http://gluon.mxnet.io/ . It’s pretty good.
Jermey / fast.ai has put in lot of thought in choosing PyTorch. May be it provides the best learning / experimentation environment. It might be better if we keep from discussing other frameworks while we are doing the course. That way we can focus on “Deep Learning” and avoid getting into this discussion on which one is better.
Once we learn to use one Framework, I am sure we can achieve the same in other new or older frameworks. That would be a valid exercise once the course is over to port the notebooks to TF or Gluon or anything else to see how it feels to work in it. Just my thoughts as a fellow student.
I think these discussions are interesting, but maybe better on #applications so that newer participants don’t get overwhelmed? There’s actually a discussion of this particular topic over at #alumni BTW: Gluon - a new deep learning library
I think Gluon looks interesting, but I don’t see any benefits vs pytorch at this stage. Still, if anyone tries it, I’d be interested in hear feedback of pros and cons.
I agree with what you said once we get the concepts right, we can work with any framework. I was just reading through the documentation and they have done good work explaining the concepts.
So I thought one can give it a read.