I see a bunch of fastai libraries are imported in the code that I am am guessing a wrapper on Pytorch
from fastai.transforms import *
from fastai.conv_learner import *
from fastai.model import *
from fastai.dataset import *
from fastai.sgdr import *
from fastai.plots import *
The problem is, if I want to learn about how some of the built in functions (e.g. ConvLearner.pretrained) are working (parameters etc.) I have no reference other than looking into fastai library code! I find this as a high barrier to my learning.
I have worked on Tensorflow and Keras before and happy to learn Pytorch. Is there any other source of these code I can follow through (either using tensorflow or pytorch or keras) rather than using fastai library?
In this case it should not take that much time for you to read the code. There were reasons why Pytorch was selected as main DL library for this cource: http://www.fast.ai/2017/09/08/introducing-pytorch-for-fastai/
Some notes you can find here Code reading.
At this point we’re teaching you fastai, and training deep learning models in general, not pytorch. In later lessons we’ll be teaching you pytorch. Therefore, I’d suggest delaying studying pytorch, unless you have a particular interest in accelerating that part of your study (in which case reading the code is the best approach).
@jeremy, Am I allowed to use the standard Fastai Library in commercial (real life) Projects ?
English is a barrier to many employees in my company. Can I translate the material to Portuguese (making the appropriate references to you and the fastai website) after it is made public ?
Thanks in advance.
Yes sure! Be sure to let us know when you have translations so we can make them available more widely.