I was trying to learn Tensorflow and high level API’s Keras/TFLearn since last few weeks, now since this course will use Pytorch, is it advisable to keep working through both deep learning frameworks - Tensorflow and Pytorch ?
My personal experience has been that without applying the knowledge, you will forget the usage of these frameworks. And it is confusing to use multiple of them at learning stage. If you are applying your TF learning in hands-on projects, if should be okay to juggle with both.
I will be sticking with PyTorch along with fast.ai for sometime now. Have heard good things about it’s design, from practitioners.
As @anandsaha said without proper knowledge and understanding of theory knowing many frameworks wont help. Its always better to start coding from scratch (Using numpy) for trivial problems and then you can always re-implement them in any framework of your choice. To understand the strengths and weakness of different framework the following would help!
It’s good to implement from scratch (and we will) but I find it best to first learn to train good networks using existing tools. This is the approach we’ll use in the course.
Hi Jeremy,
I have been unable to see or hear live stream, could be bad internet. I wish I will get to see video later and also get instructions for set up in wiki somewhere. Looking forward to learn and share.
If all videos will be posted within 24 hours, that’s going to be perfect for us Europeans. I’m actually finishing today’s workshop just now, I simply paused it and I’m now able to watch it until the end. Perfect!