Platform: Colab (Free; $10/month Pro)

I concur to this.

1 Like

Hi all. So after the dust has settled: there now a standard, working, best-practice notebook for setup in Colab, and if so, where can I find it?

@jcatanza see the top post. There’s a few methods you can work with and all do.

1 Like

Hey @imrandude, apologies for my naiveness, but can you please clarify what’s the advantage of having a “VS Code Environment” in Google Colab? I think it is still not clear for me what’s a VS Code Env, haha When googling about it, it feels like it is an IDE made by Microsoft. If yes, then which features from them are advantageous to have on Colab? Thanks!

WOW, thank you @imrandude and @deeplearner! Your little hack for accessing jupyter env on Colab is working perfectly here :slight_smile:

2 Likes

Please check this for details -> https://forums.fast.ai/t/five-ways-to-debug-in-fastai/66496
Where @ slawekbiel has mentioned 5 steps of debugging Python code using FastAI v2 as example.
He has shown on how to use VSCode to debug as well.

More over, the limitation with Colab was that you couldn’t use it as a IDE. This is now enabled with VSCode and JupyterLab (ie., I suppose using nbcovert on the notebooks created).

2 Likes

Wow, what a cool hack. Didn’t know about subprocess. :slight_smile:

Hey guys,
I’m trying to work on a project in Google Colab, but get_image_files times out every time. I googled it and it seems like Colab is no longer able to process directories with lots of files in it. Did anyone face similar issues and have any workarounds?

I tried accessing the jupyter env on colab following the instructions provided in the first post. I get the ngrok tunnel running and able to see the jupyer environment as well. But the kernel is always busy and nothing happens when I run something in a cell. Am i missing something or is there a bug? I am attaching the screenshots of what i am seeing here in case it helps. I am running this colab on a mac. I have tried to restart the kernel, interrupt, shutdown and restart but nothing works.

Can you start terminal from jupyter server and check if you are able to run any bash commands ?

You can run bash commands by prefixing your command with ‘!’.

Just execute the cells with commands like these:

!pip install fastai2
!git clone some_git_repo

# and so on...

Since @pnvijay is not able to run any commands from jupyter, I am not sure if he can run the bash commands from jupyter.

sorry, I thought you were asking a question yourself… :smiley: Carry on…

I ran lesson-2 notebook on colab without any issue. Are you having issues with the same notebook or running a custom code ?

Sorry for not being clear. I am trying to see what is causing the issue for @pnvijay. If he couldnt run the bash commands from terminal, my guess is that ngrok is not able to communicate the messages back to the colab server

Any way to store your work using this amazing hack @imrandude @deeplearner ? (i was not able to think of a way)
similarly i end up installing vim etc over again everytime i connect (not a big deal for the ability to use Jupyter Nbs) :slight_smile: Just wanted to know if i’m missing another trick from you guys :smiley:

1 Like

No apologies, entirely my fault :slight_smile:

Hope you find it, but with ngrok, nothing really for me to contribute.

By storing work, do you mean the changes you made to the notebooks or any new notebooks or scripts that you might create on the jupyter server? If so, then they should be automatically saved to the google drive location

1 Like

Are there someone use Colab pro out of US ? For example in France ? I see that it’s available now just in US but maybe it’s not very strict. Do we have some limitations using Colab out of US ? Thanks

I’m facing issues with a custom notebook. It has around 194082 image files and Colab crashes most of the time. I guess I’ll have to purchase pro