Crestle is a fraud service. I downloaded a dataset and kept it for one day (Monday). Next day(Tuesday) I deleted everything from my home directly, still it shows I have been using 25 gb disk space. Looting people for no reason.
How can I use Crestle to connect my localhost
You haven’t quite deleted everything. Try running
rm -rf .*. There’s a large
.local directory that you can view with
ls -al. Once you delete all the
. directories, the disk usage will update in a few hours.
Unfortunately, this isn’t possible with Crestle.
Hello! I’m new to using cloud service, and while I sign up for Crestle I haven’t noticed to shutdown Jupyter so before I log out, so now my free usage time for GPU is over. Now I have to pay the bills for continue using, I would like to make sure how to shutdown Jupyter after finished editing? Otherwise my account my be bankrupted in a short time if I don’t find out.
Besides, I just signed up in Crestle yesterday and didn’t do any coding yet, but my disk usage space is already over 170MB, I wonder why ?
I was a big fan of Crestle. I used it in my prev project and it worked like a Charm. I used the Paid service too and all was well.
Yesterday, I wanted to try out Crestle for my new project and it took a lot of time in creating the Jupyter CPU based notebook. I just logged out and then it said it could not create the notebook.
I logged back again and now it says I have already a Notebook running. But thats a ghost notebook. I dont have any URL where I can use that.
I hope I am not charged for this ghost notebook.
Also ,I am unable to create new notebook anymore as it says some notebook is already running… (Baaahh… Smoke outta my ears…)
I immediately wrote to Support e-mail address and there is no response on it (which is very disappointing). Is that e-mail address still functional?
I don’t know where else to ask for support.
My Crestle login is same as the e-mail address associated with this Forum account.
–EDITED-- to add the fact that I cant create new notebooks any more. I need to use the GPU for my latest work and I am running out of time. I am so used to Crestle and I dont want to try anything else. (I have used FloydHub before… Crestle was like fresh air… Hope it remains so)
2)Whenever I login, I get an immediate message “Sorry, the notebook could not be started. Please try again in a few minutes.”.
And when I start the notebook, it says “You can run only one notebook at a time.”
I got the same problem, any news?
looking into it!
working smooth again thanks!
Yep, everything is back to normal. Sorry for the trouble everyone!
Hi, I followed the guide in https://course-v3.fast.ai/start_crestle.html
When I was trying to use" fastai.structured import * " I receive the error in the bottom of this mail:
ModuleNotFoundError Traceback (most recent call last)
5 from sklearn import metrics
6 import pandas as pd
----> 7 from fastai.structured import *
ModuleNotFoundError: No module named ‘fastai.structured’
Have been using Crestle for a few days. was easy working with smaller datasets.
however, tried to load a 23GB dataset.
ended up getting ‘Connection to kernel lost’ error everytime i try loading the CSV file
df = pd.read_csv(‘train.csv’)
initally thought it was pertaining to some firewall issue. however tried the same using other laptops and even my mobile phone. getting the same error
Is there something i am missing here
Update : the exact error below
A connection to the notebook server could not be established. The notebook will continue trying to reconnect. Check your network connection or notebook server configuration.
Crestle is now owned by doc.ai, and I’d recommend reaching out to them for direct support: email@example.com or firstname.lastname@example.org.
You will be better off loading that into Amazon AWS (possibly using tools provided by Amazon that can support disconnected operations etc.) and then copying it back to the Amazon instance. Not sure if this is allowed by Crestle.
This is where floyds hub has an advantage. They have separate uploaders which take care of disconnected operations… But then, the floydhub interface is way too complicated (with a CLI etc…).