Is it recommended to install the environment before the first lecture?

Hi,

There already are a handful of threads with the setup instructions for different platforms (thanks, people!). My computer is a Mac laptop so my plan is to use either AWS or Salamander. Do you recommend to set it up before the course begins or is it just fine to wait after the first lesson?
My understanding is that the first lecture teaches how to setup the environment in Paperspace, right?

You can wait

4 Likes

Thanks Jeremy. I am trying to see if i can get started with colab. If that does not work out, i will look for other cloud options.

1 Like

I am thinking on the same lines. Google Drive is now working seamlessly with Colab. Lets collaborate to set this up once we have the instructions from @jeremy

If you’ve got fastai v1 running with a GPU, then you’ll be ready to go.

Thanks Jeremy.

I created a GPU enabled collab notebook. The installation was pretty straightforward. I’ve created a Github Gist for reference.

2 Likes

Thanks! Can you run lesson 1 pets on collab? https://github.com/fastai/course-v3/blob/master/nbs/dl1/lesson1-pets.ipynb

Let me give it a try.

1 Like

@miwojc I tried running the first lesson. But I’m stuck with untar_data. The data keeps downloading to "root/.fastai/’. Am not able to find this directory. I’ve already mounted my drive and created a seperate folder for the course. Will update here if I could get around this.

1 Like

@miwojc I was able to download the data. But now, while I’m trying to train the model using the resnet34 architecture, the pretrained model gets downloaded to “/root/.torch/models/resnet34-333f7ec4.pth”. This path not being accessible, returns a Runtime error while I try to train. Can I somehow change the pathe where this pretrained model gets downloaded to ?

Excerpt from my collab notebook:
Downloading: “https://download.pytorch.org/models/resnet34-333f7ec4.pth” to /root/.torch/models/resnet34-333f7ec4.pth 100%|██████████| 87306240/87306240 [00:01<00:00, 51999294.49it/s]