What are all the prerequisite packages I need to install? Keep running into different import errors


#1

I am using the latest anaconda version and I don’t have python 3 installed independently besides what came with anaconda. Every time I try to run my notebook locally I kept running into different import errors. I might have missed something because I didn’t do part1, I took a different course. So what are the checklists of things you need to do to start part 2’s notebook correctly? I had to install xgboost, bcolz and now I am running into the issue with nltk importing standford library. I am doing everything locally btw because I have good enough graphic cards.


(Apil Tamang) #2

Hi @inkplay

You may/not be very familiar with python environment setup using anaconda, but well worth the effort getting the basics out of the way.

  1. We use anaconda to manage a environment specific to fastai. Be sure to check it out in google. The file ‘environment.yml’ contains all the packages that will be installed once you’ve started the process for creating a environment (also called ‘fastai’) for this project.

  2. Once the environment is setup, you need to activate it by using the command (in linux) ‘source activate fastai’.

  3. Then you must run jupyter lab from the same console

You might still need to download a lib. dependency here and there, but things like bcolz or xgboost should be available once the above is done. Let us know how this works for you.


#3

I wrote a short tutorial (based on internet resources and my own experiences) on installing packages for fast.ai.
It should walk you through at least lesson 8 (that’s where I am now). Go take a look:

Install Python 3.6 + Keras+Tensorflow GPU 1.6 on Windows 10 (3/12/2018)

Hope it helps!
Neil