/’)
path_img = Path(’…/input/celebs/celebs.zip’)

and when i run:
fnames = get_image_files(path_img)
fnames[:5]

i get an error: FileNotFoundError: [Errno 2] No such file or directory: '…/input/celebs/celebs.zip’

45bad7273fe32fb258d87ecf641122fa

andresa · February 4, 2019, 6:45am

Hi,
I am in lesson 1 and when I try to download the pet data I get this error message.

gaierror Traceback (most recent call last)
/opt/conda/lib/python3.6/site-packages/urllib3/connection.py in _new_conn(self)
140 conn = connection.create_connection(
–> 141 (self.host, self.port), self.timeout, **extra_kw)
142

/opt/conda/lib/python3.6/site-packages/urllib3/util/connection.py in create_connection(address, timeout, source_address, socket_options)
59
—> 60 for res in socket.getaddrinfo(host, port, family, socket.SOCK_STREAM):

Any ideas?

Solution: Although, I had enabled the internet connection, I needed to refresh my browser for it to work. All good now.

init_27 · February 4, 2019, 6:47am

Hi @andresa,
Can you double check if the ‘internet connection’ is enabled in the kernel?

andresa · February 4, 2019, 6:48am

Yes it is?

andresa · February 4, 2019, 6:50am

Hi Sanyam, after a refresh I have it working.
Thanks

init_27 · February 4, 2019, 6:51am

Kindly edit your question and add the solution there as well, Thanks.

andresa · February 4, 2019, 6:55am

Hi Sanyam,
All done.

Simmo · February 7, 2019, 12:38am

@wdhorton @miwojc

Can you please add me too? My kaggle user is dromosys

Thanks

init_27 · February 7, 2019, 12:53am

I’ve fixed the following kernels that had BusErrors and I’ve made some minor error fixes as well:

I think these errors were due to a version difference. Please let me know if you spot any other issues.
Thanks to @Ajaxon6255 for sharing many fixes.

andresa · February 7, 2019, 12:42pm

I am trying to run lesson 2 and the notebook keeps refreshing and losing all my work. How can I fix this?

init_27 · February 7, 2019, 12:45pm

Not sure why that’s happening.
Do you have a fixed speed connection? It might be an outage at kaggle’s end.

andresa · February 7, 2019, 12:46pm

does pressing commit save my work?

andresa · February 7, 2019, 12:55pm

How do you save your work? My work is constantly lost by kaggle refreshing?

init_27 · February 7, 2019, 3:04pm

@andresa Gotcha! So by default you’ll lose the data when you refresh, you need to commit to save the work.
Warning: For Lesson 2, since the number of images downloaded>500, the kernels will give an error because this is larger than kaggle’s limit.

piaoya · February 7, 2019, 6:32pm

I am trying to train a model with
learn = create_cnn(data, models.resnet34, metrics=error_rate)

I get the following error:
OSError: [Errno 30] Read-only file system: '../input/wildkraut/wildkraut/models'

My kernel has an internet connection, I refreshed the site and my GPU is on… I uploaded the dataset in a normal way and the dataBunch shows up… Thankful for any help <3

wdhorton · February 7, 2019, 6:43pm

You should pass the ‘model_dir’ keyword arg to create_cnn. You can set it to tmp/models

piaoya · February 7, 2019, 7:05pm

Great - thanks @wdhorton that seemed to work. I guess model_dir = '/tmp/models' tells the create_cnn that it should open up a new folder and save the created model inside? (Couldn’t find an explanation for the keyword arg neither in the docs, nor anywhere else…)