Never ever give up, Colab can not download my file

Hi , I had problem with paperspace so I went to codlab which works pretty fine.
(For paperspace , I post in forum twice , try to contact one of very active people in course for mentoring , contact paperspace technical , haha , I do not give this course up)
Long story short ,I tried colab and it seems awesome, But it can not download image files for me.
It actually worked in my google drive, I mean I made all of those changes and then checked and then upload cvs file in the right place but I got the error that the directory is not existed?!
What do I do now? I listen to the video and take notes till someone help me and once in a while try to see if I could solve this myself…Thank you everyone especially Jeremy and Rachel.
You are all awesome
Happy Learning
29%20PM|690x365 14%20PM

Try skipping the /content of root drive declaration. I never do that. Eg if I want to go to a file I just start out with Path(‘gdrive/My Drive/…’)

1 Like

Thank you , I will try that. Hopefully it will work.

Let me know if it doesn’t! If it doesn’t, hit a fresh restart and do a %ls. That will show you what directory you’re currently in and where your root is.

1 Like

I delete the content part , it is still the same error and my ls command works fine?
Any idea? you can check my new screenshots.
Thanks man

I used ls , it shows following.

Do %ls, that will show you the working directory

I did ls , it shows me the following , it is correct or I should navigate to another directory?

Out of curiosity, could you post a link to entire notebook?

I could, Now I did search on fast AI, stachoverflow and I could download the first series of image, my google drive seems messy , give me a few minute , hopefully I could do the other two too. Thanks

Awesome! What was the solution?

Too be honest don`t know exactly what was the solution yet.
Here is the notebook link:
https://colab.research.google.com/drive/1d3SwpR4877vnlf2j_vyRpRuKYi_PIhXj
I am working on doing for second time, it seems google drive is not connected. Let me see if I could figure it out.

Yes it looked to me like the drive was not properly mounting. If you run into more issues I can show you the one I set up for my study group for the university I attend if you’d like.

1 Like

Even with this issue , it runs and I am now training model. haha.
I will train the model , finish it , then try to do the whole thing again. It seem putting the file on data helped me to get there. So I put all files on data, which jeremy did put files on different folder and plus the point that you mentioned about content , content should not be there at least in this version.
Thank you so much for your help and definitely ask for your setup if I could not figure out things in order.

1 Like

Wonderful! Glad to hear it :slight_smile:

1 Like

@muellerzr
One of the things that jeremy emphasize is to run the code, The problem for me is I could not set up paperspace so I set up colab.It seems colab is very slow and when I leave my laptop after an hour or two so my laptop disconnect from internet and I have to run it again. The very first lesson (pet image) talk too long even though I did bs = 16 and the second lesson seems images were way off and took too long with very bad result. Seem first lesson is much easier but still took too long to run.
Paperspace setup?
How can i stop my laptop not disconnect from net when I am not around so it runs the code?(use mac)
Sorry for asking so rudimentary question , I am searching them in internet to find some solution. (Watching lecture 4-5) do not want to watch and not to run code.
Any idea how can i make it talk less time?
Also in lesson 0 I can not find markdown tab in my notebook.
Screen shots attached which may help for all my questions.

update :
For Jeremy took less than two min to train the model in lesson one, for me with colab it took around 40min for each epoch with is very different.

First, are you sure the GPU is activated? To answer the rest, Colab is free but it’s not lightning fast. So don’t expect fast results from image classification. Small tabular datasets take not that long but it uses a different GPU than Jeremy and others. That is a caviat along with the 12 hr wipe cycle.

1 Like