Why no installation for local machine?

Why there is no clear installation and instructions for local machine?
I installed the fast.ai on windows10 (How to set up Windows 10 for fast.ai),but can’t do anything and don’t know how to start?
Why fast.ai too hard to use and messy ?

I watched the video and don’t know how to start.
I read the website,wiki,etc and still don’t know how to start.



There’s several options to use fastai and follow the courses. The easiest are using cloud platforms, like for example crestle.ai or colab.
If you want to use your local machine it can be harder and more frustrating. I think the guide you’ve linked to is obsolete, please follow this one instead.

Please try to persevere, fast.ai is worth it. Many other students managed to set up their environement to follow the fast.ai courses, there’s no reason you can’t do it !

1 Like

I installed fastai on gaming laptop with Nvidia 1070 GPU.
I use Ubuntu and I haven’t touched Windows for many years, so I can’t help with specifics, but basic steps should be the same.
Have you installed anaconda?
It’s best approach, anaconda takes care of all dependencies.

Oh my bad, as you linked the guide for Windows I thought you were using that OS. I’m also on Ubuntu.

fastai installation is very simple with anaconda on Linux. Simply type on the CLI :

conda install -c pytorch -c fastai fastai

And you should be able to import fastai.

Please note you have to have Python >= 3.6, and working CUDA drivers if you want to use your GPU. See here for troubleshooting if you’re still having issues.

1 Like

My installation steps:

refer to : How to set up Windows 10 for fast.ai

  1. install nvidia gpu drivers windows

  2. Install Anaconda (Python 3.6) using the 64-bit graphical installer(https://www.anaconda.com/download/?lang=en-us#download).
    Choose the option to install for “this user only”

  3. Install Git for Windows


  1. Run Anaconda Prompt from your start menu, and type:
    git clone https://github.com/fastai/fastai.git

  2. Once the repo is cloned, you’re ready to set up fastai:
    cd fastai
    conda env create -f environment.yml
    activate fastai

  3. open jupyter:
    (fastai)> jupyter notebook

When I import fastai. It says “no module fastai”

As I’ve said in my last message, the guide you’re using is for Windows while you said you’re on Linux Ubuntu. The instructions are not the same. Please see my last message for a guide on how to set up on Ubuntu. It’s very simple.

EDIT : I’ve just realised I’ve been talking to two different people, didn’t realise that sorry for the confusion. @hwdong I have no experience with fastai on windows, can’t help you with that sorry. My advice is look for installation troubleshooting threads on the forums (see the fastai users section) for help and similar situations.

When I import fastai . It says “no module fastai”

this means the fastai folder is not in the same folder as your notebooks. Can you confirm? If they are not place the folder where the notebooks and it should work.

For window installation follow this thread, it works Pytorch v1.0 stable is now working on Windows but fastai v1 needs some tweaks to get it work on Windows

My 2 cents … if you’re going local, go with Linux.

Once you can dual boot Ubuntu on your windows machine, search the forum for “Local Linux setup” and you’ll find plenty of good resources for installing all the bits you need to work with pytorch/fastai.

If you’re completely new to Linux, I suggest you stick with crestle or colab initially. There are really good instructions on the course wiki for these and other cloud providers.


So does fastai ONLY work using GPUs?

I have setup FastAI on Windows locally/ I am able to run PyTorch locally and use my GPU but I don’t know how do i setup my GPU for fastai while running Jupyter Notebook. Cuda device is available in Jupyter notebook. It may sound silly, but how do I enable GPU computing?