I can’t get the http://35.194.23.221:8888/tree?token=blablabla site for the jupyter notebook to load.
Anyone unable to use the new fastai functions like download_images or widgets even after updating the library?
1 Like
@bluesky314 have you update conda in addition to pulling from the git repo? See https://course-v3.fast.ai/update_crestle.html
@andavargas Try logging out and in again and starting a new instance and let me know if that fixes it.
Is crestle free for a couple weeks? How to access free service?
1 Like
Hmm, I will take a look at this later tonight, currently in class. I’ve had something similar happen to me in the past and it resolved itself so I would keep trying. Let me know if the issue resolves itself, otherwise expect to hear from me.
@bluesky314 - can you try the instructions as described here again and give it a try? https://course-v3.fast.ai/update_crestle.html
We just saw fast-ai library got updated on conda by fast.ai team - this may resolve the issue you are having.
Still no luck.
Its working! Can I know what GPU we are using and what will be the pricing once the free period is over.
Also full disclaimer, I actually used crestle.com about 6 months back and it was SO unstable, including taking a lot of time to start the server that I left. Hope this is not the case now
So it seems like there is an issue with the firewall specifically in the PG&E building. We’ll look into why this is the case, but for now it should be working once you hop on another wifi network.
Tethering to my phone fixed it.
1 Like
Pricing is $.3 per hour, check that out here. We’ve moved the entire backend over to google cloud (currently using n1-highmem-8 machines) from aws and have a whole group of us devoted to making the platform stable and easy to use =) We’re working out bugs still but your experience shouldn’t be anything like before.
In notebooks/courses/fast-ai/course-v3/nbs/dl1/lesson2-sgd.ipynb whenever I define a tensor like so:
a = tensor(3.,2)
It complains: “TypeError: tensor() takes 1 positional argument but 2 were given”
If I change that to:
a = tensor([3.,2])
It works. I feel like this is might be because something isn’t importing correctly (wrong version of Torch?), which might explain downstream errors I’m getting with numpy. Thoughts?
when doing git pull I am getting “Your local changes to the following files would be overwritten by merge:
nbs/dl1/lesson2-download.ipynb
Please commit your changes or stash them before you merge.
Aborting” even thought the file is saved and shut
try git reset --hard Also be careful if you use this elsewhere because it deletes all of the changes you’ve made since the previous commit
So if versioning is wrong that is probably because the fast.ai team needs to push an update still, some things are still streaming in. I’m going to need to go home and rewatch the lecture though before I can weigh in too much though.
We are seeing its working (see attached)
Can you make sure you follow these steps before you start the notebook:
https://course-v3.fast.ai/update_crestle.html
I suspect you have some older version of libs and conda update will likely fix this
That means you haven’t got the latest fastai version.
Yes it says the pricing but which GPU it is?