Support for Fastai Part1 v3


#1

Hey @prakash, how to use clouderizer for the coming fastai course which uses v1 of fastai library build on top of PyTorch v1 ?


#2

I am interested in this topic also!


#3

I’d also be interested in using Clouderizer for fast.ai v3


#4

Can you guys point me to url where notebooks for this is?


#5

Hi @prakash,
My understanding is that notebooks are uploaded right before each session starts.


#6

Here is the link to the notebooks I think will be used.


@prakash Would be able to use Clouderizer for v3 of the course ?


#7

Last year lesson 1 acted buggy due to last minute fastai sw changes (during the first week or so). I have higher expectations this course. It is very dynamic near to the bleeding edge.:smirk:I hope Prakash can pull us through with a V3 install walkthrough! It worked very well last year as a blog post.:crossed_fingers:


#8

Anyone here trying to set up fastai V3. When i am trying to pip install fastai,i am facing version issue.It is still pulling old version i.e. 0.7 Is it because clouderizer is configured for this version?


#9

Hey guys,

The fastai crew already made a script to configure fastai v3 to work with Google Colab, so it works perfectly with clouderizer as well.

  • Create a new project.
  • Insert the url for the code https://github.com/fastai/course-v3.git
  • Insert the following line as startup script: curl https://course-v3.fast.ai/setup/colab | bash
  • Insert any Kaggle Datasets or additional apt-packages you like

That’s it. Working perfectly

Sometimes the startup script doesn’t run completely, if that happens you just run script again in a notebook cell or in the terminal.

I hope I’ve helped.